diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,71587 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999580800670719,
+  "global_step": 11927,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-08,
+      "loss": 0.9314,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-07,
+      "loss": 1.4215,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6759776536312851e-07,
+      "loss": 1.2928,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 1.6147,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-07,
+      "loss": 1.4885,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 1.4178,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.910614525139665e-07,
+      "loss": 1.356,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 1.4244,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.027932960893855e-07,
+      "loss": 1.5311,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 0.1779,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.145251396648045e-07,
+      "loss": 1.4092,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.5496,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.262569832402236e-07,
+      "loss": 1.4841,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 1.4464,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.379888268156425e-07,
+      "loss": 1.4647,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 1.5339,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.497206703910615e-07,
+      "loss": 1.4923,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.3739,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0614525139664806e-06,
+      "loss": 1.301,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 1.5255,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1731843575418997e-06,
+      "loss": 1.2321,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 1.4058,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2849162011173185e-06,
+      "loss": 1.3553,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 1.4286,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3966480446927375e-06,
+      "loss": 1.2585,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 1.2245,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5083798882681566e-06,
+      "loss": 1.3008,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 1.3208,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6201117318435754e-06,
+      "loss": 1.2549,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 1.3884,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7318435754189947e-06,
+      "loss": 1.1995,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 1.2949,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8435754189944135e-06,
+      "loss": 1.2386,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 1.1633,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9553072625698325e-06,
+      "loss": 1.2322,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.1814,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.067039106145252e-06,
+      "loss": 1.2201,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 1.1953,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1787709497206706e-06,
+      "loss": 1.116,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 1.1151,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2905027932960895e-06,
+      "loss": 1.1199,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 1.1976,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4022346368715087e-06,
+      "loss": 1.1166,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 1.1841,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5139664804469276e-06,
+      "loss": 1.1264,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 1.0809,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6256983240223464e-06,
+      "loss": 1.1103,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 1.1004,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7374301675977656e-06,
+      "loss": 1.0877,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 1.0737,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8491620111731845e-06,
+      "loss": 1.2165,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 1.1238,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9608938547486037e-06,
+      "loss": 1.1461,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 1.1401,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0726256983240226e-06,
+      "loss": 1.0933,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 1.1497,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1843575418994414e-06,
+      "loss": 1.0388,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 0.2153,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2960893854748607e-06,
+      "loss": 1.1273,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 1.1429,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4078212290502795e-06,
+      "loss": 0.9513,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 1.1108,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5195530726256988e-06,
+      "loss": 1.0703,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 1.1256,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6312849162011176e-06,
+      "loss": 0.2069,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 1.044,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7430167597765364e-06,
+      "loss": 0.9957,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 1.0629,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.854748603351956e-06,
+      "loss": 0.2042,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 1.1075,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9664804469273745e-06,
+      "loss": 1.0811,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 1.0491,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.078212290502794e-06,
+      "loss": 1.1748,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.954,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189944134078213e-06,
+      "loss": 1.0246,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 1.0995,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.301675977653632e-06,
+      "loss": 1.1043,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 0.2189,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.413407821229051e-06,
+      "loss": 1.0916,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 1.1773,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5251396648044695e-06,
+      "loss": 0.9576,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 0.9542,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.636871508379888e-06,
+      "loss": 1.0192,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 1.0859,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748603351955308e-06,
+      "loss": 0.9724,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.0695,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.860335195530727e-06,
+      "loss": 1.1345,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 1.079,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972067039106146e-06,
+      "loss": 1.0732,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 1.0833,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0837988826815645e-06,
+      "loss": 1.0326,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 0.2037,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.195530726256983e-06,
+      "loss": 1.1395,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 0.9644,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.307262569832403e-06,
+      "loss": 1.0242,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 0.9973,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.418994413407822e-06,
+      "loss": 0.1916,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 0.997,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.530726256983241e-06,
+      "loss": 1.1156,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 1.0953,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6424581005586595e-06,
+      "loss": 1.0272,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 1.0845,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.754189944134078e-06,
+      "loss": 1.0214,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 0.9858,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.865921787709497e-06,
+      "loss": 1.0793,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 1.0154,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.977653631284917e-06,
+      "loss": 0.2107,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 0.9486,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.089385474860336e-06,
+      "loss": 1.0494,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 1.0223,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2011173184357546e-06,
+      "loss": 1.0109,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 0.1937,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.312849162011173e-06,
+      "loss": 0.9855,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 0.999,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.424581005586593e-06,
+      "loss": 1.175,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 1.1116,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.536312849162012e-06,
+      "loss": 1.0592,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 1.0216,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.648044692737431e-06,
+      "loss": 1.0425,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 1.1677,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7597765363128496e-06,
+      "loss": 0.9902,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 1.0318,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.871508379888268e-06,
+      "loss": 1.0109,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 1.151,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.983240223463687e-06,
+      "loss": 1.0826,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 0.9601,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.094972067039106e-06,
+      "loss": 1.0194,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 0.8922,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.206703910614526e-06,
+      "loss": 0.9891,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 1.0923,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.318435754189945e-06,
+      "loss": 1.1238,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 1.0032,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.430167597765364e-06,
+      "loss": 1.0138,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 1.2532,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.541899441340783e-06,
+      "loss": 0.8773,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 0.9193,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.653631284916202e-06,
+      "loss": 1.1311,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 1.0122,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.76536312849162e-06,
+      "loss": 0.2472,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 0.9413,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.87709497206704e-06,
+      "loss": 1.0146,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 1.1776,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.988826815642458e-06,
+      "loss": 1.0559,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 0.9001,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.100558659217877e-06,
+      "loss": 0.9583,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 1.1353,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.212290502793296e-06,
+      "loss": 0.9371,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 0.9634,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.324022346368715e-06,
+      "loss": 1.0425,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 0.9869,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.435754189944135e-06,
+      "loss": 0.8885,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 1.0297,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.547486033519554e-06,
+      "loss": 1.0448,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 1.0042,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.659217877094973e-06,
+      "loss": 1.0992,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 0.969,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.770949720670392e-06,
+      "loss": 1.0782,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 1.0061,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88268156424581e-06,
+      "loss": 0.9968,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 1.0381,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.99441340782123e-06,
+      "loss": 0.96,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 0.2186,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.106145251396648e-06,
+      "loss": 1.0794,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 0.9792,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.217877094972067e-06,
+      "loss": 0.2267,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 0.9027,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.329608938547486e-06,
+      "loss": 1.0019,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 1.0964,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.441340782122905e-06,
+      "loss": 0.2019,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 0.927,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.553072625698325e-06,
+      "loss": 1.0306,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 1.0558,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.664804469273744e-06,
+      "loss": 1.1042,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 1.0055,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.776536312849163e-06,
+      "loss": 1.0531,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 1.0338,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.888268156424582e-06,
+      "loss": 1.0214,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.9941,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9323,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.005586592178771e-05,
+      "loss": 1.0741,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0111731843575421e-05,
+      "loss": 0.1802,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0167597765363129e-05,
+      "loss": 0.9972,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0223463687150838e-05,
+      "loss": 1.0266,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0279329608938548e-05,
+      "loss": 0.9546,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0335195530726259e-05,
+      "loss": 0.9966,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0391061452513967e-05,
+      "loss": 1.0471,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0446927374301676e-05,
+      "loss": 0.9982,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 1.0186,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0558659217877097e-05,
+      "loss": 1.035,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0614525139664806e-05,
+      "loss": 1.0549,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0670391061452514e-05,
+      "loss": 1.0567,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0726256983240225e-05,
+      "loss": 1.0139,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0782122905027934e-05,
+      "loss": 1.0113,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0837988826815644e-05,
+      "loss": 0.9035,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0893854748603351e-05,
+      "loss": 0.9385,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0949720670391063e-05,
+      "loss": 0.8853,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1005586592178772e-05,
+      "loss": 1.0278,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 0.9733,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1117318435754193e-05,
+      "loss": 0.947,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.11731843575419e-05,
+      "loss": 1.0651,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.122905027932961e-05,
+      "loss": 1.0206,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1284916201117319e-05,
+      "loss": 1.0126,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.134078212290503e-05,
+      "loss": 0.9073,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1396648044692738e-05,
+      "loss": 1.0903,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1452513966480447e-05,
+      "loss": 0.1832,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1508379888268157e-05,
+      "loss": 0.9412,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1564245810055868e-05,
+      "loss": 1.0425,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 1.0476,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1675977653631285e-05,
+      "loss": 1.0948,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1731843575418994e-05,
+      "loss": 0.965,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1787709497206706e-05,
+      "loss": 1.0294,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1843575418994415e-05,
+      "loss": 1.0488,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1899441340782123e-05,
+      "loss": 0.9868,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1955307262569834e-05,
+      "loss": 0.9192,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2011173184357543e-05,
+      "loss": 0.9583,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2067039106145253e-05,
+      "loss": 1.0631,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2122905027932964e-05,
+      "loss": 0.1844,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 0.9894,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2234636871508381e-05,
+      "loss": 1.0409,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.229050279329609e-05,
+      "loss": 1.0217,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2346368715083801e-05,
+      "loss": 1.0154,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2402234636871509e-05,
+      "loss": 0.9154,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2458100558659219e-05,
+      "loss": 1.006,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2513966480446928e-05,
+      "loss": 0.9552,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2569832402234639e-05,
+      "loss": 0.9235,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625698324022347e-05,
+      "loss": 0.9416,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2681564245810056e-05,
+      "loss": 0.9946,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 0.9341,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2793296089385477e-05,
+      "loss": 0.9343,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2849162011173186e-05,
+      "loss": 0.9567,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2905027932960894e-05,
+      "loss": 1.0067,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2960893854748603e-05,
+      "loss": 1.1058,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3016759776536314e-05,
+      "loss": 0.9515,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3072625698324024e-05,
+      "loss": 0.9506,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3128491620111732e-05,
+      "loss": 1.0134,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3184357541899443e-05,
+      "loss": 0.9636,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3240223463687152e-05,
+      "loss": 0.959,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 0.861,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3351955307262573e-05,
+      "loss": 0.9879,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.340782122905028e-05,
+      "loss": 1.0152,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.346368715083799e-05,
+      "loss": 1.0519,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3519553072625699e-05,
+      "loss": 0.1983,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.357541899441341e-05,
+      "loss": 0.9999,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3631284916201118e-05,
+      "loss": 0.9776,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3687150837988827e-05,
+      "loss": 1.049,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3743016759776537e-05,
+      "loss": 1.0319,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3798882681564248e-05,
+      "loss": 0.9575,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 0.9151,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3910614525139665e-05,
+      "loss": 1.0236,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3966480446927374e-05,
+      "loss": 0.9507,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4022346368715086e-05,
+      "loss": 0.899,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4078212290502795e-05,
+      "loss": 0.9309,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4134078212290503e-05,
+      "loss": 1.0593,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4189944134078212e-05,
+      "loss": 1.1875,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4245810055865923e-05,
+      "loss": 1.1699,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4301675977653633e-05,
+      "loss": 0.9709,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4357541899441342e-05,
+      "loss": 0.9304,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 1.0573,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4469273743016761e-05,
+      "loss": 1.0144,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.452513966480447e-05,
+      "loss": 1.008,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4581005586592181e-05,
+      "loss": 0.9932,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.463687150837989e-05,
+      "loss": 0.9238,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4692737430167599e-05,
+      "loss": 0.9255,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4748603351955308e-05,
+      "loss": 0.964,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4804469273743019e-05,
+      "loss": 0.1935,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4860335195530729e-05,
+      "loss": 1.1248,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4916201117318436e-05,
+      "loss": 1.0125,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 0.9567,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5027932960893857e-05,
+      "loss": 0.9805,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083798882681566e-05,
+      "loss": 1.0103,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5139664804469274e-05,
+      "loss": 1.0956,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5195530726256983e-05,
+      "loss": 1.0938,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5251396648044694e-05,
+      "loss": 0.9764,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5307262569832404e-05,
+      "loss": 1.0169,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5363128491620113e-05,
+      "loss": 0.9593,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5418994413407823e-05,
+      "loss": 0.9509,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5474860335195532e-05,
+      "loss": 1.0337,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 1.0392,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.558659217877095e-05,
+      "loss": 1.0789,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564245810055866e-05,
+      "loss": 1.0124,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.569832402234637e-05,
+      "loss": 0.8999,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575418994413408e-05,
+      "loss": 1.0383,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.581005586592179e-05,
+      "loss": 1.0325,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5865921787709498e-05,
+      "loss": 0.8527,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5921787709497207e-05,
+      "loss": 1.015,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5977653631284917e-05,
+      "loss": 1.0366,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6033519553072626e-05,
+      "loss": 0.9469,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 0.9527,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6145251396648045e-05,
+      "loss": 1.0765,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6201117318435755e-05,
+      "loss": 0.2142,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6256983240223464e-05,
+      "loss": 1.1231,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6312849162011177e-05,
+      "loss": 1.1407,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6368715083798886e-05,
+      "loss": 1.051,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6424581005586592e-05,
+      "loss": 1.0164,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64804469273743e-05,
+      "loss": 0.9084,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6536312849162014e-05,
+      "loss": 0.9345,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6592178770949724e-05,
+      "loss": 0.9162,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 0.1864,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.670391061452514e-05,
+      "loss": 1.0709,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6759776536312852e-05,
+      "loss": 0.8969,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.681564245810056e-05,
+      "loss": 0.9988,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.687150837988827e-05,
+      "loss": 0.951,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6927374301675977e-05,
+      "loss": 1.044,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698324022346369e-05,
+      "loss": 0.9243,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.70391061452514e-05,
+      "loss": 0.1826,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.709497206703911e-05,
+      "loss": 1.0227,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7150837988826815e-05,
+      "loss": 0.9631,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 1.0391,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7262569832402237e-05,
+      "loss": 0.2273,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7318435754189946e-05,
+      "loss": 0.9402,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7374301675977656e-05,
+      "loss": 1.0353,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7430167597765365e-05,
+      "loss": 1.0124,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7486033519553075e-05,
+      "loss": 0.993,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7541899441340784e-05,
+      "loss": 1.0223,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7597765363128493e-05,
+      "loss": 1.0243,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7653631284916203e-05,
+      "loss": 0.9974,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7709497206703912e-05,
+      "loss": 0.9304,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 1.032,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.782122905027933e-05,
+      "loss": 1.0842,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.787709497206704e-05,
+      "loss": 0.9783,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.793296089385475e-05,
+      "loss": 0.9203,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.798882681564246e-05,
+      "loss": 0.9897,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.804469273743017e-05,
+      "loss": 0.9505,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8100558659217878e-05,
+      "loss": 1.0161,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8156424581005588e-05,
+      "loss": 1.0541,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8212290502793297e-05,
+      "loss": 1.0022,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8268156424581006e-05,
+      "loss": 1.0707,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 1.0286,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8379888268156425e-05,
+      "loss": 1.0089,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8435754189944135e-05,
+      "loss": 1.0583,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8491620111731844e-05,
+      "loss": 0.9652,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8547486033519553e-05,
+      "loss": 0.9785,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8603351955307266e-05,
+      "loss": 1.0436,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8659217877094972e-05,
+      "loss": 0.9847,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.871508379888268e-05,
+      "loss": 1.0315,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8770949720670394e-05,
+      "loss": 0.9165,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8826815642458104e-05,
+      "loss": 1.0304,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 1.087,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.893854748603352e-05,
+      "loss": 1.0355,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8994413407821232e-05,
+      "loss": 1.0431,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.905027932960894e-05,
+      "loss": 1.0288,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.910614525139665e-05,
+      "loss": 1.0094,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9162011173184357e-05,
+      "loss": 1.142,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.921787709497207e-05,
+      "loss": 0.9099,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.927374301675978e-05,
+      "loss": 0.9614,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.932960893854749e-05,
+      "loss": 0.7849,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9385474860335195e-05,
+      "loss": 1.0167,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 1.036,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9497206703910617e-05,
+      "loss": 0.9667,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9553072625698326e-05,
+      "loss": 0.9516,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9608938547486036e-05,
+      "loss": 1.0268,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9664804469273745e-05,
+      "loss": 0.9991,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9720670391061455e-05,
+      "loss": 1.0234,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776536312849164e-05,
+      "loss": 1.095,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9832402234636873e-05,
+      "loss": 0.9758,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9888268156424583e-05,
+      "loss": 0.9713,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944134078212292e-05,
+      "loss": 1.0102,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8511,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999631295782e-05,
+      "loss": 0.8865,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999852518316e-05,
+      "loss": 1.0088,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999668166221e-05,
+      "loss": 0.9451,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994100733073e-05,
+      "loss": 0.9277,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999990782395935e-05,
+      "loss": 1.0489,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986726651043e-05,
+      "loss": 0.2171,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981933498694e-05,
+      "loss": 0.998,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976402939244e-05,
+      "loss": 0.8945,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999701349731e-05,
+      "loss": 1.0044,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963129600727e-05,
+      "loss": 0.8466,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999955386822632e-05,
+      "loss": 0.9487,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999469066394e-05,
+      "loss": 0.2525,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999937689051642e-05,
+      "loss": 1.0604,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999992773406005e-05,
+      "loss": 1.1009,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999917041665352e-05,
+      "loss": 0.2159,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999905611868336e-05,
+      "loss": 0.981,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999893444669843e-05,
+      "loss": 0.9868,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988054007078e-05,
+      "loss": 1.1362,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999866898072088e-05,
+      "loss": 0.9766,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999852518674777e-05,
+      "loss": 0.9077,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999983740187991e-05,
+      "loss": 1.0782,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999821547688595e-05,
+      "loss": 1.1466,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999804956102004e-05,
+      "loss": 0.9246,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999787627121367e-05,
+      "loss": 0.9506,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999769560747954e-05,
+      "loss": 0.1804,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999750756983097e-05,
+      "loss": 0.9555,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973121582819e-05,
+      "loss": 0.9614,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999710937284665e-05,
+      "loss": 1.0695,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999689921354025e-05,
+      "loss": 0.8755,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999668168037815e-05,
+      "loss": 0.9574,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999964567733764e-05,
+      "loss": 0.9211,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999622449255156e-05,
+      "loss": 0.979,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999598483792086e-05,
+      "loss": 0.2602,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999573780950185e-05,
+      "loss": 1.0073,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999954834073128e-05,
+      "loss": 0.9483,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999522163137245e-05,
+      "loss": 0.9712,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999495248170015e-05,
+      "loss": 1.0108,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999946759583157e-05,
+      "loss": 0.9598,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943920612395e-05,
+      "loss": 1.1141,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999941007904925e-05,
+      "loss": 0.9129,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938021460962e-05,
+      "loss": 0.9951,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999349612807254e-05,
+      "loss": 0.9501,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999318273644418e-05,
+      "loss": 0.2242,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999286197123417e-05,
+      "loss": 1.0935,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999253383246618e-05,
+      "loss": 0.9779,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999219832016444e-05,
+      "loss": 0.8792,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999185543435365e-05,
+      "loss": 1.1296,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915051750591e-05,
+      "loss": 1.0471,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911475423066e-05,
+      "loss": 1.0179,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999078253612255e-05,
+      "loss": 0.9494,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999041015653387e-05,
+      "loss": 0.9893,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999003040356804e-05,
+      "loss": 0.9953,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998964327725302e-05,
+      "loss": 1.046,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892487776174e-05,
+      "loss": 1.0414,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888469046902e-05,
+      "loss": 0.9698,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998843765850108e-05,
+      "loss": 0.9856,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999880210390803e-05,
+      "loss": 1.0179,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998759704645845e-05,
+      "loss": 1.0607,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999871656806669e-05,
+      "loss": 1.0649,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998672694173738e-05,
+      "loss": 0.9846,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998628082970237e-05,
+      "loss": 1.0195,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999858273445946e-05,
+      "loss": 0.9993,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998536648644763e-05,
+      "loss": 0.9575,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999848982552954e-05,
+      "loss": 0.9288,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998442265117243e-05,
+      "loss": 0.9746,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839396741138e-05,
+      "loss": 0.982,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998344932415517e-05,
+      "loss": 0.9866,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998295160133264e-05,
+      "loss": 0.9655,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999824465056829e-05,
+      "loss": 0.9875,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998193403724328e-05,
+      "loss": 0.8723,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999814141960515e-05,
+      "loss": 1.022,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999808869821459e-05,
+      "loss": 1.0028,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998035239556536e-05,
+      "loss": 1.0612,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999798104363493e-05,
+      "loss": 0.9579,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997926110453772e-05,
+      "loss": 0.9913,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997870440017114e-05,
+      "loss": 0.984,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999781403232905e-05,
+      "loss": 0.9814,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997756887393752e-05,
+      "loss": 1.0222,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997699005215426e-05,
+      "loss": 1.0015,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997640385798346e-05,
+      "loss": 1.1377,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997581029146825e-05,
+      "loss": 0.9767,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997520935265253e-05,
+      "loss": 0.2284,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997460104158056e-05,
+      "loss": 0.976,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997398535829716e-05,
+      "loss": 0.9818,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999733623028478e-05,
+      "loss": 0.9521,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999727318752783e-05,
+      "loss": 0.8369,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999720940756353e-05,
+      "loss": 0.9949,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997144890396574e-05,
+      "loss": 1.1124,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997079636031725e-05,
+      "loss": 0.9807,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999701364447379e-05,
+      "loss": 0.95,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999694691572764e-05,
+      "loss": 0.9291,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999687944979819e-05,
+      "loss": 1.0848,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996811246690416e-05,
+      "loss": 0.947,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999674230640935e-05,
+      "loss": 1.0687,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996672628960074e-05,
+      "loss": 0.9574,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999660221434773e-05,
+      "loss": 0.9766,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996531062577505e-05,
+      "loss": 0.9326,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999645917365465e-05,
+      "loss": 0.978,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996386547584465e-05,
+      "loss": 1.0005,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996313184372306e-05,
+      "loss": 0.9795,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996239084023578e-05,
+      "loss": 1.1227,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999616424654375e-05,
+      "loss": 0.9376,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996088671938342e-05,
+      "loss": 0.9509,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996012360212923e-05,
+      "loss": 0.9537,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999593531137312e-05,
+      "loss": 0.9417,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995857525424615e-05,
+      "loss": 0.9603,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999577900237315e-05,
+      "loss": 1.0472,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999569974222451e-05,
+      "loss": 0.9242,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999561974498454e-05,
+      "loss": 0.9957,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995539010659136e-05,
+      "loss": 1.0182,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999545753925426e-05,
+      "loss": 0.9059,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995375330775914e-05,
+      "loss": 0.913,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995292385230157e-05,
+      "loss": 0.9145,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995208702623113e-05,
+      "loss": 0.9512,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995124282960944e-05,
+      "loss": 0.9164,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995039126249886e-05,
+      "loss": 0.9171,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999495323249621e-05,
+      "loss": 0.8273,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994866601706257e-05,
+      "loss": 1.0206,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994779233886406e-05,
+      "loss": 1.0072,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994691129043107e-05,
+      "loss": 1.051,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994602287182854e-05,
+      "loss": 1.0446,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994512708312196e-05,
+      "loss": 0.9969,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994422392437744e-05,
+      "loss": 1.0282,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994331339566155e-05,
+      "loss": 0.2463,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994239549704148e-05,
+      "loss": 1.0439,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994147022858483e-05,
+      "loss": 0.9891,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994053759035986e-05,
+      "loss": 0.9283,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999395975824354e-05,
+      "loss": 1.0218,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993865020488066e-05,
+      "loss": 1.0289,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999376954577656e-05,
+      "loss": 0.9503,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993673334116057e-05,
+      "loss": 1.0353,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993576385513655e-05,
+      "loss": 1.0264,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993478699976502e-05,
+      "loss": 0.9985,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99933802775118e-05,
+      "loss": 0.9885,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993281118126807e-05,
+      "loss": 0.8684,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999318122182884e-05,
+      "loss": 0.9575,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993080588625257e-05,
+      "loss": 0.2259,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992979218523488e-05,
+      "loss": 1.0464,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992877111531e-05,
+      "loss": 0.216,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992774267655325e-05,
+      "loss": 0.9855,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999267068690405e-05,
+      "loss": 0.9811,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992566369284806e-05,
+      "loss": 0.9935,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992461314805293e-05,
+      "loss": 0.9776,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999235552347325e-05,
+      "loss": 1.0547,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999224899529649e-05,
+      "loss": 0.9183,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999214173028286e-05,
+      "loss": 0.9099,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999203372844027e-05,
+      "loss": 0.9727,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991924989776684e-05,
+      "loss": 1.0674,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991815514300125e-05,
+      "loss": 0.9604,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999170530201866e-05,
+      "loss": 0.9193,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999159435294042e-05,
+      "loss": 0.9907,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991482667073585e-05,
+      "loss": 1.0276,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991370244426394e-05,
+      "loss": 0.8919,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999125708500713e-05,
+      "loss": 0.214,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991143188824146e-05,
+      "loss": 1.0629,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999102855588583e-05,
+      "loss": 0.9216,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990913186200648e-05,
+      "loss": 1.0583,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99907970797771e-05,
+      "loss": 0.9293,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990680236623746e-05,
+      "loss": 0.9719,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999056265674921e-05,
+      "loss": 0.9272,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999044434016215e-05,
+      "loss": 1.0649,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990325286871305e-05,
+      "loss": 0.9528,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990205496885445e-05,
+      "loss": 0.8737,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990084970213405e-05,
+      "loss": 1.006,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989963706864074e-05,
+      "loss": 0.947,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989841706846394e-05,
+      "loss": 1.0011,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998971897016936e-05,
+      "loss": 0.9615,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989595496842024e-05,
+      "loss": 0.9735,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998947128687349e-05,
+      "loss": 0.9527,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989346340272923e-05,
+      "loss": 0.9874,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989220657049525e-05,
+      "loss": 0.9917,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989094237212575e-05,
+      "loss": 0.8891,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988967080771392e-05,
+      "loss": 0.9943,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988839187735346e-05,
+      "loss": 1.0452,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998871055811388e-05,
+      "loss": 0.9387,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988581191916468e-05,
+      "loss": 0.9558,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988451089152657e-05,
+      "loss": 0.948,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988320249832037e-05,
+      "loss": 1.1246,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988188673964255e-05,
+      "loss": 0.8971,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988056361559018e-05,
+      "loss": 1.0206,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987923312626083e-05,
+      "loss": 0.9556,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987789527175256e-05,
+      "loss": 1.1045,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987655005216407e-05,
+      "loss": 1.032,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987519746759456e-05,
+      "loss": 0.9582,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998738375181437e-05,
+      "loss": 1.0119,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987247020391186e-05,
+      "loss": 0.9108,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987109552499986e-05,
+      "loss": 0.9829,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99869713481509e-05,
+      "loss": 1.0486,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986832407354123e-05,
+      "loss": 1.0211,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986692730119904e-05,
+      "loss": 0.9303,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998655231645854e-05,
+      "loss": 0.9414,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986411166380387e-05,
+      "loss": 0.9433,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986269279895848e-05,
+      "loss": 0.9431,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986126657015393e-05,
+      "loss": 1.1189,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985983297749535e-05,
+      "loss": 0.1881,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985839202108845e-05,
+      "loss": 0.9791,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998569437010395e-05,
+      "loss": 0.9786,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998554880174553e-05,
+      "loss": 1.0371,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998540249704432e-05,
+      "loss": 1.0003,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985255456011107e-05,
+      "loss": 0.9661,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985107678656738e-05,
+      "loss": 1.1008,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984959164992105e-05,
+      "loss": 1.0273,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998480991502816e-05,
+      "loss": 0.9847,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984659928775912e-05,
+      "loss": 1.0858,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998450920624642e-05,
+      "loss": 0.9858,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99843577474508e-05,
+      "loss": 1.0214,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984205552400215e-05,
+      "loss": 0.9871,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984052621105895e-05,
+      "loss": 0.8939,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998389895357911e-05,
+      "loss": 0.9216,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983744549831197e-05,
+      "loss": 0.9969,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983589409873544e-05,
+      "loss": 0.938,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983433533717584e-05,
+      "loss": 1.0426,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983276921374817e-05,
+      "loss": 1.0259,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998311957285679e-05,
+      "loss": 0.9297,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982961488175105e-05,
+      "loss": 0.9515,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998280266734142e-05,
+      "loss": 0.9639,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982643110367443e-05,
+      "loss": 0.9607,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998248281726495e-05,
+      "loss": 0.2147,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998232178804575e-05,
+      "loss": 1.0057,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982160022721724e-05,
+      "loss": 0.9514,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99819975213048e-05,
+      "loss": 0.9783,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981834283806956e-05,
+      "loss": 0.9323,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981670310240237e-05,
+      "loss": 1.0129,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981505600616726e-05,
+      "loss": 1.0705,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981340154948578e-05,
+      "loss": 0.998,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981173973247986e-05,
+      "loss": 1.0316,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981007055527207e-05,
+      "loss": 1.1569,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980839401798552e-05,
+      "loss": 0.9146,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998067101207438e-05,
+      "loss": 0.9264,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998050188636711e-05,
+      "loss": 0.8957,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998033202468921e-05,
+      "loss": 0.9365,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980161427053215e-05,
+      "loss": 1.0128,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979990093471696e-05,
+      "loss": 0.8971,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997981802395729e-05,
+      "loss": 1.038,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979645218522687e-05,
+      "loss": 0.9581,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979471677180627e-05,
+      "loss": 1.0435,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979297399943908e-05,
+      "loss": 0.9789,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997912238682538e-05,
+      "loss": 1.0222,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978946637837956e-05,
+      "loss": 1.0842,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978770152994588e-05,
+      "loss": 0.8684,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997859293230829e-05,
+      "loss": 0.8847,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978414975792137e-05,
+      "loss": 0.931,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978236283459247e-05,
+      "loss": 0.9781,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99780568553228e-05,
+      "loss": 1.0065,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977876691396022e-05,
+      "loss": 0.9761,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99776957916922e-05,
+      "loss": 0.9651,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977514156224678e-05,
+      "loss": 1.0311,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977331785006843e-05,
+      "loss": 0.9001,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997714867805215e-05,
+      "loss": 0.9962,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976964835374098e-05,
+      "loss": 0.9594,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976780256986242e-05,
+      "loss": 0.9764,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99765949429022e-05,
+      "loss": 1.1146,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976408893135632e-05,
+      "loss": 0.9825,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976222107700253e-05,
+      "loss": 1.0084,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976034586609848e-05,
+      "loss": 0.9365,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975846329878233e-05,
+      "loss": 1.1001,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975657337519297e-05,
+      "loss": 1.0435,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975467609546978e-05,
+      "loss": 0.924,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997527714597526e-05,
+      "loss": 0.9444,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99750859468182e-05,
+      "loss": 0.9858,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974894012089882e-05,
+      "loss": 1.0295,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997470134180447e-05,
+      "loss": 0.9588,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974507935976163e-05,
+      "loss": 0.9836,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997431379461923e-05,
+      "loss": 1.0393,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997411891774799e-05,
+      "loss": 1.0415,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973923305376806e-05,
+      "loss": 0.9689,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973726957520104e-05,
+      "loss": 1.0856,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973529874192365e-05,
+      "loss": 0.914,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997333205540812e-05,
+      "loss": 1.0767,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997313350118196e-05,
+      "loss": 1.005,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972934211528524e-05,
+      "loss": 1.1042,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972734186462505e-05,
+      "loss": 0.9641,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972533425998655e-05,
+      "loss": 1.0929,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972331930151782e-05,
+      "loss": 1.0912,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972129698936742e-05,
+      "loss": 0.8552,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971926732368446e-05,
+      "loss": 0.9091,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971723030461866e-05,
+      "loss": 1.0519,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971518593232016e-05,
+      "loss": 1.063,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971313420693975e-05,
+      "loss": 0.9323,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971107512862873e-05,
+      "loss": 0.9432,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970900869753893e-05,
+      "loss": 0.9301,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997069349138227e-05,
+      "loss": 0.9947,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970485377763308e-05,
+      "loss": 0.9394,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970276528912338e-05,
+      "loss": 1.0506,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997006694484477e-05,
+      "loss": 1.0154,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969856625576054e-05,
+      "loss": 1.0201,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969645571121705e-05,
+      "loss": 1.0029,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996943378149728e-05,
+      "loss": 0.9859,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969221256718403e-05,
+      "loss": 0.2934,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996900799680074e-05,
+      "loss": 0.9776,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996879400176002e-05,
+      "loss": 0.9455,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996857927161202e-05,
+      "loss": 1.0053,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996836380637258e-05,
+      "loss": 0.2017,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968147606057588e-05,
+      "loss": 1.0471,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996793067068298e-05,
+      "loss": 0.9258,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967713000264756e-05,
+      "loss": 0.9193,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996749459481897e-05,
+      "loss": 0.9431,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967275454361727e-05,
+      "loss": 0.94,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967055578909185e-05,
+      "loss": 0.9791,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966834968477556e-05,
+      "loss": 1.0593,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966613623083113e-05,
+      "loss": 0.9387,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966391542742176e-05,
+      "loss": 0.9892,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966168727471117e-05,
+      "loss": 1.0249,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965945177286375e-05,
+      "loss": 0.9173,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965720892204427e-05,
+      "loss": 0.9748,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965495872241817e-05,
+      "loss": 0.9711,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965270117415136e-05,
+      "loss": 0.9606,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996504362774103e-05,
+      "loss": 1.0236,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964816403236204e-05,
+      "loss": 0.8964,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964588443917414e-05,
+      "loss": 0.9995,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964359749801462e-05,
+      "loss": 0.9722,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996413032090522e-05,
+      "loss": 0.9777,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996390015724561e-05,
+      "loss": 0.9806,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996366925883959e-05,
+      "loss": 1.0576,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99634376257042e-05,
+      "loss": 1.0513,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963205257856513e-05,
+      "loss": 1.0161,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962972155313674e-05,
+      "loss": 0.978,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996273831809286e-05,
+      "loss": 0.9362,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996250374621132e-05,
+      "loss": 0.9199,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962268439686352e-05,
+      "loss": 1.0559,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962032398535308e-05,
+      "loss": 0.9357,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996179562277559e-05,
+      "loss": 0.9609,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996155811242466e-05,
+      "loss": 0.9544,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961319867500038e-05,
+      "loss": 0.9054,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961080888019286e-05,
+      "loss": 0.903,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960841174000025e-05,
+      "loss": 0.946,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960600725459938e-05,
+      "loss": 0.9235,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960359542416752e-05,
+      "loss": 0.2802,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960117624888252e-05,
+      "loss": 0.9777,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959874972892276e-05,
+      "loss": 1.0732,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959631586446723e-05,
+      "loss": 1.0082,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959387465569534e-05,
+      "loss": 0.9568,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959142610278716e-05,
+      "loss": 1.1384,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958897020592318e-05,
+      "loss": 0.9394,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958650696528456e-05,
+      "loss": 0.9392,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995840363810529e-05,
+      "loss": 1.0902,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958155845341048e-05,
+      "loss": 0.8555,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995790731825399e-05,
+      "loss": 1.0003,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957658056862445e-05,
+      "loss": 1.0297,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957408061184803e-05,
+      "loss": 1.028,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957157331239486e-05,
+      "loss": 0.9472,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956905867044994e-05,
+      "loss": 0.8709,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956653668619863e-05,
+      "loss": 0.8828,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956400735982694e-05,
+      "loss": 0.9084,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956147069152138e-05,
+      "loss": 0.9546,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955892668146898e-05,
+      "loss": 0.9525,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955637532985738e-05,
+      "loss": 1.0187,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995538166368747e-05,
+      "loss": 0.9867,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955125060270958e-05,
+      "loss": 0.9533,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954867722755125e-05,
+      "loss": 1.0043,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995460965115896e-05,
+      "loss": 0.9352,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954350845501474e-05,
+      "loss": 0.2012,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954091305801765e-05,
+      "loss": 0.9973,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995383103207897e-05,
+      "loss": 1.0732,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953570024352275e-05,
+      "loss": 0.8756,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953308282640935e-05,
+      "loss": 0.9513,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953045806964247e-05,
+      "loss": 0.8346,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952782597341564e-05,
+      "loss": 1.0135,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952518653792303e-05,
+      "loss": 1.0099,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952253976335915e-05,
+      "loss": 1.0122,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995198856499193e-05,
+      "loss": 1.0134,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951722419779914e-05,
+      "loss": 0.928,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951455540719495e-05,
+      "loss": 0.8479,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951187927830347e-05,
+      "loss": 0.8902,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995091958113221e-05,
+      "loss": 1.0134,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995065050064487e-05,
+      "loss": 1.1048,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995038068638817e-05,
+      "loss": 0.9698,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950110138382007e-05,
+      "loss": 0.9449,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949838856646328e-05,
+      "loss": 0.9886,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949566841201142e-05,
+      "loss": 0.9824,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949294092066504e-05,
+      "loss": 0.8604,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949020609262528e-05,
+      "loss": 0.9515,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948746392809385e-05,
+      "loss": 0.9525,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948471442727287e-05,
+      "loss": 1.0404,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948195759036517e-05,
+      "loss": 1.0797,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99479193417574e-05,
+      "loss": 1.0457,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994764219091032e-05,
+      "loss": 0.9869,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947364306515714e-05,
+      "loss": 0.9461,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994708568859408e-05,
+      "loss": 0.9599,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994680633716595e-05,
+      "loss": 1.0727,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946526252251935e-05,
+      "loss": 1.0029,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946245433872687e-05,
+      "loss": 0.9366,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994596388204891e-05,
+      "loss": 0.9489,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994568159680137e-05,
+      "loss": 1.049,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945398578150878e-05,
+      "loss": 1.0037,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945114826118306e-05,
+      "loss": 0.9037,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994483034072458e-05,
+      "loss": 0.9895,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944545121990676e-05,
+      "loss": 0.934,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994425916993763e-05,
+      "loss": 0.9524,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943972484586523e-05,
+      "loss": 1.024,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99436850659585e-05,
+      "loss": 0.9705,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943396914074753e-05,
+      "loss": 0.9632,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943108028956534e-05,
+      "loss": 0.9437,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942818410625134e-05,
+      "loss": 0.9668,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942528059101928e-05,
+      "loss": 0.8692,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942236974408314e-05,
+      "loss": 0.9556,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994194515656576e-05,
+      "loss": 0.9496,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941652605595782e-05,
+      "loss": 0.9832,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994135932151996e-05,
+      "loss": 0.9357,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941065304359916e-05,
+      "loss": 0.9803,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994077055413733e-05,
+      "loss": 0.2228,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940475070873942e-05,
+      "loss": 0.9936,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940178854591536e-05,
+      "loss": 0.9208,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939881905311958e-05,
+      "loss": 0.9403,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939584223057105e-05,
+      "loss": 0.9452,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939285807848928e-05,
+      "loss": 1.0959,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993898665970943e-05,
+      "loss": 1.0536,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993868677866068e-05,
+      "loss": 0.9944,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938386164724778e-05,
+      "loss": 1.0456,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938084817923902e-05,
+      "loss": 1.0389,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937782738280268e-05,
+      "loss": 0.9891,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937479925816152e-05,
+      "loss": 0.9642,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993717638055389e-05,
+      "loss": 0.9972,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936872102515854e-05,
+      "loss": 0.9568,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993656709172449e-05,
+      "loss": 0.8977,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993626134820229e-05,
+      "loss": 1.0329,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935954871971797e-05,
+      "loss": 0.8493,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993564766305561e-05,
+      "loss": 1.0132,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935339721476385e-05,
+      "loss": 1.0157,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935031047256828e-05,
+      "loss": 1.0026,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993472164041971e-05,
+      "loss": 1.0203,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934411500987832e-05,
+      "loss": 1.0529,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993410062898407e-05,
+      "loss": 0.2273,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933789024431353e-05,
+      "loss": 1.0001,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933476687352653e-05,
+      "loss": 0.8804,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933163617771005e-05,
+      "loss": 0.9829,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932849815709493e-05,
+      "loss": 0.9213,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932535281191257e-05,
+      "loss": 0.2539,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932220014239495e-05,
+      "loss": 1.0428,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993190401487745e-05,
+      "loss": 1.0364,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993158728312843e-05,
+      "loss": 0.9761,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931269819015783e-05,
+      "loss": 1.0056,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930951622562923e-05,
+      "loss": 1.0693,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930632693793316e-05,
+      "loss": 0.9057,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930313032730478e-05,
+      "loss": 0.9974,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929992639397984e-05,
+      "loss": 0.8646,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929671513819456e-05,
+      "loss": 0.9823,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929349656018574e-05,
+      "loss": 0.9646,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992902706601908e-05,
+      "loss": 1.0334,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928703743844747e-05,
+      "loss": 1.0447,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928379689519432e-05,
+      "loss": 0.9466,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928054903067023e-05,
+      "loss": 0.9008,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927729384511472e-05,
+      "loss": 1.0204,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927403133876783e-05,
+      "loss": 0.9336,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992707615118701e-05,
+      "loss": 0.9948,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926748436466274e-05,
+      "loss": 0.9123,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926419989738735e-05,
+      "loss": 0.9459,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926090811028614e-05,
+      "loss": 1.0032,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925760900360185e-05,
+      "loss": 0.1948,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925430257757777e-05,
+      "loss": 0.9647,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992509888324577e-05,
+      "loss": 0.973,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924766776848597e-05,
+      "loss": 1.098,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924433938590753e-05,
+      "loss": 0.9748,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992410036849678e-05,
+      "loss": 0.9774,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923766066591278e-05,
+      "loss": 0.9732,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923431032898893e-05,
+      "loss": 1.0328,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923095267444333e-05,
+      "loss": 1.0293,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922758770252362e-05,
+      "loss": 1.0015,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992242154134779e-05,
+      "loss": 0.2425,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922083580755483e-05,
+      "loss": 0.9565,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921744888500366e-05,
+      "loss": 1.0048,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921405464607413e-05,
+      "loss": 0.9588,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992106530910165e-05,
+      "loss": 0.9991,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920724422008165e-05,
+      "loss": 0.9111,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920382803352096e-05,
+      "loss": 1.0562,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920040453158627e-05,
+      "loss": 0.9816,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919697371453014e-05,
+      "loss": 0.9161,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991935355826055e-05,
+      "loss": 1.1249,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919009013606586e-05,
+      "loss": 0.9886,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918663737516532e-05,
+      "loss": 1.006,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991831773001585e-05,
+      "loss": 0.9829,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917970991130053e-05,
+      "loss": 0.9667,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917623520884707e-05,
+      "loss": 0.869,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991727531930544e-05,
+      "loss": 0.969,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916926386417928e-05,
+      "loss": 0.907,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916576722247897e-05,
+      "loss": 0.9232,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991622632682114e-05,
+      "loss": 1.0111,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915875200163482e-05,
+      "loss": 1.0087,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991552334230083e-05,
+      "loss": 1.0324,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915170753259122e-05,
+      "loss": 0.9633,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991481743306436e-05,
+      "loss": 0.8981,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914463381742597e-05,
+      "loss": 0.9162,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914108599319942e-05,
+      "loss": 0.9743,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991375308582256e-05,
+      "loss": 0.9301,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913396841276664e-05,
+      "loss": 0.9708,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991303986570852e-05,
+      "loss": 0.9183,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991268215914446e-05,
+      "loss": 0.9337,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912323721610856e-05,
+      "loss": 1.0117,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911964553134138e-05,
+      "loss": 0.9449,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911604653740796e-05,
+      "loss": 0.9797,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911244023457365e-05,
+      "loss": 0.2413,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910882662310444e-05,
+      "loss": 0.9863,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910520570326674e-05,
+      "loss": 0.9538,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991015774753276e-05,
+      "loss": 1.0054,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909794193955455e-05,
+      "loss": 0.9666,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909429909621566e-05,
+      "loss": 0.9841,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990906489455796e-05,
+      "loss": 0.8981,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908699148791548e-05,
+      "loss": 1.0125,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908332672349307e-05,
+      "loss": 0.973,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907965465258258e-05,
+      "loss": 0.9564,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907597527545476e-05,
+      "loss": 0.9161,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99072288592381e-05,
+      "loss": 0.9354,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906859460363307e-05,
+      "loss": 0.9824,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906489330948346e-05,
+      "loss": 0.9012,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906118471020505e-05,
+      "loss": 1.0459,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905746880607135e-05,
+      "loss": 1.0206,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905374559735634e-05,
+      "loss": 1.0343,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905001508433457e-05,
+      "loss": 0.2615,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904627726728115e-05,
+      "loss": 0.9613,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904253214647175e-05,
+      "loss": 1.0471,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903877972218245e-05,
+      "loss": 1.0373,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903501999469e-05,
+      "loss": 0.9347,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903125296427167e-05,
+      "loss": 0.8879,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990274786312052e-05,
+      "loss": 0.9338,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902369699576894e-05,
+      "loss": 1.0227,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901990805824177e-05,
+      "loss": 0.9942,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901611181890303e-05,
+      "loss": 0.9052,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901230827803274e-05,
+      "loss": 0.8765,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900849743591127e-05,
+      "loss": 1.0224,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900467929281974e-05,
+      "loss": 1.0248,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900085384903963e-05,
+      "loss": 0.2627,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899702110485308e-05,
+      "loss": 0.9763,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899318106054268e-05,
+      "loss": 0.8605,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898933371639162e-05,
+      "loss": 0.9289,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898547907268358e-05,
+      "loss": 0.8921,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898161712970287e-05,
+      "loss": 0.9506,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989777478877342e-05,
+      "loss": 0.9525,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897387134706294e-05,
+      "loss": 1.0489,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896998750797493e-05,
+      "loss": 0.9657,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896609637075658e-05,
+      "loss": 0.9383,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896219793569478e-05,
+      "loss": 1.0662,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895829220307706e-05,
+      "loss": 0.8335,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989543791731914e-05,
+      "loss": 1.0887,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895045884632636e-05,
+      "loss": 1.0505,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894653122277103e-05,
+      "loss": 0.944,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894259630281503e-05,
+      "loss": 0.2327,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893865408674854e-05,
+      "loss": 0.8817,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893470457486225e-05,
+      "loss": 0.9296,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989307477674474e-05,
+      "loss": 1.0224,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892678366479575e-05,
+      "loss": 0.9409,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989228122671997e-05,
+      "loss": 0.9337,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891883357495197e-05,
+      "loss": 1.0641,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891484758834605e-05,
+      "loss": 1.0161,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891085430767586e-05,
+      "loss": 0.2584,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890685373323583e-05,
+      "loss": 0.2412,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890284586532096e-05,
+      "loss": 0.9887,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889883070422686e-05,
+      "loss": 1.0656,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889480825024953e-05,
+      "loss": 0.9282,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988907785036857e-05,
+      "loss": 0.2303,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988867414648324e-05,
+      "loss": 0.9386,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988826971339874e-05,
+      "loss": 0.9763,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887864551144892e-05,
+      "loss": 1.057,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887458659751568e-05,
+      "loss": 0.8724,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887052039248708e-05,
+      "loss": 0.9459,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988664468966629e-05,
+      "loss": 0.9824,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886236611034357e-05,
+      "loss": 1.0168,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885827803382994e-05,
+      "loss": 1.012,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988541826674235e-05,
+      "loss": 0.9276,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988500800114263e-05,
+      "loss": 0.8921,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884597006614078e-05,
+      "loss": 0.9764,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988418528318701e-05,
+      "loss": 0.9992,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883772830891783e-05,
+      "loss": 0.8319,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988335964975881e-05,
+      "loss": 0.9129,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882945739818558e-05,
+      "loss": 0.2323,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882531101101554e-05,
+      "loss": 1.0895,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988211573363837e-05,
+      "loss": 0.9333,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881699637459642e-05,
+      "loss": 1.0244,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881282812596043e-05,
+      "loss": 0.9235,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880865259078317e-05,
+      "loss": 0.2157,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880446976937257e-05,
+      "loss": 0.9929,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880027966203696e-05,
+      "loss": 1.0225,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879608226908543e-05,
+      "loss": 0.2377,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987918775908275e-05,
+      "loss": 1.0002,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878766562757317e-05,
+      "loss": 0.9073,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878344637963304e-05,
+      "loss": 0.9289,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877921984731826e-05,
+      "loss": 0.9221,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877498603094054e-05,
+      "loss": 1.0875,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98770744930812e-05,
+      "loss": 1.057,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876649654724545e-05,
+      "loss": 1.0353,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987622408805541e-05,
+      "loss": 0.9579,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875797793105187e-05,
+      "loss": 0.9636,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875370769905303e-05,
+      "loss": 1.0789,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874943018487246e-05,
+      "loss": 0.9665,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874514538882567e-05,
+      "loss": 0.9652,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874085331122854e-05,
+      "loss": 0.9437,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873655395239762e-05,
+      "loss": 1.0068,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873224731264994e-05,
+      "loss": 1.0057,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872793339230307e-05,
+      "loss": 0.9346,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987236121916751e-05,
+      "loss": 0.9344,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871928371108477e-05,
+      "loss": 0.9789,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871494795085113e-05,
+      "loss": 0.9983,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98710604911294e-05,
+      "loss": 1.029,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870625459273357e-05,
+      "loss": 0.8836,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870189699549073e-05,
+      "loss": 0.9376,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869753211988673e-05,
+      "loss": 0.9073,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869315996624347e-05,
+      "loss": 1.0381,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868878053488337e-05,
+      "loss": 0.9541,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868439382612935e-05,
+      "loss": 0.9453,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867999984030488e-05,
+      "loss": 0.9554,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98675598577734e-05,
+      "loss": 0.9679,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867119003874126e-05,
+      "loss": 1.0238,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866677422365174e-05,
+      "loss": 0.9165,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866235113279112e-05,
+      "loss": 0.9585,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865792076648547e-05,
+      "loss": 1.0667,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986534831250615e-05,
+      "loss": 1.0093,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864903820884652e-05,
+      "loss": 1.0438,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986445860181683e-05,
+      "loss": 1.017,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864012655335504e-05,
+      "loss": 1.0212,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863565981473568e-05,
+      "loss": 0.9819,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863118580263955e-05,
+      "loss": 0.9278,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862670451739662e-05,
+      "loss": 0.8771,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862221595933732e-05,
+      "loss": 0.8663,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986177201287926e-05,
+      "loss": 0.9897,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861321702609405e-05,
+      "loss": 0.923,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986087066515737e-05,
+      "loss": 0.2297,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860418900556417e-05,
+      "loss": 1.074,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859966408839856e-05,
+      "loss": 0.9461,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859513190041058e-05,
+      "loss": 1.0333,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859059244193438e-05,
+      "loss": 0.8254,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858604571330476e-05,
+      "loss": 0.277,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98581491714857e-05,
+      "loss": 1.0307,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857693044692686e-05,
+      "loss": 0.8915,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857236190985074e-05,
+      "loss": 0.929,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856778610396552e-05,
+      "loss": 0.9462,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856320302960862e-05,
+      "loss": 1.0803,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98558612687118e-05,
+      "loss": 1.0685,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855401507683213e-05,
+      "loss": 0.9637,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854941019909012e-05,
+      "loss": 0.8899,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854479805423145e-05,
+      "loss": 0.9213,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854017864259625e-05,
+      "loss": 0.9396,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853555196452517e-05,
+      "loss": 0.955,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853091802035938e-05,
+      "loss": 0.9385,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852627681044062e-05,
+      "loss": 0.9342,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985216283351111e-05,
+      "loss": 0.9895,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985169725947136e-05,
+      "loss": 0.9728,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985123095895914e-05,
+      "loss": 0.196,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850763932008846e-05,
+      "loss": 0.9736,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850296178654913e-05,
+      "loss": 0.9696,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984982769893183e-05,
+      "loss": 1.0169,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849358492874145e-05,
+      "loss": 1.0594,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848888560516456e-05,
+      "loss": 0.9166,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984841790189342e-05,
+      "loss": 0.9075,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984794651703974e-05,
+      "loss": 0.8636,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847474405990178e-05,
+      "loss": 1.0104,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984700156877955e-05,
+      "loss": 0.9797,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984652800544272e-05,
+      "loss": 1.0211,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846053716014607e-05,
+      "loss": 0.9285,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984557870053019e-05,
+      "loss": 0.9862,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98451029590245e-05,
+      "loss": 0.981,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844626491532614e-05,
+      "loss": 0.9809,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984414929808966e-05,
+      "loss": 0.9577,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984367137873084e-05,
+      "loss": 1.0276,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984319273349139e-05,
+      "loss": 0.8517,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842713362406605e-05,
+      "loss": 0.8054,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842233265511837e-05,
+      "loss": 0.9318,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841752442842485e-05,
+      "loss": 0.966,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841270894434006e-05,
+      "loss": 0.9254,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984078862032191e-05,
+      "loss": 0.9862,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840305620541766e-05,
+      "loss": 1.0195,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839821895129182e-05,
+      "loss": 0.9104,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839337444119835e-05,
+      "loss": 0.9084,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838852267549444e-05,
+      "loss": 1.0325,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838366365453788e-05,
+      "loss": 0.263,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837879737868704e-05,
+      "loss": 1.0594,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837392384830066e-05,
+      "loss": 0.9223,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836904306373817e-05,
+      "loss": 0.9695,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983641550253595e-05,
+      "loss": 0.9377,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835925973352505e-05,
+      "loss": 0.952,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835435718859582e-05,
+      "loss": 0.892,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834944739093336e-05,
+      "loss": 0.9661,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983445303408997e-05,
+      "loss": 0.8627,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983396060388574e-05,
+      "loss": 1.0266,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833467448516965e-05,
+      "loss": 0.9985,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832973568020004e-05,
+      "loss": 0.9868,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983247896243128e-05,
+      "loss": 0.8878,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983198363178726e-05,
+      "loss": 1.0028,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983148757612448e-05,
+      "loss": 0.9455,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830990795479513e-05,
+      "loss": 0.9475,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830493289888993e-05,
+      "loss": 0.9969,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829995059389608e-05,
+      "loss": 0.8789,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982949610401809e-05,
+      "loss": 0.9715,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828996423811246e-05,
+      "loss": 0.9625,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828496018805918e-05,
+      "loss": 1.0022,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827994889038996e-05,
+      "loss": 0.9682,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982749303454745e-05,
+      "loss": 1.0095,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982699045536827e-05,
+      "loss": 0.9982,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982648715153853e-05,
+      "loss": 0.9037,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982598312309534e-05,
+      "loss": 0.2312,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982547837007587e-05,
+      "loss": 1.0167,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824972892517333e-05,
+      "loss": 0.8559,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982446669045701e-05,
+      "loss": 1.0674,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982395976393223e-05,
+      "loss": 1.0479,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823452112980364e-05,
+      "loss": 1.0042,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982294373763886e-05,
+      "loss": 1.0083,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822434637945197e-05,
+      "loss": 0.8691,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982192481393692e-05,
+      "loss": 0.9206,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821414265651628e-05,
+      "loss": 0.967,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820902993126958e-05,
+      "loss": 0.9835,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982039099640062e-05,
+      "loss": 1.0117,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819878275510368e-05,
+      "loss": 0.9017,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819364830494013e-05,
+      "loss": 0.949,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818850661389408e-05,
+      "loss": 0.9914,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818335768234477e-05,
+      "loss": 1.0524,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817820151067184e-05,
+      "loss": 0.9023,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981730380992555e-05,
+      "loss": 0.9953,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981678674484766e-05,
+      "loss": 1.0015,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981626895587163e-05,
+      "loss": 0.9013,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815750443035654e-05,
+      "loss": 0.9973,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815231206377956e-05,
+      "loss": 0.2555,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981471124593683e-05,
+      "loss": 0.8964,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814190561750624e-05,
+      "loss": 0.983,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813669153857726e-05,
+      "loss": 0.8944,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813147022296584e-05,
+      "loss": 0.9419,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981262416710571e-05,
+      "loss": 0.9276,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981210058832365e-05,
+      "loss": 1.0709,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811576285989016e-05,
+      "loss": 0.8676,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811051260140474e-05,
+      "loss": 1.0069,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981052551081674e-05,
+      "loss": 1.0284,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809999038056573e-05,
+      "loss": 0.9374,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809471841898807e-05,
+      "loss": 1.0588,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808943922382315e-05,
+      "loss": 1.0084,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808415279546022e-05,
+      "loss": 1.0712,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980788591342892e-05,
+      "loss": 0.8787,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807355824070033e-05,
+      "loss": 0.9104,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806825011508457e-05,
+      "loss": 0.9293,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806293475783334e-05,
+      "loss": 0.8617,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805761216933858e-05,
+      "loss": 1.0155,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980522823499928e-05,
+      "loss": 1.0129,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98046945300189e-05,
+      "loss": 0.9343,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980416010203208e-05,
+      "loss": 0.8716,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803624951078223e-05,
+      "loss": 0.8712,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803089077196794e-05,
+      "loss": 0.9,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980255248042731e-05,
+      "loss": 1.0305,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980201516080934e-05,
+      "loss": 1.0148,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98014771183825e-05,
+      "loss": 0.9519,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980093835318647e-05,
+      "loss": 0.9885,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800398865260987e-05,
+      "loss": 0.9653,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799858654645817e-05,
+      "loss": 0.961,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979931772138081e-05,
+      "loss": 0.9984,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798776065505847e-05,
+      "loss": 1.0154,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979823368706087e-05,
+      "loss": 1.0367,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797690586085882e-05,
+      "loss": 0.9426,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797146762620927e-05,
+      "loss": 1.0023,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796602216706104e-05,
+      "loss": 0.878,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979605694838157e-05,
+      "loss": 1.0161,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795510957687537e-05,
+      "loss": 0.9693,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979496424466426e-05,
+      "loss": 0.9579,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794416809352065e-05,
+      "loss": 0.9589,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793868651791307e-05,
+      "loss": 0.8786,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793319772022418e-05,
+      "loss": 0.9778,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979277017008587e-05,
+      "loss": 1.0238,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792219846022188e-05,
+      "loss": 0.9962,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791668799871957e-05,
+      "loss": 0.9048,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791117031675808e-05,
+      "loss": 0.9885,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790564541474432e-05,
+      "loss": 0.9163,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790011329308573e-05,
+      "loss": 0.9502,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789457395219017e-05,
+      "loss": 0.9761,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788902739246617e-05,
+      "loss": 1.0115,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788347361432276e-05,
+      "loss": 0.8657,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978779126181694e-05,
+      "loss": 0.9979,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787234440441624e-05,
+      "loss": 0.799,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786676897347384e-05,
+      "loss": 0.9046,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978611863257534e-05,
+      "loss": 0.9112,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978555964616665e-05,
+      "loss": 1.015,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978499993816254e-05,
+      "loss": 0.9817,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978443950860428e-05,
+      "loss": 0.9808,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97838783575332e-05,
+      "loss": 0.9003,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783316484990677e-05,
+      "loss": 1.0402,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782753891018146e-05,
+      "loss": 0.9341,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978219057565709e-05,
+      "loss": 0.9401,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978162653894905e-05,
+      "loss": 0.9742,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978106178093562e-05,
+      "loss": 0.9033,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978049630165845e-05,
+      "loss": 0.9911,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977993010115923e-05,
+      "loss": 1.0187,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779363179479713e-05,
+      "loss": 0.9873,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977879553666171e-05,
+      "loss": 0.8613,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778227172747074e-05,
+      "loss": 0.9742,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777658087777723e-05,
+      "loss": 0.2563,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777088281795615e-05,
+      "loss": 1.119,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977651775484277e-05,
+      "loss": 0.8781,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775946506961264e-05,
+      "loss": 1.003,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775374538193218e-05,
+      "loss": 0.9953,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774801848580804e-05,
+      "loss": 0.8445,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774228438166262e-05,
+      "loss": 0.2445,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773654306991866e-05,
+      "loss": 0.9858,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773079455099965e-05,
+      "loss": 0.9957,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772503882532934e-05,
+      "loss": 0.9307,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977192758933323e-05,
+      "loss": 0.9757,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771350575543345e-05,
+      "loss": 1.054,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770772841205822e-05,
+      "loss": 1.0084,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770194386363274e-05,
+      "loss": 0.9993,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976961521105835e-05,
+      "loss": 0.9847,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769035315333757e-05,
+      "loss": 0.8961,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768454699232264e-05,
+      "loss": 0.9798,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976787336279668e-05,
+      "loss": 0.9897,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976729130606988e-05,
+      "loss": 0.9878,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766708529094777e-05,
+      "loss": 0.9486,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976612503191435e-05,
+      "loss": 0.8869,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765540814571628e-05,
+      "loss": 0.9507,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976495587710969e-05,
+      "loss": 1.0002,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976437021957167e-05,
+      "loss": 0.922,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763783842000754e-05,
+      "loss": 0.9226,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763196744440184e-05,
+      "loss": 1.0266,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762608926933248e-05,
+      "loss": 0.2577,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762020389523296e-05,
+      "loss": 0.9543,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761431132253734e-05,
+      "loss": 0.9835,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760841155168e-05,
+      "loss": 0.95,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976025045830961e-05,
+      "loss": 0.8831,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759659041722117e-05,
+      "loss": 0.9564,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759066905449136e-05,
+      "loss": 0.9728,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975847404953433e-05,
+      "loss": 0.9368,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757880474021418e-05,
+      "loss": 0.9381,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975728617895417e-05,
+      "loss": 0.9557,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975669116437641e-05,
+      "loss": 0.9872,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975609543033201e-05,
+      "loss": 0.8902,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975549897686491e-05,
+      "loss": 0.2355,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975490180401909e-05,
+      "loss": 0.9054,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754303911838577e-05,
+      "loss": 0.9934,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975370530036747e-05,
+      "loss": 0.9631,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753105969649904e-05,
+      "loss": 0.9026,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975250591973008e-05,
+      "loss": 0.9022,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751905150652247e-05,
+      "loss": 0.8281,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97513036624607e-05,
+      "loss": 1.1003,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97507014551998e-05,
+      "loss": 0.938,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750098528913946e-05,
+      "loss": 0.8974,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749494883647605e-05,
+      "loss": 1.0076,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974889051944529e-05,
+      "loss": 0.9905,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748285436351565e-05,
+      "loss": 0.9356,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974767963441105e-05,
+      "loss": 1.0058,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747073113668412e-05,
+      "loss": 0.8991,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974646587416839e-05,
+      "loss": 0.967,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974585791595575e-05,
+      "loss": 0.9701,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974524923907533e-05,
+      "loss": 0.9266,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974463984357201e-05,
+      "loss": 0.9204,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974402972949073e-05,
+      "loss": 0.2568,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743418896876476e-05,
+      "loss": 0.895,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742807345774298e-05,
+      "loss": 1.0085,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742195076229288e-05,
+      "loss": 0.9028,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741582088286598e-05,
+      "loss": 0.976,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740968381991427e-05,
+      "loss": 0.8475,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740353957389034e-05,
+      "loss": 0.9207,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973973881452472e-05,
+      "loss": 1.088,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739122953443856e-05,
+      "loss": 0.2869,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738506374191852e-05,
+      "loss": 1.0927,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973788907681417e-05,
+      "loss": 0.9495,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737271061356335e-05,
+      "loss": 0.9504,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736652327863922e-05,
+      "loss": 0.9343,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736032876382554e-05,
+      "loss": 0.9935,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973541270695791e-05,
+      "loss": 0.9662,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734791819635722e-05,
+      "loss": 1.0482,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734170214461776e-05,
+      "loss": 1.0827,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733547891481904e-05,
+      "loss": 0.8979,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732924850742004e-05,
+      "loss": 1.0516,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732301092288014e-05,
+      "loss": 0.9569,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973167661616594e-05,
+      "loss": 0.9766,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731051422421817e-05,
+      "loss": 0.9467,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730425511101756e-05,
+      "loss": 1.0798,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729798882251913e-05,
+      "loss": 0.2519,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729171535918496e-05,
+      "loss": 0.9081,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972854347214776e-05,
+      "loss": 0.9114,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727914690986022e-05,
+      "loss": 1.0305,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727285192479657e-05,
+      "loss": 1.0391,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726654976675073e-05,
+      "loss": 1.0255,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726024043618747e-05,
+      "loss": 1.0296,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725392393357206e-05,
+      "loss": 0.903,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724760025937028e-05,
+      "loss": 0.95,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724126941404843e-05,
+      "loss": 0.9705,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723493139807338e-05,
+      "loss": 0.8721,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722858621191245e-05,
+      "loss": 0.9693,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972222338560336e-05,
+      "loss": 1.0192,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721587433090516e-05,
+      "loss": 1.0172,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972095076369962e-05,
+      "loss": 1.0334,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720313377477618e-05,
+      "loss": 1.0853,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719675274471504e-05,
+      "loss": 1.0117,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971903645472834e-05,
+      "loss": 0.9081,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971839691829523e-05,
+      "loss": 0.969,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971775666521934e-05,
+      "loss": 0.2582,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717115695547873e-05,
+      "loss": 0.9425,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716474009328098e-05,
+      "loss": 0.9376,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715831606607338e-05,
+      "loss": 0.9312,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715188487432957e-05,
+      "loss": 0.8838,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714544651852385e-05,
+      "loss": 0.9511,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713900099913097e-05,
+      "loss": 0.8598,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971325483166262e-05,
+      "loss": 0.9738,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712608847148544e-05,
+      "loss": 1.0319,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711962146418498e-05,
+      "loss": 0.8879,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971131472952017e-05,
+      "loss": 0.9967,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710666596501304e-05,
+      "loss": 0.8766,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710017747409695e-05,
+      "loss": 0.9014,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709368182293187e-05,
+      "loss": 1.0127,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708717901199677e-05,
+      "loss": 1.0006,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708066904177126e-05,
+      "loss": 0.9345,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970741519127353e-05,
+      "loss": 1.0233,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970676276253695e-05,
+      "loss": 1.0323,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706109618015498e-05,
+      "loss": 0.9173,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705455757757338e-05,
+      "loss": 0.9284,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704801181810683e-05,
+      "loss": 0.9536,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704145890223804e-05,
+      "loss": 0.2366,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703489883045022e-05,
+      "loss": 0.9241,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970283316032271e-05,
+      "loss": 1.0591,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702175722105302e-05,
+      "loss": 0.8073,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970151756844127e-05,
+      "loss": 0.8879,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970085869937915e-05,
+      "loss": 0.9883,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970019911496753e-05,
+      "loss": 1.0753,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699538815255047e-05,
+      "loss": 0.9532,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969887780029039e-05,
+      "loss": 0.9774,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96982160701223e-05,
+      "loss": 1.0459,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969755362479958e-05,
+      "loss": 0.8973,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696890464371075e-05,
+      "loss": 0.9409,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696226588885688e-05,
+      "loss": 0.8845,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969556199839238e-05,
+      "loss": 0.996,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969489669294015e-05,
+      "loss": 1.0884,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969423067257806e-05,
+      "loss": 0.9667,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969356393735522e-05,
+      "loss": 1.0033,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692896487320804e-05,
+      "loss": 1.0213,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692228322524028e-05,
+      "loss": 1.0389,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969155944301416e-05,
+      "loss": 0.9624,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969088984884052e-05,
+      "loss": 0.9385,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969021954005249e-05,
+      "loss": 0.9149,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689548516699503e-05,
+      "loss": 1.0108,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688876778831035e-05,
+      "loss": 1.0125,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688204326496622e-05,
+      "loss": 0.8266,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968753115974585e-05,
+      "loss": 0.9399,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686857278628356e-05,
+      "loss": 0.8945,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968618268319384e-05,
+      "loss": 0.7922,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968550737349204e-05,
+      "loss": 1.042,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968483134957276e-05,
+      "loss": 1.0045,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968415461148585e-05,
+      "loss": 1.0072,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683477159281204e-05,
+      "loss": 0.9608,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968279899300879e-05,
+      "loss": 0.9972,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682120112718615e-05,
+      "loss": 0.9139,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681440518460733e-05,
+      "loss": 0.9266,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968076021028526e-05,
+      "loss": 0.9955,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968007918824237e-05,
+      "loss": 1.0215,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967939745238227e-05,
+      "loss": 0.8808,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967871500275524e-05,
+      "loss": 0.2231,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678031839411604e-05,
+      "loss": 0.9292,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677347962401736e-05,
+      "loss": 0.881,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967666337177607e-05,
+      "loss": 0.9974,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675978067585083e-05,
+      "loss": 1.0847,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967529204987931e-05,
+      "loss": 0.9972,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674605318709347e-05,
+      "loss": 1.0139,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673917874125823e-05,
+      "loss": 1.0945,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673229716179438e-05,
+      "loss": 1.0054,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672540844920936e-05,
+      "loss": 0.9876,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671851260401113e-05,
+      "loss": 0.9839,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967116096267082e-05,
+      "loss": 1.0211,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967046995178096e-05,
+      "loss": 1.0395,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669778227782493e-05,
+      "loss": 0.8938,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669085790726422e-05,
+      "loss": 1.0137,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668392640663806e-05,
+      "loss": 0.8813,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966769877764577e-05,
+      "loss": 0.9787,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966700420172347e-05,
+      "loss": 0.9618,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666308912948123e-05,
+      "loss": 0.8941,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665612911371005e-05,
+      "loss": 0.9157,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966491619704344e-05,
+      "loss": 0.9003,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96642187700168e-05,
+      "loss": 1.0547,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966352063034252e-05,
+      "loss": 0.9162,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966282177807208e-05,
+      "loss": 1.078,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966212221325701e-05,
+      "loss": 1.0178,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661421935948898e-05,
+      "loss": 0.9802,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660720946199385e-05,
+      "loss": 0.9481,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660019244060164e-05,
+      "loss": 0.9213,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965931682958297e-05,
+      "loss": 0.9299,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658613702819612e-05,
+      "loss": 1.0354,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965790986382193e-05,
+      "loss": 0.9466,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657205312641832e-05,
+      "loss": 1.0263,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656500049331267e-05,
+      "loss": 0.9275,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655794073942245e-05,
+      "loss": 0.98,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655087386526825e-05,
+      "loss": 0.8853,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654379987137115e-05,
+      "loss": 0.941,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653671875825283e-05,
+      "loss": 0.9106,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652963052643547e-05,
+      "loss": 1.0864,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965225351764417e-05,
+      "loss": 0.9005,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965154327087948e-05,
+      "loss": 1.0717,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965083231240185e-05,
+      "loss": 0.9758,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650120642263703e-05,
+      "loss": 0.9167,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649408260517523e-05,
+      "loss": 0.9111,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964869516721584e-05,
+      "loss": 0.9625,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647981362411236e-05,
+      "loss": 1.0873,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964726684615635e-05,
+      "loss": 1.0517,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964655161850387e-05,
+      "loss": 0.9914,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645835679506537e-05,
+      "loss": 0.8807,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645119029217146e-05,
+      "loss": 0.9687,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964440166768854e-05,
+      "loss": 1.0165,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643683594973625e-05,
+      "loss": 0.9722,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964296481112535e-05,
+      "loss": 0.9838,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964224531619671e-05,
+      "loss": 1.0955,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641525110240772e-05,
+      "loss": 0.9016,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964080419331064e-05,
+      "loss": 0.2352,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640082565459476e-05,
+      "loss": 1.0537,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639360226740492e-05,
+      "loss": 1.0258,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638637177206953e-05,
+      "loss": 0.9846,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637913416912184e-05,
+      "loss": 0.9146,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963718894590955e-05,
+      "loss": 0.9825,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963646376425247e-05,
+      "loss": 0.9684,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963573787199443e-05,
+      "loss": 1.0542,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635011269188952e-05,
+      "loss": 0.943,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634283955889613e-05,
+      "loss": 0.8407,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963355593215005e-05,
+      "loss": 1.1188,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963282719802395e-05,
+      "loss": 0.9661,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632097753565043e-05,
+      "loss": 0.9847,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963136759882713e-05,
+      "loss": 0.8715,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630636733864046e-05,
+      "loss": 0.8958,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629905158729685e-05,
+      "loss": 0.8818,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.9993,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628439878162977e-05,
+      "loss": 1.0525,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962770617283868e-05,
+      "loss": 0.9061,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962697175755921e-05,
+      "loss": 0.8785,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626236632378726e-05,
+      "loss": 1.0487,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962550079735143e-05,
+      "loss": 0.2463,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962476425253159e-05,
+      "loss": 1.0454,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624026997973512e-05,
+      "loss": 0.9376,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962328903373157e-05,
+      "loss": 1.0448,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622550359860176e-05,
+      "loss": 1.0076,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621810976413808e-05,
+      "loss": 0.9964,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621070883446977e-05,
+      "loss": 1.1165,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620330081014268e-05,
+      "loss": 0.8602,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619588569170302e-05,
+      "loss": 0.9215,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618846347969763e-05,
+      "loss": 0.891,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618103417467384e-05,
+      "loss": 0.8505,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617359777717945e-05,
+      "loss": 0.9714,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616615428776283e-05,
+      "loss": 0.9197,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961587037069729e-05,
+      "loss": 1.0373,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615124603535905e-05,
+      "loss": 1.0246,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614378127347124e-05,
+      "loss": 0.9911,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961363094218599e-05,
+      "loss": 0.2531,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96128830481076e-05,
+      "loss": 1.0154,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961213444516711e-05,
+      "loss": 0.8707,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611385133419716e-05,
+      "loss": 0.9765,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610635112920677e-05,
+      "loss": 1.0002,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609884383725298e-05,
+      "loss": 0.8613,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960913294588894e-05,
+      "loss": 0.9391,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608380799467015e-05,
+      "loss": 0.9543,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607627944514983e-05,
+      "loss": 0.9685,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960687438108837e-05,
+      "loss": 0.9484,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960612010924273e-05,
+      "loss": 0.9477,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605365129033697e-05,
+      "loss": 0.863,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604609440516933e-05,
+      "loss": 1.0238,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960385304374817e-05,
+      "loss": 0.9662,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603095938783185e-05,
+      "loss": 1.0652,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602338125677803e-05,
+      "loss": 0.8678,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601579604487914e-05,
+      "loss": 0.8417,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960082037526944e-05,
+      "loss": 0.9405,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600060438078382e-05,
+      "loss": 1.0182,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599299792970766e-05,
+      "loss": 0.967,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959853844000269e-05,
+      "loss": 0.886,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597776379230286e-05,
+      "loss": 1.0706,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597013610709763e-05,
+      "loss": 0.9479,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959625013449736e-05,
+      "loss": 0.8517,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959548595064938e-05,
+      "loss": 1.0237,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594721059222173e-05,
+      "loss": 0.9619,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593955460272146e-05,
+      "loss": 0.8782,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593189153855746e-05,
+      "loss": 1.1028,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592422140029487e-05,
+      "loss": 0.8828,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591654418849934e-05,
+      "loss": 0.9521,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590885990373688e-05,
+      "loss": 0.9681,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590116854657424e-05,
+      "loss": 0.8896,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958934701175785e-05,
+      "loss": 0.9608,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958857646173175e-05,
+      "loss": 0.8834,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587805204635922e-05,
+      "loss": 0.8807,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958703324052726e-05,
+      "loss": 0.9663,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958626056946268e-05,
+      "loss": 0.9986,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958548719149916e-05,
+      "loss": 1.0475,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584713106693728e-05,
+      "loss": 0.9661,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583938315103468e-05,
+      "loss": 0.9255,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958316281678551e-05,
+      "loss": 1.0026,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958238661179705e-05,
+      "loss": 0.9733,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581609700195317e-05,
+      "loss": 0.8374,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580832082037606e-05,
+      "loss": 0.9401,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580053757381253e-05,
+      "loss": 0.9588,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579274726283653e-05,
+      "loss": 0.9774,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957849498880226e-05,
+      "loss": 1.009,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577714544994567e-05,
+      "loss": 1.0133,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576933394918126e-05,
+      "loss": 0.9707,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576151538630538e-05,
+      "loss": 0.8409,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957536897618946e-05,
+      "loss": 0.9341,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574585707652595e-05,
+      "loss": 0.2433,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573801733077707e-05,
+      "loss": 0.9882,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573017052522606e-05,
+      "loss": 0.9607,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957223166604515e-05,
+      "loss": 0.941,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957144557370326e-05,
+      "loss": 0.9635,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570658775554897e-05,
+      "loss": 1.1287,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569871271658087e-05,
+      "loss": 0.9567,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95690830620709e-05,
+      "loss": 0.8949,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568294146851455e-05,
+      "loss": 0.9686,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956750452605793e-05,
+      "loss": 0.9307,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566714199748548e-05,
+      "loss": 0.8825,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565923167981597e-05,
+      "loss": 0.2247,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565131430815402e-05,
+      "loss": 1.0679,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956433898830835e-05,
+      "loss": 0.8893,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956354584051887e-05,
+      "loss": 0.9653,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562751987505458e-05,
+      "loss": 0.9191,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561957429326644e-05,
+      "loss": 0.9964,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561162166041027e-05,
+      "loss": 0.8937,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560366197707248e-05,
+      "loss": 0.9931,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559569524384003e-05,
+      "loss": 0.9211,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955877214613004e-05,
+      "loss": 0.845,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557974063004154e-05,
+      "loss": 0.9479,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95571752750652e-05,
+      "loss": 1.018,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955637578237208e-05,
+      "loss": 0.9287,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955557558498375e-05,
+      "loss": 0.9794,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955477468295922e-05,
+      "loss": 0.8965,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553973076357546e-05,
+      "loss": 0.9814,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553170765237834e-05,
+      "loss": 0.9548,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552367749659255e-05,
+      "loss": 0.9881,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551564029681023e-05,
+      "loss": 0.8565,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550759605362403e-05,
+      "loss": 0.9868,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549954476762715e-05,
+      "loss": 0.9156,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549148643941328e-05,
+      "loss": 0.9641,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548342106957665e-05,
+      "loss": 0.9394,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547534865871202e-05,
+      "loss": 1.0345,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954672692074147e-05,
+      "loss": 0.9588,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545918271628038e-05,
+      "loss": 0.9686,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545108918590544e-05,
+      "loss": 0.921,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544298861688667e-05,
+      "loss": 0.8905,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954348810098214e-05,
+      "loss": 1.0029,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542676636530755e-05,
+      "loss": 1.0148,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541864468394344e-05,
+      "loss": 0.9529,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541051596632804e-05,
+      "loss": 0.9951,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540238021306067e-05,
+      "loss": 0.9042,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539423742474134e-05,
+      "loss": 0.955,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538608760197045e-05,
+      "loss": 0.8979,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537793074534907e-05,
+      "loss": 0.9423,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536976685547858e-05,
+      "loss": 0.9004,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536159593296105e-05,
+      "loss": 0.9271,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535341797839904e-05,
+      "loss": 0.9156,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534523299239553e-05,
+      "loss": 0.8974,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533704097555417e-05,
+      "loss": 0.9687,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532884192847898e-05,
+      "loss": 0.9096,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532063585177456e-05,
+      "loss": 0.2315,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531242274604607e-05,
+      "loss": 0.9878,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530420261189914e-05,
+      "loss": 1.0508,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529597544993994e-05,
+      "loss": 1.024,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952877412607751e-05,
+      "loss": 0.9284,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527950004501194e-05,
+      "loss": 0.8646,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95271251803258e-05,
+      "loss": 0.9674,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526299653612164e-05,
+      "loss": 0.9445,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525473424421153e-05,
+      "loss": 0.9917,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95246464928137e-05,
+      "loss": 0.8959,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952381885885078e-05,
+      "loss": 0.9327,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522990522593428e-05,
+      "loss": 0.9114,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522161484102722e-05,
+      "loss": 0.9313,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95213317434398e-05,
+      "loss": 1.1042,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520501300665837e-05,
+      "loss": 0.9428,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519670155842082e-05,
+      "loss": 0.9332,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951883830902982e-05,
+      "loss": 0.8973,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518005760290396e-05,
+      "loss": 0.957,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95171725096852e-05,
+      "loss": 0.9773,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951633855727567e-05,
+      "loss": 0.2348,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515503903123314e-05,
+      "loss": 1.0417,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514668547289672e-05,
+      "loss": 0.9398,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951383248983635e-05,
+      "loss": 0.9479,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951299573082499e-05,
+      "loss": 0.9646,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951215827031731e-05,
+      "loss": 0.9855,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511320108375045e-05,
+      "loss": 1.022,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510481245060022e-05,
+      "loss": 1.0459,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950964168043409e-05,
+      "loss": 0.9869,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950880141455916e-05,
+      "loss": 0.8855,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950796044749719e-05,
+      "loss": 0.2459,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507118779310197e-05,
+      "loss": 0.9876,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950627641006025e-05,
+      "loss": 1.0116,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505433339809464e-05,
+      "loss": 0.9492,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504589568620003e-05,
+      "loss": 0.9459,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503745096554093e-05,
+      "loss": 0.9509,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502899923674003e-05,
+      "loss": 0.9094,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502054050042058e-05,
+      "loss": 0.9421,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950120747572063e-05,
+      "loss": 0.9839,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500360200772156e-05,
+      "loss": 1.0054,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499512225259103e-05,
+      "loss": 0.9735,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949866354924401e-05,
+      "loss": 0.9699,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497814172789453e-05,
+      "loss": 0.8229,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496964095958067e-05,
+      "loss": 1.0413,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496113318812542e-05,
+      "loss": 0.9608,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949526184141561e-05,
+      "loss": 0.9692,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494409663830064e-05,
+      "loss": 1.0566,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493556786118738e-05,
+      "loss": 0.9776,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492703208344527e-05,
+      "loss": 0.9841,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491848930570376e-05,
+      "loss": 0.9457,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949099395285928e-05,
+      "loss": 0.9972,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490138275274285e-05,
+      "loss": 0.9465,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948928189787849e-05,
+      "loss": 0.927,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488424820735044e-05,
+      "loss": 0.9754,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487567043907148e-05,
+      "loss": 1.0788,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948670856745806e-05,
+      "loss": 0.9636,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485849391451076e-05,
+      "loss": 0.2278,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948498951594956e-05,
+      "loss": 0.8549,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484128941016916e-05,
+      "loss": 0.9222,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483267666716607e-05,
+      "loss": 0.9677,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948240569311214e-05,
+      "loss": 1.0364,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481543020267083e-05,
+      "loss": 0.9318,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480679648245045e-05,
+      "loss": 0.2818,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947981557710969e-05,
+      "loss": 0.8377,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478950806924742e-05,
+      "loss": 0.9887,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478085337753966e-05,
+      "loss": 1.0074,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477219169661183e-05,
+      "loss": 0.9764,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476352302710267e-05,
+      "loss": 0.8618,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947548473696514e-05,
+      "loss": 1.0042,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474616472489773e-05,
+      "loss": 0.2735,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94737475093482e-05,
+      "loss": 0.2315,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472877847604493e-05,
+      "loss": 1.0628,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472007487322786e-05,
+      "loss": 0.8969,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471136428567256e-05,
+      "loss": 0.9404,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947026467140214e-05,
+      "loss": 0.9696,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469392215891715e-05,
+      "loss": 1.1042,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468519062100326e-05,
+      "loss": 1.0154,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467645210092353e-05,
+      "loss": 0.2797,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946677065993224e-05,
+      "loss": 1.0114,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946589541168447e-05,
+      "loss": 0.9601,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465019465413592e-05,
+      "loss": 0.2336,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464142821184192e-05,
+      "loss": 0.9778,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946326547906092e-05,
+      "loss": 0.8842,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462387439108473e-05,
+      "loss": 1.037,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461508701391594e-05,
+      "loss": 0.9431,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460629265975082e-05,
+      "loss": 0.9669,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945974913292379e-05,
+      "loss": 1.121,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458868302302617e-05,
+      "loss": 0.9882,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457986774176517e-05,
+      "loss": 1.0446,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94571045486105e-05,
+      "loss": 0.9926,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456221625669614e-05,
+      "loss": 0.9293,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945533800541897e-05,
+      "loss": 1.0398,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454453687923723e-05,
+      "loss": 0.8666,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453568673249093e-05,
+      "loss": 1.0004,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945268296146033e-05,
+      "loss": 1.0529,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451796552622758e-05,
+      "loss": 0.9642,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450909446801736e-05,
+      "loss": 1.0118,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450021644062676e-05,
+      "loss": 0.9198,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449133144471054e-05,
+      "loss": 0.8658,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448243948092386e-05,
+      "loss": 0.9421,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447354054992232e-05,
+      "loss": 0.9401,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944646346523623e-05,
+      "loss": 0.8087,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944557217889004e-05,
+      "loss": 0.925,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444680196019395e-05,
+      "loss": 0.9938,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443787516690063e-05,
+      "loss": 1.0342,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442894140967875e-05,
+      "loss": 0.9769,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944200006891871e-05,
+      "loss": 0.9988,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94411053006085e-05,
+      "loss": 0.9253,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944020983610322e-05,
+      "loss": 1.0051,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94393136754689e-05,
+      "loss": 0.8805,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438416818771633e-05,
+      "loss": 1.1036,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943751926607755e-05,
+      "loss": 0.8315,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436621017452833e-05,
+      "loss": 0.9851,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435722072963724e-05,
+      "loss": 1.0272,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943482243267651e-05,
+      "loss": 0.8354,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433922096657532e-05,
+      "loss": 0.9427,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433021064973186e-05,
+      "loss": 0.966,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943211933768991e-05,
+      "loss": 0.891,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431216914874196e-05,
+      "loss": 0.9338,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430313796592594e-05,
+      "loss": 0.9918,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429409982911698e-05,
+      "loss": 0.2273,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942850547389816e-05,
+      "loss": 1.0183,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942760026961867e-05,
+      "loss": 1.0046,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942669437013999e-05,
+      "loss": 0.9942,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425787775528913e-05,
+      "loss": 0.9506,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424880485852298e-05,
+      "loss": 0.8992,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423972501177048e-05,
+      "loss": 0.2636,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423063821570113e-05,
+      "loss": 0.9805,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942215444709851e-05,
+      "loss": 1.0306,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942124437782929e-05,
+      "loss": 0.9697,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420333613829562e-05,
+      "loss": 0.9967,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419422155166484e-05,
+      "loss": 1.0094,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941851000190728e-05,
+      "loss": 1.0153,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94175971541192e-05,
+      "loss": 0.8793,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941668361186956e-05,
+      "loss": 0.9052,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415769375225733e-05,
+      "loss": 1.0168,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941485444425513e-05,
+      "loss": 0.9097,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941393881902522e-05,
+      "loss": 0.873,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941302249960352e-05,
+      "loss": 0.8452,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412105486057602e-05,
+      "loss": 0.9122,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941118777845509e-05,
+      "loss": 0.8617,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941026937686365e-05,
+      "loss": 1.051,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940935028135101e-05,
+      "loss": 0.9299,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408430491984943e-05,
+      "loss": 1.0042,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940751000883328e-05,
+      "loss": 0.948,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406588831963894e-05,
+      "loss": 0.8988,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405666961444713e-05,
+      "loss": 0.9041,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404744397343716e-05,
+      "loss": 0.9278,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403821139728936e-05,
+      "loss": 0.9142,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402897188668453e-05,
+      "loss": 0.9238,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94019725442304e-05,
+      "loss": 0.8439,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401047206482964e-05,
+      "loss": 0.9817,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400121175494376e-05,
+      "loss": 0.9741,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399194451332926e-05,
+      "loss": 0.9179,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398267034066954e-05,
+      "loss": 0.9451,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939733892376484e-05,
+      "loss": 0.9474,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396410120495024e-05,
+      "loss": 1.0194,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395480624326006e-05,
+      "loss": 0.9407,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939455043532632e-05,
+      "loss": 1.0225,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939361955356456e-05,
+      "loss": 0.8762,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392687979109378e-05,
+      "loss": 0.9799,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391755712029458e-05,
+      "loss": 0.8889,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939082275239355e-05,
+      "loss": 0.9381,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938988910027045e-05,
+      "loss": 0.9604,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388954755729014e-05,
+      "loss": 0.9116,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938801971883813e-05,
+      "loss": 0.9505,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387083989666756e-05,
+      "loss": 0.979,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386147568283894e-05,
+      "loss": 0.8826,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938521045475859e-05,
+      "loss": 1.0116,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384272649159952e-05,
+      "loss": 1.0222,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383334151557136e-05,
+      "loss": 1.0071,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382394962019343e-05,
+      "loss": 0.9194,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381455080615833e-05,
+      "loss": 0.9645,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380514507415914e-05,
+      "loss": 0.9886,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379573242488944e-05,
+      "loss": 0.9324,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937863128590433e-05,
+      "loss": 0.9676,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937768863773154e-05,
+      "loss": 0.8282,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376745298040072e-05,
+      "loss": 0.9774,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375801266899503e-05,
+      "loss": 0.9316,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937485654437944e-05,
+      "loss": 0.9005,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373911130549547e-05,
+      "loss": 0.9155,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372965025479544e-05,
+      "loss": 0.9157,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372018229239194e-05,
+      "loss": 0.9962,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371070741898316e-05,
+      "loss": 0.953,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937012256352677e-05,
+      "loss": 0.2494,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369173694194492e-05,
+      "loss": 0.91,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936822413397144e-05,
+      "loss": 0.8996,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936727388292764e-05,
+      "loss": 0.8824,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366322941133166e-05,
+      "loss": 0.9527,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365371308658135e-05,
+      "loss": 1.0208,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364418985572727e-05,
+      "loss": 0.9508,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363465971947165e-05,
+      "loss": 0.8609,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362512267851722e-05,
+      "loss": 1.0005,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361557873356732e-05,
+      "loss": 0.9961,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360602788532567e-05,
+      "loss": 0.9272,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359647013449655e-05,
+      "loss": 0.9798,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935869054817848e-05,
+      "loss": 0.9394,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935773339278957e-05,
+      "loss": 1.0581,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935677554735351e-05,
+      "loss": 0.924,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355817011940927e-05,
+      "loss": 0.9434,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935485778662251e-05,
+      "loss": 0.9827,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353897871468983e-05,
+      "loss": 1.0402,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935293726655114e-05,
+      "loss": 0.9741,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351975971939816e-05,
+      "loss": 0.9367,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93510139877059e-05,
+      "loss": 0.9923,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935005131392032e-05,
+      "loss": 0.8856,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934908795065407e-05,
+      "loss": 0.9482,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348123897978196e-05,
+      "loss": 1.0181,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347159155963775e-05,
+      "loss": 0.9491,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346193724681957e-05,
+      "loss": 0.9009,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934522760420393e-05,
+      "loss": 1.0118,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344260794600938e-05,
+      "loss": 0.984,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343293295944275e-05,
+      "loss": 0.9755,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934232510830528e-05,
+      "loss": 0.9982,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341356231755357e-05,
+      "loss": 1.0341,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340386666365943e-05,
+      "loss": 0.9753,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933941641220854e-05,
+      "loss": 0.9201,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338445469354696e-05,
+      "loss": 0.9869,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337473837876005e-05,
+      "loss": 0.9199,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336501517844116e-05,
+      "loss": 0.95,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933552850933073e-05,
+      "loss": 0.946,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334554812407603e-05,
+      "loss": 0.9008,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333580427146524e-05,
+      "loss": 0.9522,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933260535361936e-05,
+      "loss": 0.9657,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331629591898e-05,
+      "loss": 0.8597,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330653142054407e-05,
+      "loss": 1.0854,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329676004160582e-05,
+      "loss": 0.9495,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328698178288578e-05,
+      "loss": 0.9486,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327719664510503e-05,
+      "loss": 0.8791,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326740462898513e-05,
+      "loss": 1.0067,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325760573524814e-05,
+      "loss": 0.9872,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324779996461666e-05,
+      "loss": 1.0245,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323798731781373e-05,
+      "loss": 0.9188,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93228167795563e-05,
+      "loss": 0.88,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321834139858856e-05,
+      "loss": 1.0572,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93208508127615e-05,
+      "loss": 1.0242,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931986679833674e-05,
+      "loss": 0.9818,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318882096657145e-05,
+      "loss": 0.9691,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317896707795326e-05,
+      "loss": 0.9001,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931691063182394e-05,
+      "loss": 1.0606,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315923868815713e-05,
+      "loss": 0.8762,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93149364188434e-05,
+      "loss": 0.999,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313948281979816e-05,
+      "loss": 0.9158,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931295945829783e-05,
+      "loss": 0.8755,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311969947870362e-05,
+      "loss": 0.9379,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310979750770375e-05,
+      "loss": 0.9691,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309988867070887e-05,
+      "loss": 0.9929,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308997296844973e-05,
+      "loss": 1.0706,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930800504016574e-05,
+      "loss": 0.9875,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307012097106367e-05,
+      "loss": 0.9878,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306018467740074e-05,
+      "loss": 0.9736,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930502415214013e-05,
+      "loss": 1.0584,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304029150379854e-05,
+      "loss": 0.9409,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303033462532625e-05,
+      "loss": 1.076,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930203708867186e-05,
+      "loss": 1.0022,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301040028871036e-05,
+      "loss": 0.9648,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300042283203673e-05,
+      "loss": 0.7603,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299043851743352e-05,
+      "loss": 0.226,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298044734563693e-05,
+      "loss": 0.9573,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297044931738372e-05,
+      "loss": 0.9363,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296044443341117e-05,
+      "loss": 0.9321,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295043269445705e-05,
+      "loss": 0.9896,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294041410125964e-05,
+      "loss": 0.9914,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293038865455765e-05,
+      "loss": 0.9219,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929203563550905e-05,
+      "loss": 1.0283,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291031720359787e-05,
+      "loss": 0.8709,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290027120082008e-05,
+      "loss": 0.9803,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289021834749792e-05,
+      "loss": 0.9479,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288015864437272e-05,
+      "loss": 0.9683,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928700920921863e-05,
+      "loss": 0.9539,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286001869168097e-05,
+      "loss": 0.8724,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284993844359952e-05,
+      "loss": 0.9381,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283985134868532e-05,
+      "loss": 0.9036,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282975740768215e-05,
+      "loss": 0.9224,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928196566213344e-05,
+      "loss": 0.9417,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280954899038688e-05,
+      "loss": 0.9202,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279943451558495e-05,
+      "loss": 0.9255,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927893131976744e-05,
+      "loss": 0.894,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927791850374017e-05,
+      "loss": 0.9608,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276905003551364e-05,
+      "loss": 0.8495,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275890819275754e-05,
+      "loss": 0.9986,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927487595098813e-05,
+      "loss": 0.954,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273860398763338e-05,
+      "loss": 0.969,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272844162676255e-05,
+      "loss": 0.9149,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271827242801823e-05,
+      "loss": 0.9846,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927080963921503e-05,
+      "loss": 0.8388,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269791351990913e-05,
+      "loss": 0.819,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268772381204567e-05,
+      "loss": 0.2522,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267752726931127e-05,
+      "loss": 0.2456,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266732389245782e-05,
+      "loss": 1.0249,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926571136822378e-05,
+      "loss": 0.9628,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264689663940405e-05,
+      "loss": 1.0413,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263667276470998e-05,
+      "loss": 0.8973,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262644205890953e-05,
+      "loss": 0.8902,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261620452275712e-05,
+      "loss": 1.0545,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260596015700768e-05,
+      "loss": 0.926,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259570896241664e-05,
+      "loss": 0.9437,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258545093973994e-05,
+      "loss": 0.8444,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257518608973395e-05,
+      "loss": 1.0299,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925649144131557e-05,
+      "loss": 0.9656,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255463591076258e-05,
+      "loss": 0.9429,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254435058331255e-05,
+      "loss": 0.9264,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253405843156404e-05,
+      "loss": 1.0183,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252375945627603e-05,
+      "loss": 1.0238,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251345365820794e-05,
+      "loss": 0.9475,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250314103811974e-05,
+      "loss": 0.9436,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249282159677193e-05,
+      "loss": 1.1273,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248249533492545e-05,
+      "loss": 0.9767,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247216225334177e-05,
+      "loss": 0.9278,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246182235278285e-05,
+      "loss": 0.9956,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245147563401114e-05,
+      "loss": 1.0397,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244112209778965e-05,
+      "loss": 1.0292,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243076174488187e-05,
+      "loss": 0.9037,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242039457605175e-05,
+      "loss": 0.941,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241002059206376e-05,
+      "loss": 0.8882,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239963979368294e-05,
+      "loss": 0.9323,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238925218167476e-05,
+      "loss": 0.8678,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237885775680518e-05,
+      "loss": 1.0482,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923684565198407e-05,
+      "loss": 0.9386,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235804847154836e-05,
+      "loss": 0.9188,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923476336126956e-05,
+      "loss": 0.9848,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923372119440505e-05,
+      "loss": 0.9856,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232678346638147e-05,
+      "loss": 0.8634,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231634818045756e-05,
+      "loss": 0.959,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923059060870483e-05,
+      "loss": 0.9581,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229545718692368e-05,
+      "loss": 0.868,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228500148085416e-05,
+      "loss": 1.0312,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227453896961084e-05,
+      "loss": 0.9243,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226406965396517e-05,
+      "loss": 1.0258,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225359353468914e-05,
+      "loss": 1.0138,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224311061255537e-05,
+      "loss": 1.0263,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223262088833684e-05,
+      "loss": 1.0042,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222212436280704e-05,
+      "loss": 0.961,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221162103673998e-05,
+      "loss": 1.0618,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220111091091024e-05,
+      "loss": 1.0538,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921905939860928e-05,
+      "loss": 0.8562,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921800702630632e-05,
+      "loss": 0.8861,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921695397425975e-05,
+      "loss": 0.8026,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215900242547217e-05,
+      "loss": 1.0864,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921484583124643e-05,
+      "loss": 0.9506,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213790740435137e-05,
+      "loss": 1.0691,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212734970191148e-05,
+      "loss": 0.9249,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921167852059231e-05,
+      "loss": 0.8657,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210621391716525e-05,
+      "loss": 1.0655,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209563583641752e-05,
+      "loss": 0.9561,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920850509644599e-05,
+      "loss": 0.8976,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207445930207305e-05,
+      "loss": 0.9791,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206386085003786e-05,
+      "loss": 0.9132,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920532556091359e-05,
+      "loss": 0.9816,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204264358014924e-05,
+      "loss": 1.0115,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203202476386046e-05,
+      "loss": 0.9249,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920213991610525e-05,
+      "loss": 0.9637,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201076677250898e-05,
+      "loss": 0.273,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920001275990139e-05,
+      "loss": 1.0173,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919894816413518e-05,
+      "loss": 1.0741,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197882890030778e-05,
+      "loss": 0.9535,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196816937666732e-05,
+      "loss": 0.9503,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919575030712165e-05,
+      "loss": 0.9449,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194682998474183e-05,
+      "loss": 0.8831,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193615011803035e-05,
+      "loss": 0.8983,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192546347186966e-05,
+      "loss": 0.9954,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919147700470477e-05,
+      "loss": 0.2616,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190406984435313e-05,
+      "loss": 0.9868,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918933628645749e-05,
+      "loss": 0.9243,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918826491085026e-05,
+      "loss": 0.9999,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918719285769263e-05,
+      "loss": 0.9381,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186120127063647e-05,
+      "loss": 0.9149,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185046719042416e-05,
+      "loss": 0.9355,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183972633708094e-05,
+      "loss": 0.9409,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918289787113989e-05,
+      "loss": 0.9811,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181822431417045e-05,
+      "loss": 0.91,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180746314618872e-05,
+      "loss": 1.0059,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917966952082472e-05,
+      "loss": 0.9507,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178592050114e-05,
+      "loss": 0.967,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177513902566155e-05,
+      "loss": 0.2263,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91764350782607e-05,
+      "loss": 0.9836,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917535557727718e-05,
+      "loss": 0.9918,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174275399695202e-05,
+      "loss": 0.9807,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173194545594418e-05,
+      "loss": 1.0164,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917211301505453e-05,
+      "loss": 0.2569,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171030808155296e-05,
+      "loss": 0.9431,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169947924976512e-05,
+      "loss": 0.9819,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168864365598035e-05,
+      "loss": 0.9719,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167780130099766e-05,
+      "loss": 0.9706,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916669521856166e-05,
+      "loss": 0.9459,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165609631063716e-05,
+      "loss": 1.0651,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164523367685986e-05,
+      "loss": 1.0297,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163436428508576e-05,
+      "loss": 0.8925,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162348813611632e-05,
+      "loss": 0.9359,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916126052307536e-05,
+      "loss": 0.9894,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916017155698001e-05,
+      "loss": 1.0383,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159081915405883e-05,
+      "loss": 0.9037,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915799159843333e-05,
+      "loss": 0.9912,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156900606142757e-05,
+      "loss": 0.9088,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915580893861461e-05,
+      "loss": 1.0009,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154716595929385e-05,
+      "loss": 0.9035,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915362357816764e-05,
+      "loss": 0.8546,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152529885409973e-05,
+      "loss": 0.8818,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915143551773703e-05,
+      "loss": 0.9603,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150340475229514e-05,
+      "loss": 0.9788,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149244757968175e-05,
+      "loss": 0.8365,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914814836603381e-05,
+      "loss": 1.0444,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914705129950727e-05,
+      "loss": 0.9904,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914595355846945e-05,
+      "loss": 0.8675,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144855143001303e-05,
+      "loss": 0.9514,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143756053183824e-05,
+      "loss": 0.2179,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914265628909806e-05,
+      "loss": 0.976,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141555850825114e-05,
+      "loss": 0.9907,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140454738446127e-05,
+      "loss": 1.1182,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91393529520423e-05,
+      "loss": 0.9479,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138250491694873e-05,
+      "loss": 0.8104,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913714735748515e-05,
+      "loss": 0.9318,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913604354949448e-05,
+      "loss": 0.9609,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134939067804248e-05,
+      "loss": 1.0519,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133833912495905e-05,
+      "loss": 0.9456,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132728083650945e-05,
+      "loss": 0.9337,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131621581350915e-05,
+      "loss": 1.0017,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913051440567741e-05,
+      "loss": 0.8441,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129406556712066e-05,
+      "loss": 0.9137,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912829803453659e-05,
+      "loss": 0.9725,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912718883923271e-05,
+      "loss": 0.2765,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126078970882232e-05,
+      "loss": 0.9277,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124968429566992e-05,
+      "loss": 1.1144,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123857215368885e-05,
+      "loss": 0.955,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912274532836985e-05,
+      "loss": 0.8987,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121632768651884e-05,
+      "loss": 1.0831,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912051953629702e-05,
+      "loss": 0.9678,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119405631387353e-05,
+      "loss": 0.9152,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118291054005027e-05,
+      "loss": 0.9642,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117175804232226e-05,
+      "loss": 0.9276,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911605988215119e-05,
+      "loss": 0.9104,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911494328784421e-05,
+      "loss": 0.9468,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113826021393627e-05,
+      "loss": 1.0638,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112708082881824e-05,
+      "loss": 0.8589,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111589472391242e-05,
+      "loss": 0.9932,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110470190004368e-05,
+      "loss": 0.928,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109350235803737e-05,
+      "loss": 0.9633,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910822960987194e-05,
+      "loss": 0.8951,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107108312291606e-05,
+      "loss": 0.9893,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105986343145425e-05,
+      "loss": 0.9263,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104863702516132e-05,
+      "loss": 1.0076,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103740390486506e-05,
+      "loss": 0.9747,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102616407139393e-05,
+      "loss": 0.9672,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101491752557662e-05,
+      "loss": 0.8849,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100366426824254e-05,
+      "loss": 0.9908,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099240430022152e-05,
+      "loss": 0.9389,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098113762234387e-05,
+      "loss": 0.881,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096986423544038e-05,
+      "loss": 0.8611,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095858414034238e-05,
+      "loss": 0.8727,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094729733788167e-05,
+      "loss": 0.8832,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093600382889056e-05,
+      "loss": 1.0054,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092470361420182e-05,
+      "loss": 0.8716,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909133966946487e-05,
+      "loss": 0.9955,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909020830710651e-05,
+      "loss": 1.0372,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908907627442852e-05,
+      "loss": 1.0398,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908794357151438e-05,
+      "loss": 1.074,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086810198447618e-05,
+      "loss": 0.904,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085676155311806e-05,
+      "loss": 0.8336,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908454144219057e-05,
+      "loss": 0.9133,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908340605916759e-05,
+      "loss": 0.9175,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908227000632658e-05,
+      "loss": 0.8542,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081133283751324e-05,
+      "loss": 1.0073,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907999589152564e-05,
+      "loss": 0.9352,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90788578297334e-05,
+      "loss": 0.8441,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077719098458528e-05,
+      "loss": 0.2344,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076579697784995e-05,
+      "loss": 0.968,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075439627796816e-05,
+      "loss": 0.9759,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074298888578063e-05,
+      "loss": 0.2401,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907315748021286e-05,
+      "loss": 0.2487,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907201540278537e-05,
+      "loss": 1.0124,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070872656379814e-05,
+      "loss": 0.8967,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069729241080458e-05,
+      "loss": 1.0718,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906858515697162e-05,
+      "loss": 0.2474,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067440404137657e-05,
+      "loss": 0.8614,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066294982662998e-05,
+      "loss": 0.7773,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065148892632095e-05,
+      "loss": 0.9267,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906400213412947e-05,
+      "loss": 0.9599,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062854707239684e-05,
+      "loss": 0.8951,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061706612047346e-05,
+      "loss": 0.2742,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060557848637123e-05,
+      "loss": 0.9919,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905940841709372e-05,
+      "loss": 1.0063,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90582583175019e-05,
+      "loss": 1.026,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057107549946468e-05,
+      "loss": 0.9204,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905595611451229e-05,
+      "loss": 0.8753,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054804011284268e-05,
+      "loss": 0.9034,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053651240347363e-05,
+      "loss": 0.9455,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905249780178658e-05,
+      "loss": 0.9421,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051343695686973e-05,
+      "loss": 0.8857,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050188922133646e-05,
+      "loss": 0.928,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904903348121176e-05,
+      "loss": 0.969,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904787737300651e-05,
+      "loss": 0.9598,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046720597603154e-05,
+      "loss": 0.8954,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045563155086987e-05,
+      "loss": 0.9297,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044405045543363e-05,
+      "loss": 0.9591,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043246269057688e-05,
+      "loss": 0.8467,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90420868257154e-05,
+      "loss": 0.8815,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904092671560201e-05,
+      "loss": 0.9963,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039765938803053e-05,
+      "loss": 0.9292,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038604495404135e-05,
+      "loss": 1.0338,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037442385490895e-05,
+      "loss": 0.9619,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903627960914903e-05,
+      "loss": 0.9536,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035116166464292e-05,
+      "loss": 0.9441,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903395205752246e-05,
+      "loss": 0.9389,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903278728240939e-05,
+      "loss": 0.981,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031621841210965e-05,
+      "loss": 0.8502,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030455734013128e-05,
+      "loss": 0.8367,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029288960901868e-05,
+      "loss": 0.9941,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028121521963223e-05,
+      "loss": 0.9411,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026953417283284e-05,
+      "loss": 0.9734,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902578464694819e-05,
+      "loss": 0.9467,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902461521104412e-05,
+      "loss": 0.982,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023445109657314e-05,
+      "loss": 0.8695,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022274342874053e-05,
+      "loss": 0.9251,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021102910780676e-05,
+      "loss": 0.9705,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019930813463557e-05,
+      "loss": 0.9589,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018758051009137e-05,
+      "loss": 0.2568,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901758462350389e-05,
+      "loss": 0.9756,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016410531034347e-05,
+      "loss": 0.8684,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015235773687087e-05,
+      "loss": 1.01,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014060351548738e-05,
+      "loss": 0.9166,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012884264705977e-05,
+      "loss": 0.9432,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011707513245525e-05,
+      "loss": 0.9807,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010530097254164e-05,
+      "loss": 0.97,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009352016818713e-05,
+      "loss": 1.019,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008173272026042e-05,
+      "loss": 0.9669,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006993862963083e-05,
+      "loss": 0.2193,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005813789716795e-05,
+      "loss": 0.8541,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004633052374203e-05,
+      "loss": 0.9523,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003451651022376e-05,
+      "loss": 0.8638,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900226958574843e-05,
+      "loss": 0.2525,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001086856639533e-05,
+      "loss": 0.8847,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89999034637829e-05,
+      "loss": 0.9211,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998719407265794e-05,
+      "loss": 0.9103,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997534687175528e-05,
+      "loss": 0.9191,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899634930359947e-05,
+      "loss": 0.8724,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995163256625023e-05,
+      "loss": 0.9295,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899397654633965e-05,
+      "loss": 1.0432,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992789172830865e-05,
+      "loss": 0.9949,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991601136186214e-05,
+      "loss": 1.0618,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899041243649332e-05,
+      "loss": 0.2614,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989223073839827e-05,
+      "loss": 0.9294,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988033048313443e-05,
+      "loss": 0.9901,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898684236000192e-05,
+      "loss": 0.9689,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985651008993063e-05,
+      "loss": 0.9787,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898445899537472e-05,
+      "loss": 0.9781,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983266319234794e-05,
+      "loss": 0.9216,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982072980661237e-05,
+      "loss": 0.9874,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898087897974204e-05,
+      "loss": 0.9708,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897968431656525e-05,
+      "loss": 0.8543,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978488991218967e-05,
+      "loss": 0.9318,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977293003791333e-05,
+      "loss": 0.8776,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897609635437054e-05,
+      "loss": 0.8867,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974899043044833e-05,
+      "loss": 0.92,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973701069902502e-05,
+      "loss": 0.8893,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972502435031884e-05,
+      "loss": 0.2744,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971303138521367e-05,
+      "loss": 0.9029,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970103180459395e-05,
+      "loss": 1.0291,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968902560934443e-05,
+      "loss": 0.8853,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967701280035056e-05,
+      "loss": 0.8958,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896649933784981e-05,
+      "loss": 0.8585,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965296734467343e-05,
+      "loss": 0.807,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964093469976334e-05,
+      "loss": 1.0382,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962889544465512e-05,
+      "loss": 0.8671,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896168495802365e-05,
+      "loss": 0.9536,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960479710739585e-05,
+      "loss": 0.809,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895927380270219e-05,
+      "loss": 1.0552,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958067234000387e-05,
+      "loss": 0.9904,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895686000472315e-05,
+      "loss": 0.9337,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955652114959502e-05,
+      "loss": 0.9496,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954443564798513e-05,
+      "loss": 0.8892,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953234354329308e-05,
+      "loss": 0.9359,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952024483641044e-05,
+      "loss": 0.9946,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950813952822948e-05,
+      "loss": 0.8465,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949602761964278e-05,
+      "loss": 0.9176,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948390911154357e-05,
+      "loss": 0.9265,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894717840048254e-05,
+      "loss": 1.0287,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945965230038243e-05,
+      "loss": 1.109,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944751399910927e-05,
+      "loss": 0.9791,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943536910190093e-05,
+      "loss": 0.9531,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942321760965307e-05,
+      "loss": 0.988,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941105952326174e-05,
+      "loss": 1.0716,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939889484362345e-05,
+      "loss": 0.8819,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938672357163527e-05,
+      "loss": 0.9507,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937454570819468e-05,
+      "loss": 0.8879,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893623612541997e-05,
+      "loss": 0.9076,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893501702105489e-05,
+      "loss": 0.9525,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933797257814112e-05,
+      "loss": 1.0891,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932576835787592e-05,
+      "loss": 0.9666,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893135575506532e-05,
+      "loss": 1.0477,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930134015737344e-05,
+      "loss": 0.9523,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928911617893755e-05,
+      "loss": 0.9501,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892768856162469e-05,
+      "loss": 1.0058,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892646484702034e-05,
+      "loss": 0.8731,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925240474170947e-05,
+      "loss": 0.9833,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892401544316679e-05,
+      "loss": 0.9855,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892278975409821e-05,
+      "loss": 1.0377,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921563407055585e-05,
+      "loss": 0.8513,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920336402129353e-05,
+      "loss": 0.9651,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891910873940999e-05,
+      "loss": 0.922,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917880418988028e-05,
+      "loss": 0.9999,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891665144095404e-05,
+      "loss": 0.992,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915421805398655e-05,
+      "loss": 0.9314,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914191512412546e-05,
+      "loss": 0.8992,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912960562086436e-05,
+      "loss": 0.9878,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89117289545111e-05,
+      "loss": 1.0447,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891049668977735e-05,
+      "loss": 0.9673,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909263767976063e-05,
+      "loss": 0.9055,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890803018919815e-05,
+      "loss": 1.0298,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906795953534577e-05,
+      "loss": 0.905,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890556106107636e-05,
+      "loss": 0.9644,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904325511914556e-05,
+      "loss": 0.9194,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890308930614028e-05,
+      "loss": 0.2681,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890185244384469e-05,
+      "loss": 0.9756,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900614925118995e-05,
+      "loss": 0.9668,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889937675005445e-05,
+      "loss": 0.9104,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889813791874235e-05,
+      "loss": 1.0143,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889689843127406e-05,
+      "loss": 0.839,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895658287740973e-05,
+      "loss": 0.9959,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894417488234547e-05,
+      "loss": 0.9549,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889317603284627e-05,
+      "loss": 1.0206,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889193392166769e-05,
+      "loss": 0.9342,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890691154790403e-05,
+      "loss": 0.9293,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889447732306056e-05,
+      "loss": 1.0312,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888203654306333e-05,
+      "loss": 1.0259,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886958920882974e-05,
+      "loss": 0.931,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888571353212777e-05,
+      "loss": 1.0323,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888446748813256e-05,
+      "loss": 0.9806,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888322078898922e-05,
+      "loss": 1.0033,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881973434789685e-05,
+      "loss": 0.9467,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880725425625944e-05,
+      "loss": 0.8871,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879476761590015e-05,
+      "loss": 1.0848,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887822744277398e-05,
+      "loss": 0.999,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876977469269967e-05,
+      "loss": 0.9483,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887572684117015e-05,
+      "loss": 0.9288,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874475558566748e-05,
+      "loss": 0.8758,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873223621552035e-05,
+      "loss": 0.9732,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871971030218326e-05,
+      "loss": 0.9641,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870717784657992e-05,
+      "loss": 0.8575,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869463884963447e-05,
+      "loss": 0.9385,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868209331227153e-05,
+      "loss": 0.8546,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866954123541624e-05,
+      "loss": 0.9523,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886569826199942e-05,
+      "loss": 0.8626,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864441746693147e-05,
+      "loss": 0.9349,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863184577715466e-05,
+      "loss": 1.0261,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861926755159076e-05,
+      "loss": 1.0151,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860668279116736e-05,
+      "loss": 0.9927,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885940914968124e-05,
+      "loss": 1.0564,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858149366945443e-05,
+      "loss": 1.0114,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856888931002243e-05,
+      "loss": 0.8586,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885562784194458e-05,
+      "loss": 0.2436,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885436609986545e-05,
+      "loss": 0.8869,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853103704857897e-05,
+      "loss": 1.0153,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885184065701501e-05,
+      "loss": 0.9555,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885057695642993e-05,
+      "loss": 0.889,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849312603195837e-05,
+      "loss": 0.881,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848047597405973e-05,
+      "loss": 0.917,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846781939153613e-05,
+      "loss": 0.9311,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845515628532093e-05,
+      "loss": 0.9445,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884424866563479e-05,
+      "loss": 0.958,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842981050555128e-05,
+      "loss": 0.9348,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884171278338659e-05,
+      "loss": 0.9342,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884044386422269e-05,
+      "loss": 0.9678,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839174293157005e-05,
+      "loss": 1.035,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883790407028315e-05,
+      "loss": 1.0309,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836633195694798e-05,
+      "loss": 0.8762,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835361669485662e-05,
+      "loss": 0.9409,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88340894917495e-05,
+      "loss": 0.9266,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832816662580133e-05,
+      "loss": 0.9464,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831543182071414e-05,
+      "loss": 0.891,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883026905031725e-05,
+      "loss": 0.7893,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88289942674116e-05,
+      "loss": 0.988,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882771883344847e-05,
+      "loss": 0.9124,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826442748521902e-05,
+      "loss": 0.9766,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825166012726004e-05,
+      "loss": 0.9691,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823888626154922e-05,
+      "loss": 0.9305,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822610588902847e-05,
+      "loss": 0.9883,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821331901064027e-05,
+      "loss": 0.9004,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820052562732757e-05,
+      "loss": 0.9324,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818772574003366e-05,
+      "loss": 0.9912,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881749193497025e-05,
+      "loss": 0.2589,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816210645727844e-05,
+      "loss": 0.8907,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881492870637063e-05,
+      "loss": 0.9549,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881364611699313e-05,
+      "loss": 0.9677,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812362877689942e-05,
+      "loss": 0.9771,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811078988555676e-05,
+      "loss": 0.9856,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809794449685015e-05,
+      "loss": 1.0167,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808509261172685e-05,
+      "loss": 0.9013,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880722342311345e-05,
+      "loss": 0.9902,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880593693560213e-05,
+      "loss": 0.8973,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804649798733597e-05,
+      "loss": 0.9274,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880336201260276e-05,
+      "loss": 1.0182,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802073577304582e-05,
+      "loss": 0.9136,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800784492934077e-05,
+      "loss": 0.9669,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87994947595863e-05,
+      "loss": 0.9485,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879820437735636e-05,
+      "loss": 0.9526,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796913346339404e-05,
+      "loss": 0.9365,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879562166663064e-05,
+      "loss": 0.2234,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879432933832532e-05,
+      "loss": 0.9756,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793036361518734e-05,
+      "loss": 0.9394,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791742736306225e-05,
+      "loss": 0.9106,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87904484627832e-05,
+      "loss": 0.8069,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789153541045088e-05,
+      "loss": 0.91,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878785797118738e-05,
+      "loss": 0.8149,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786561753305618e-05,
+      "loss": 1.0726,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785264887495375e-05,
+      "loss": 0.8808,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783967373852294e-05,
+      "loss": 0.9069,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878266921247205e-05,
+      "loss": 0.9489,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878137040345037e-05,
+      "loss": 0.8643,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878007094688303e-05,
+      "loss": 0.9814,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778770842865852e-05,
+      "loss": 0.2806,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877747009149471e-05,
+      "loss": 0.9574,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776168692865518e-05,
+      "loss": 1.0146,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774866647074244e-05,
+      "loss": 0.9593,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773563954216908e-05,
+      "loss": 1.0106,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772260614389562e-05,
+      "loss": 1.126,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877095662768832e-05,
+      "loss": 0.7817,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876965199420934e-05,
+      "loss": 0.9104,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768346714048822e-05,
+      "loss": 0.9058,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767040787303024e-05,
+      "loss": 0.9067,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765734214068247e-05,
+      "loss": 0.9803,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876442699444083e-05,
+      "loss": 0.956,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876311912851718e-05,
+      "loss": 0.9131,
+      "step": 2209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761810616393734e-05,
+      "loss": 0.839,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760501458166983e-05,
+      "loss": 0.9074,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759191653933464e-05,
+      "loss": 0.9988,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757881203789764e-05,
+      "loss": 0.2327,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875657010783252e-05,
+      "loss": 0.9897,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875525836615841e-05,
+      "loss": 0.94,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753945978864162e-05,
+      "loss": 0.9867,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752632946046557e-05,
+      "loss": 0.9599,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751319267802412e-05,
+      "loss": 0.9904,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875000494422861e-05,
+      "loss": 1.0891,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748689975422057e-05,
+      "loss": 0.9319,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874737436147973e-05,
+      "loss": 0.9054,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874605810249864e-05,
+      "loss": 0.8789,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744741198575845e-05,
+      "loss": 0.8843,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743423649808465e-05,
+      "loss": 0.9232,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742105456293645e-05,
+      "loss": 0.8939,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87407866181286e-05,
+      "loss": 0.9632,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739467135410575e-05,
+      "loss": 0.8506,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873814700823687e-05,
+      "loss": 0.8889,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873682623670484e-05,
+      "loss": 0.8941,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735504820911876e-05,
+      "loss": 1.0244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873418276095541e-05,
+      "loss": 0.9736,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732860056932946e-05,
+      "loss": 0.8583,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731536708942017e-05,
+      "loss": 0.989,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730212717080202e-05,
+      "loss": 0.9326,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728888081445142e-05,
+      "loss": 0.9624,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727562802134515e-05,
+      "loss": 0.8499,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872623687924604e-05,
+      "loss": 0.8107,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724910312877503e-05,
+      "loss": 0.9238,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872358310312672e-05,
+      "loss": 0.9623,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722255250091555e-05,
+      "loss": 0.9724,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720926753869938e-05,
+      "loss": 0.9089,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719597614559826e-05,
+      "loss": 0.9272,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718267832259232e-05,
+      "loss": 0.9751,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716937407066214e-05,
+      "loss": 0.9442,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715606339078877e-05,
+      "loss": 0.9936,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871427462839538e-05,
+      "loss": 0.9028,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712942275113923e-05,
+      "loss": 0.8272,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711609279332754e-05,
+      "loss": 0.8448,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871027564115017e-05,
+      "loss": 0.947,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870894136066451e-05,
+      "loss": 0.8653,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870760643797417e-05,
+      "loss": 0.9458,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870627087317759e-05,
+      "loss": 0.9394,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870493466637325e-05,
+      "loss": 0.2455,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703597817659688e-05,
+      "loss": 0.9337,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870226032713548e-05,
+      "loss": 0.9641,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700922194899258e-05,
+      "loss": 0.9768,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699583421049693e-05,
+      "loss": 0.9051,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869824400568551e-05,
+      "loss": 0.9619,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696903948905477e-05,
+      "loss": 0.9142,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695563250808414e-05,
+      "loss": 0.9919,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869422191149318e-05,
+      "loss": 1.0041,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869287993105869e-05,
+      "loss": 1.0161,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691537309603905e-05,
+      "loss": 0.9996,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690194047227823e-05,
+      "loss": 1.0033,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688850144029506e-05,
+      "loss": 0.9415,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868750560010805e-05,
+      "loss": 0.2458,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686160415562605e-05,
+      "loss": 0.9659,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684814590492366e-05,
+      "loss": 0.9361,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683468124996575e-05,
+      "loss": 0.8393,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682121019174515e-05,
+      "loss": 0.9695,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680773273125532e-05,
+      "loss": 1.017,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679424886949004e-05,
+      "loss": 0.8673,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678075860744366e-05,
+      "loss": 0.877,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676726194611096e-05,
+      "loss": 0.929,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867537588864872e-05,
+      "loss": 0.9601,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674024942956806e-05,
+      "loss": 0.952,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672673357634977e-05,
+      "loss": 0.9993,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671321132782897e-05,
+      "loss": 1.0104,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669968268500288e-05,
+      "loss": 0.874,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668614764886904e-05,
+      "loss": 0.9544,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866726062204256e-05,
+      "loss": 0.2596,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665905840067105e-05,
+      "loss": 0.876,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664550419060444e-05,
+      "loss": 0.9275,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663194359122527e-05,
+      "loss": 0.9312,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661837660353353e-05,
+      "loss": 0.9991,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866048032285296e-05,
+      "loss": 0.937,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659122346721447e-05,
+      "loss": 1.0052,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865776373205895e-05,
+      "loss": 0.9109,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865640447896565e-05,
+      "loss": 0.8975,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655044587541785e-05,
+      "loss": 0.8314,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653684057887635e-05,
+      "loss": 1.0161,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652322890103517e-05,
+      "loss": 0.8381,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650961084289815e-05,
+      "loss": 0.8536,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649598640546948e-05,
+      "loss": 0.9081,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864823555897538e-05,
+      "loss": 0.8857,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864687183967563e-05,
+      "loss": 1.0263,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645507482748256e-05,
+      "loss": 0.9402,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644142488293865e-05,
+      "loss": 1.0101,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642776856413124e-05,
+      "loss": 1.0017,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864141058720672e-05,
+      "loss": 0.934,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864004368077542e-05,
+      "loss": 0.907,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638676137220004e-05,
+      "loss": 0.9316,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637307956641326e-05,
+      "loss": 0.9052,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635939139140275e-05,
+      "loss": 0.9669,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634569684817787e-05,
+      "loss": 0.8665,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863319959377485e-05,
+      "loss": 1.1197,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863182886611249e-05,
+      "loss": 1.0132,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630457501931792e-05,
+      "loss": 0.994,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862908550133388e-05,
+      "loss": 0.8192,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627712864419926e-05,
+      "loss": 0.9825,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626339591291144e-05,
+      "loss": 0.871,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624965682048808e-05,
+      "loss": 0.8383,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623591136794226e-05,
+      "loss": 1.0119,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622215955628765e-05,
+      "loss": 0.8911,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862084013865382e-05,
+      "loss": 1.0231,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861946368597086e-05,
+      "loss": 1.0385,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618086597681375e-05,
+      "loss": 0.863,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616708873886913e-05,
+      "loss": 0.9296,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615330514689074e-05,
+      "loss": 0.9689,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613951520189493e-05,
+      "loss": 1.1196,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612571890489867e-05,
+      "loss": 1.0068,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611191625691923e-05,
+      "loss": 1.05,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609810725897447e-05,
+      "loss": 0.2389,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608429191208264e-05,
+      "loss": 0.261,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860704702172625e-05,
+      "loss": 0.9172,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605664217553337e-05,
+      "loss": 1.0217,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860428077879148e-05,
+      "loss": 1.0367,
+      "step": 2327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602896705542704e-05,
+      "loss": 0.9775,
+      "step": 2328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601511997909066e-05,
+      "loss": 0.9524,
+      "step": 2329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600126655992685e-05,
+      "loss": 0.9447,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598740679895706e-05,
+      "loss": 0.9176,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597354069720336e-05,
+      "loss": 0.8865,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859596682556883e-05,
+      "loss": 0.2603,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859457894754348e-05,
+      "loss": 1.0103,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593190435746623e-05,
+      "loss": 0.9889,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591801290280664e-05,
+      "loss": 0.2553,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590411511248024e-05,
+      "loss": 0.8671,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85890210987512e-05,
+      "loss": 0.9172,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587630052892715e-05,
+      "loss": 0.8813,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586238373775144e-05,
+      "loss": 0.9518,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584846061501115e-05,
+      "loss": 0.9467,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583453116173297e-05,
+      "loss": 1.0099,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858205953789441e-05,
+      "loss": 0.9627,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858066532676721e-05,
+      "loss": 0.9001,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579270482894508e-05,
+      "loss": 0.3001,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577875006379173e-05,
+      "loss": 0.9769,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576478897324096e-05,
+      "loss": 0.2372,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575082155832234e-05,
+      "loss": 0.9359,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857368478200658e-05,
+      "loss": 0.2421,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572286775950176e-05,
+      "loss": 0.9487,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857088813776612e-05,
+      "loss": 0.9108,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569488867557544e-05,
+      "loss": 0.9999,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568088965427628e-05,
+      "loss": 0.966,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566688431479603e-05,
+      "loss": 0.9665,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565287265816755e-05,
+      "loss": 0.9359,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563885468542393e-05,
+      "loss": 0.9492,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562483039759897e-05,
+      "loss": 0.9884,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561079979572682e-05,
+      "loss": 1.0324,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855967628808421e-05,
+      "loss": 0.8974,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855827196539798e-05,
+      "loss": 1.0318,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556867011617563e-05,
+      "loss": 0.9858,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855546142684656e-05,
+      "loss": 0.2722,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855405521118861e-05,
+      "loss": 0.8904,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855264836474741e-05,
+      "loss": 0.2909,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551240887626712e-05,
+      "loss": 0.8498,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85498327799303e-05,
+      "loss": 0.9985,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548424041762003e-05,
+      "loss": 1.0085,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854701467322571e-05,
+      "loss": 1.0136,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545604674425348e-05,
+      "loss": 0.8484,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544194045464888e-05,
+      "loss": 0.863,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542782786448352e-05,
+      "loss": 0.8865,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854137089747981e-05,
+      "loss": 0.8914,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539958378663372e-05,
+      "loss": 0.9352,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85385452301032e-05,
+      "loss": 0.9242,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537131451903506e-05,
+      "loss": 0.9474,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535717044168532e-05,
+      "loss": 0.9897,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534302007002586e-05,
+      "loss": 0.9241,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532886340510015e-05,
+      "loss": 0.9277,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531470044795207e-05,
+      "loss": 1.0779,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530053119962602e-05,
+      "loss": 0.9545,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528635566116683e-05,
+      "loss": 0.9792,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852721738336199e-05,
+      "loss": 0.9208,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525798571803092e-05,
+      "loss": 0.8914,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852437913154461e-05,
+      "loss": 0.9908,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852295906269123e-05,
+      "loss": 0.9296,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852153836534766e-05,
+      "loss": 1.0087,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852011703961866e-05,
+      "loss": 0.8573,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518695085609043e-05,
+      "loss": 0.9957,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517272503423668e-05,
+      "loss": 0.9385,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515849293167434e-05,
+      "loss": 1.0217,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514425454945294e-05,
+      "loss": 1.0089,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513000988862235e-05,
+      "loss": 1.0189,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511575895023306e-05,
+      "loss": 0.9885,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510150173533592e-05,
+      "loss": 1.0183,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508723824498226e-05,
+      "loss": 0.9729,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507296848022387e-05,
+      "loss": 1.0091,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85058692442113e-05,
+      "loss": 0.9999,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850444101317025e-05,
+      "loss": 0.9661,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850301215500454e-05,
+      "loss": 1.0256,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501582669819543e-05,
+      "loss": 0.9267,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500152557720674e-05,
+      "loss": 0.9438,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849872181881338e-05,
+      "loss": 1.0516,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849729045320318e-05,
+      "loss": 1.0186,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495858460995607e-05,
+      "loss": 0.9526,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849442584229627e-05,
+      "loss": 0.9963,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84929925972108e-05,
+      "loss": 0.8836,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84915587258449e-05,
+      "loss": 1.0136,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490124228304297e-05,
+      "loss": 0.8962,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848868910469477e-05,
+      "loss": 0.951,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848725335512215e-05,
+      "loss": 0.8916,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848581697969231e-05,
+      "loss": 0.9454,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484379978511168e-05,
+      "loss": 0.9476,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848294235168469e-05,
+      "loss": 0.8716,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848150409931889e-05,
+      "loss": 0.9072,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480065221519824e-05,
+      "loss": 0.8663,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478625718393594e-05,
+      "loss": 0.9717,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477185590046355e-05,
+      "loss": 0.9579,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84757448365843e-05,
+      "loss": 0.8991,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474303458113675e-05,
+      "loss": 0.9599,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847286145474076e-05,
+      "loss": 0.893,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471418826571897e-05,
+      "loss": 0.8962,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469975573713466e-05,
+      "loss": 0.9257,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468531696271893e-05,
+      "loss": 0.2384,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846708719435365e-05,
+      "loss": 0.8117,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465642068065257e-05,
+      "loss": 0.923,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464196317513276e-05,
+      "loss": 0.9448,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846274994280432e-05,
+      "loss": 0.8942,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461302944045045e-05,
+      "loss": 0.2986,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459855321342155e-05,
+      "loss": 0.8951,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84584070748024e-05,
+      "loss": 0.9171,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456958204532573e-05,
+      "loss": 0.8698,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455508710639513e-05,
+      "loss": 1.045,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454058593230108e-05,
+      "loss": 1.0245,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452607852411297e-05,
+      "loss": 0.8751,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845115648829005e-05,
+      "loss": 0.9634,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449704500973396e-05,
+      "loss": 0.2697,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448251890568408e-05,
+      "loss": 1.0026,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446798657182197e-05,
+      "loss": 1.039,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445344800921932e-05,
+      "loss": 0.8162,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443890321894815e-05,
+      "loss": 0.2674,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844243522020811e-05,
+      "loss": 0.9525,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440979495969108e-05,
+      "loss": 0.8404,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439523149285154e-05,
+      "loss": 0.9257,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843806618026365e-05,
+      "loss": 0.9046,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436608589012028e-05,
+      "loss": 0.8667,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435150375637773e-05,
+      "loss": 0.9681,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433691540248417e-05,
+      "loss": 1.0139,
+      "step": 2447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843223208295153e-05,
+      "loss": 0.8154,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843077200385474e-05,
+      "loss": 1.0194,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429311303065712e-05,
+      "loss": 0.8994,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427849980692155e-05,
+      "loss": 0.856,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426388036841838e-05,
+      "loss": 0.9837,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424925471622557e-05,
+      "loss": 0.9598,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423462285142166e-05,
+      "loss": 1.0045,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842199847750856e-05,
+      "loss": 0.8933,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420534048829685e-05,
+      "loss": 1.0113,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841906899921353e-05,
+      "loss": 0.9165,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417603328768123e-05,
+      "loss": 0.8905,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416137037601547e-05,
+      "loss": 0.2338,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841467012582193e-05,
+      "loss": 0.9539,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413202593537437e-05,
+      "loss": 0.8282,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841173444085629e-05,
+      "loss": 0.9884,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841026566788675e-05,
+      "loss": 0.8188,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840879627473713e-05,
+      "loss": 1.0139,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407326261515778e-05,
+      "loss": 1.0037,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405855628331097e-05,
+      "loss": 1.0601,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404384375291533e-05,
+      "loss": 0.8854,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402912502505578e-05,
+      "loss": 0.9073,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401440010081765e-05,
+      "loss": 0.9379,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399966898128682e-05,
+      "loss": 0.888,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398493166754954e-05,
+      "loss": 0.8832,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397018816069254e-05,
+      "loss": 0.906,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395543846180308e-05,
+      "loss": 0.9723,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839406825719688e-05,
+      "loss": 0.9055,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839259204922777e-05,
+      "loss": 0.9564,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839111522238185e-05,
+      "loss": 0.9119,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838963777676802e-05,
+      "loss": 0.2258,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388159712495217e-05,
+      "loss": 0.9541,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386681029672448e-05,
+      "loss": 0.9185,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385201728408742e-05,
+      "loss": 0.9608,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838372180881319e-05,
+      "loss": 0.8928,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382241270994924e-05,
+      "loss": 0.8929,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380760115063113e-05,
+      "loss": 0.9208,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379278341126983e-05,
+      "loss": 1.0202,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83777959492958e-05,
+      "loss": 0.8861,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837631293967888e-05,
+      "loss": 0.902,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374829312385576e-05,
+      "loss": 0.7916,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83733450675253e-05,
+      "loss": 0.8983,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371860205207494e-05,
+      "loss": 0.9643,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370374725541655e-05,
+      "loss": 0.9948,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368888628637323e-05,
+      "loss": 0.9037,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367401914604088e-05,
+      "loss": 0.9305,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365914583551577e-05,
+      "loss": 0.2972,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836442663558947e-05,
+      "loss": 0.9361,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362938070827483e-05,
+      "loss": 1.0293,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361448889375394e-05,
+      "loss": 0.966,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835995909134301e-05,
+      "loss": 0.8915,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358468676840196e-05,
+      "loss": 0.8141,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356977645976847e-05,
+      "loss": 0.9078,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835548599886292e-05,
+      "loss": 0.9265,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353993735608407e-05,
+      "loss": 0.9768,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835250085632335e-05,
+      "loss": 0.9526,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835100736111784e-05,
+      "loss": 0.9761,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349513250102e-05,
+      "loss": 0.949,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834801852338601e-05,
+      "loss": 0.9389,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834652318108009e-05,
+      "loss": 0.9345,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834502722329452e-05,
+      "loss": 0.9171,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343530650139597e-05,
+      "loss": 0.9443,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834203346172569e-05,
+      "loss": 0.9951,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340535658163198e-05,
+      "loss": 0.9765,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833903723956257e-05,
+      "loss": 0.9721,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833753820603431e-05,
+      "loss": 0.9019,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336038557688945e-05,
+      "loss": 0.984,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334538294637066e-05,
+      "loss": 0.8809,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333037416989306e-05,
+      "loss": 0.8644,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331535924856337e-05,
+      "loss": 0.9755,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330033818348884e-05,
+      "loss": 0.9532,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328531097577708e-05,
+      "loss": 0.8797,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327027762653625e-05,
+      "loss": 0.9975,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325523813687495e-05,
+      "loss": 1.006,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324019250790214e-05,
+      "loss": 1.0121,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322514074072738e-05,
+      "loss": 1.074,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832100828364605e-05,
+      "loss": 0.9312,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319501879621194e-05,
+      "loss": 0.875,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317994862109253e-05,
+      "loss": 1.0006,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831648723122135e-05,
+      "loss": 1.0209,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314978987068672e-05,
+      "loss": 0.9129,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831347012976243e-05,
+      "loss": 0.7848,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311960659413883e-05,
+      "loss": 0.8286,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310450576134352e-05,
+      "loss": 0.9884,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308939880035184e-05,
+      "loss": 0.9333,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830742857122778e-05,
+      "loss": 0.8799,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305916649823593e-05,
+      "loss": 0.8896,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83044041159341e-05,
+      "loss": 1.0108,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302890969670846e-05,
+      "loss": 0.9445,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830137721114541e-05,
+      "loss": 0.8695,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299862840469414e-05,
+      "loss": 0.9357,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298347857754534e-05,
+      "loss": 1.0207,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296832263112484e-05,
+      "loss": 0.9667,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295316056655026e-05,
+      "loss": 1.0673,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293799238493963e-05,
+      "loss": 0.9205,
+      "step": 2541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829228180874115e-05,
+      "loss": 0.9607,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290763767508483e-05,
+      "loss": 0.9995,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289245114907904e-05,
+      "loss": 0.9354,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82877258510514e-05,
+      "loss": 0.8893,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286205976050998e-05,
+      "loss": 1.0148,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284685490018782e-05,
+      "loss": 1.002,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828316439306687e-05,
+      "loss": 0.2867,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281642685307428e-05,
+      "loss": 0.9416,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828012036685267e-05,
+      "loss": 0.945,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827859743781485e-05,
+      "loss": 0.9344,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277073898306275e-05,
+      "loss": 0.9498,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275549748439294e-05,
+      "loss": 0.8621,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827402498832629e-05,
+      "loss": 1.0298,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272499618079703e-05,
+      "loss": 0.9097,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270973637812023e-05,
+      "loss": 0.8282,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269447047635768e-05,
+      "loss": 0.9763,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267919847663513e-05,
+      "loss": 0.9725,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266392038007873e-05,
+      "loss": 0.8857,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264863618781516e-05,
+      "loss": 1.0316,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263334590097142e-05,
+      "loss": 0.9666,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826180495206751e-05,
+      "loss": 1.0125,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260274704805406e-05,
+      "loss": 0.9029,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258743848423683e-05,
+      "loss": 1.0665,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257212383035222e-05,
+      "loss": 0.9486,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825568030875296e-05,
+      "loss": 0.9525,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254147625689862e-05,
+      "loss": 0.8228,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825261433395896e-05,
+      "loss": 0.898,
+      "step": 2568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251080433673318e-05,
+      "loss": 1.0171,
+      "step": 2569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824954592494604e-05,
+      "loss": 0.9523,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248010807890294e-05,
+      "loss": 0.9933,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824647508261927e-05,
+      "loss": 0.923,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244938749246222e-05,
+      "loss": 0.2756,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243401807884435e-05,
+      "loss": 0.976,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241864258647247e-05,
+      "loss": 0.9719,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240326101648036e-05,
+      "loss": 0.9809,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238787337000227e-05,
+      "loss": 0.8949,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823724796481729e-05,
+      "loss": 0.934,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235707985212744e-05,
+      "loss": 0.9101,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234167398300146e-05,
+      "loss": 0.2651,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232626204193094e-05,
+      "loss": 0.9543,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231084403005247e-05,
+      "loss": 1.0155,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229541994850294e-05,
+      "loss": 0.7975,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822799897984197e-05,
+      "loss": 0.9395,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226455358094064e-05,
+      "loss": 0.9005,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82249111297204e-05,
+      "loss": 0.9228,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223366294834855e-05,
+      "loss": 1.0381,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822182085355134e-05,
+      "loss": 0.9397,
+      "step": 2588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220274805983828e-05,
+      "loss": 0.9254,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821872815224631e-05,
+      "loss": 0.8776,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821718089245285e-05,
+      "loss": 0.8668,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215633026717543e-05,
+      "loss": 0.9877,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214084555154523e-05,
+      "loss": 0.9499,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212535477877982e-05,
+      "loss": 0.9842,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210985795002147e-05,
+      "loss": 0.9078,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209435506641297e-05,
+      "loss": 0.9336,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207884612909746e-05,
+      "loss": 0.8939,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820633311392186e-05,
+      "loss": 0.9766,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204781009792054e-05,
+      "loss": 0.9984,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203228300634773e-05,
+      "loss": 0.892,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201674986564516e-05,
+      "loss": 1.0023,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820012106769583e-05,
+      "loss": 0.9453,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198566544143302e-05,
+      "loss": 0.8448,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197011416021562e-05,
+      "loss": 0.96,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195455683445283e-05,
+      "loss": 0.9359,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193899346529194e-05,
+      "loss": 0.274,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192342405388058e-05,
+      "loss": 0.9561,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190784860136677e-05,
+      "loss": 0.9069,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189226710889917e-05,
+      "loss": 0.8945,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818766795776267e-05,
+      "loss": 0.9489,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186108600869883e-05,
+      "loss": 0.9745,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184548640326543e-05,
+      "loss": 0.8731,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182988076247686e-05,
+      "loss": 0.8891,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181426908748384e-05,
+      "loss": 0.8159,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817986513794376e-05,
+      "loss": 0.8583,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178302763948984e-05,
+      "loss": 0.9201,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176739786879264e-05,
+      "loss": 0.9259,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175176206849856e-05,
+      "loss": 0.8147,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173612023976058e-05,
+      "loss": 1.013,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172047238373215e-05,
+      "loss": 0.9727,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817048185015672e-05,
+      "loss": 0.894,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168915859442e-05,
+      "loss": 1.0548,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167349266344535e-05,
+      "loss": 0.9615,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816578207097985e-05,
+      "loss": 0.2251,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81642142734635e-05,
+      "loss": 1.0357,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816264587391111e-05,
+      "loss": 0.9954,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161076872438327e-05,
+      "loss": 0.9636,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159507269160854e-05,
+      "loss": 0.9629,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157937064194427e-05,
+      "loss": 0.8815,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156366257654844e-05,
+      "loss": 1.0013,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154794849657938e-05,
+      "loss": 0.9598,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815322284031958e-05,
+      "loss": 0.8864,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815165022975569e-05,
+      "loss": 1.0749,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150077018082237e-05,
+      "loss": 0.8515,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814850320541523e-05,
+      "loss": 0.973,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146928791870727e-05,
+      "loss": 0.9601,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145353777564826e-05,
+      "loss": 1.0351,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814377816261366e-05,
+      "loss": 0.9125,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814220194713343e-05,
+      "loss": 0.8076,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814062513124036e-05,
+      "loss": 0.9041,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813904771505073e-05,
+      "loss": 0.9013,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137469698680848e-05,
+      "loss": 0.9336,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813589108224709e-05,
+      "loss": 0.9703,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134311865865868e-05,
+      "loss": 0.9993,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132732049653625e-05,
+      "loss": 0.978,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813115163372686e-05,
+      "loss": 0.948,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129570618202116e-05,
+      "loss": 0.9235,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127989003195977e-05,
+      "loss": 0.9293,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126406788825076e-05,
+      "loss": 0.9926,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124823975206084e-05,
+      "loss": 0.8505,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123240562455717e-05,
+      "loss": 0.9333,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812165655069074e-05,
+      "loss": 0.9558,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120071940027963e-05,
+      "loss": 0.8955,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118486730584228e-05,
+      "loss": 0.8773,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116900922476436e-05,
+      "loss": 0.8998,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115314515821527e-05,
+      "loss": 0.9395,
+      "step": 2656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811372751073648e-05,
+      "loss": 0.9059,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811213990733832e-05,
+      "loss": 0.9587,
+      "step": 2658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110551705744123e-05,
+      "loss": 1.0507,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108962906071e-05,
+      "loss": 0.9551,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107373508436116e-05,
+      "loss": 0.92,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810578351295667e-05,
+      "loss": 0.9952,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104192919749916e-05,
+      "loss": 0.9561,
+      "step": 2663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102601728933137e-05,
+      "loss": 1.0142,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810100994062367e-05,
+      "loss": 1.0616,
+      "step": 2665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099417554938902e-05,
+      "loss": 0.9127,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097824571996247e-05,
+      "loss": 0.914,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096230991913183e-05,
+      "loss": 0.8672,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094636814807216e-05,
+      "loss": 0.8553,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80930420407959e-05,
+      "loss": 0.9738,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091446669996844e-05,
+      "loss": 0.9035,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089850702527684e-05,
+      "loss": 0.9548,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808825413850611e-05,
+      "loss": 0.9393,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086656978049853e-05,
+      "loss": 0.9463,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085059221276687e-05,
+      "loss": 0.9961,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808346086830444e-05,
+      "loss": 0.8851,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808186191925097e-05,
+      "loss": 0.8752,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080262374234186e-05,
+      "loss": 0.9207,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078662233372037e-05,
+      "loss": 0.8311,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077061496782523e-05,
+      "loss": 0.9481,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807546016458368e-05,
+      "loss": 0.9303,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80738582368936e-05,
+      "loss": 1.0024,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072255713830396e-05,
+      "loss": 0.918,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807065259551225e-05,
+      "loss": 0.9322,
+      "step": 2684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069048882057377e-05,
+      "loss": 0.9855,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806744457358403e-05,
+      "loss": 0.9081,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806583967021052e-05,
+      "loss": 0.9909,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806423417205519e-05,
+      "loss": 0.9582,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062628079236428e-05,
+      "loss": 0.9925,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061021391872667e-05,
+      "loss": 0.971,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059414110082398e-05,
+      "loss": 0.9506,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805780623398413e-05,
+      "loss": 0.8934,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056197763696435e-05,
+      "loss": 0.9132,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054588699337925e-05,
+      "loss": 0.9885,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052979041027248e-05,
+      "loss": 0.9021,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051368788883106e-05,
+      "loss": 0.7982,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049757943024237e-05,
+      "loss": 0.89,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804814650356943e-05,
+      "loss": 0.9927,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804653447063751e-05,
+      "loss": 0.8859,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044921844347353e-05,
+      "loss": 0.9102,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804330862481787e-05,
+      "loss": 0.9011,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041694812168035e-05,
+      "loss": 0.9753,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040080406516836e-05,
+      "loss": 0.9369,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038465407983327e-05,
+      "loss": 0.9047,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036849816686597e-05,
+      "loss": 0.9906,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035233632745782e-05,
+      "loss": 0.9483,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033616856280067e-05,
+      "loss": 0.9342,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031999487408667e-05,
+      "loss": 0.9405,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803038152625085e-05,
+      "loss": 0.9239,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028762972925925e-05,
+      "loss": 0.9859,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802714382755325e-05,
+      "loss": 0.8847,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802552409025222e-05,
+      "loss": 0.9896,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802390376114227e-05,
+      "loss": 0.865,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802228284034289e-05,
+      "loss": 0.9576,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802066132797361e-05,
+      "loss": 1.0925,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019039224153996e-05,
+      "loss": 0.97,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017416529003668e-05,
+      "loss": 0.9595,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015793242642283e-05,
+      "loss": 0.9886,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014169365189544e-05,
+      "loss": 0.9525,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012544896765197e-05,
+      "loss": 0.9653,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801091983748903e-05,
+      "loss": 0.9925,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009294187480877e-05,
+      "loss": 0.9728,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800766794686062e-05,
+      "loss": 0.8644,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800604111574817e-05,
+      "loss": 0.9001,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004413694263497e-05,
+      "loss": 0.9685,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002785682526608e-05,
+      "loss": 0.9781,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800115708065755e-05,
+      "loss": 0.9591,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999527888776425e-05,
+      "loss": 0.8797,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997898107003365e-05,
+      "loss": 0.9692,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799626773545855e-05,
+      "loss": 0.8788,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994636774262213e-05,
+      "loss": 0.2604,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993005223534613e-05,
+      "loss": 0.8953,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799137308339607e-05,
+      "loss": 0.9411,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989740353966935e-05,
+      "loss": 1.0043,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988107035367607e-05,
+      "loss": 1.0746,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798647312771853e-05,
+      "loss": 0.8275,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984838631140185e-05,
+      "loss": 0.9819,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798320354575311e-05,
+      "loss": 1.0086,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981567871677866e-05,
+      "loss": 0.936,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797993160903508e-05,
+      "loss": 0.9758,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978294757945404e-05,
+      "loss": 0.9738,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976657318529548e-05,
+      "loss": 0.9482,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797501929090825e-05,
+      "loss": 0.9307,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973380675202304e-05,
+      "loss": 0.9537,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971741471532544e-05,
+      "loss": 0.9717,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797010168001984e-05,
+      "loss": 0.2243,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796846130078512e-05,
+      "loss": 0.9233,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966820333949343e-05,
+      "loss": 1.0062,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796517877963351e-05,
+      "loss": 0.8837,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796353663795868e-05,
+      "loss": 0.9619,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961893909045943e-05,
+      "loss": 0.2411,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796025059301643e-05,
+      "loss": 0.855,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795860668999133e-05,
+      "loss": 0.9502,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956962200091856e-05,
+      "loss": 0.9235,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795531712343928e-05,
+      "loss": 0.8779,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953671460154905e-05,
+      "loss": 0.962,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79520252103601e-05,
+      "loss": 0.9429,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950378374176235e-05,
+      "loss": 0.9567,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794873095172477e-05,
+      "loss": 0.893,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947082943127182e-05,
+      "loss": 0.9391,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945434348504993e-05,
+      "loss": 1.0259,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943785167979774e-05,
+      "loss": 0.9808,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794213540167314e-05,
+      "loss": 0.9962,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794048504970674e-05,
+      "loss": 0.9357,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793883411220227e-05,
+      "loss": 0.9906,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937182589281485e-05,
+      "loss": 0.8795,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793553048106616e-05,
+      "loss": 0.9054,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793387778767812e-05,
+      "loss": 1.0589,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932224509239243e-05,
+      "loss": 0.9547,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930570645871444e-05,
+      "loss": 0.9819,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928916197696674e-05,
+      "loss": 1.0599,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927261164836935e-05,
+      "loss": 0.9991,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792560554741427e-05,
+      "loss": 0.9943,
+      "step": 2773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923949345550774e-05,
+      "loss": 1.0228,
+      "step": 2774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922292559368565e-05,
+      "loss": 1.0319,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792063518898982e-05,
+      "loss": 0.9508,
+      "step": 2776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918977234536758e-05,
+      "loss": 0.9113,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917318696131634e-05,
+      "loss": 0.8572,
+      "step": 2778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915659573896755e-05,
+      "loss": 0.927,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791399986795446e-05,
+      "loss": 0.888,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912339578427142e-05,
+      "loss": 0.9481,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791067870543723e-05,
+      "loss": 0.8954,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909017249107193e-05,
+      "loss": 0.9303,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790735520955956e-05,
+      "loss": 0.897,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905692586916883e-05,
+      "loss": 0.9264,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904029381301767e-05,
+      "loss": 0.9576,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902365592836853e-05,
+      "loss": 0.9376,
+      "step": 2787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790070122164484e-05,
+      "loss": 0.9602,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899036267848458e-05,
+      "loss": 0.9977,
+      "step": 2789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897370731570475e-05,
+      "loss": 0.9208,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895704612933715e-05,
+      "loss": 0.8589,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789403791206104e-05,
+      "loss": 0.9378,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892370629075347e-05,
+      "loss": 0.9672,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789070276409959e-05,
+      "loss": 0.9172,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889034317256754e-05,
+      "loss": 0.8737,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887365288669874e-05,
+      "loss": 0.9465,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885695678462024e-05,
+      "loss": 0.9938,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884025486756328e-05,
+      "loss": 0.9334,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788235471367594e-05,
+      "loss": 0.9613,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880683359344065e-05,
+      "loss": 0.8907,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879011423883957e-05,
+      "loss": 0.9569,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877338907418898e-05,
+      "loss": 1.0114,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787566581007223e-05,
+      "loss": 0.9573,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873992131967317e-05,
+      "loss": 1.0242,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872317873227585e-05,
+      "loss": 0.8619,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870643033976493e-05,
+      "loss": 1.0257,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868967614337544e-05,
+      "loss": 0.917,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867291614434288e-05,
+      "loss": 0.8748,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865615034390312e-05,
+      "loss": 0.9492,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863937874329253e-05,
+      "loss": 0.923,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862260134374778e-05,
+      "loss": 0.9848,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786058181465061e-05,
+      "loss": 0.9103,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858902915280514e-05,
+      "loss": 0.8341,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857223436388283e-05,
+      "loss": 0.8962,
+      "step": 2814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785554337809777e-05,
+      "loss": 0.9777,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853862740532867e-05,
+      "loss": 0.9082,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852181523817497e-05,
+      "loss": 0.9057,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785049972807564e-05,
+      "loss": 0.8296,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784881735343131e-05,
+      "loss": 0.2629,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847134400008572e-05,
+      "loss": 0.9755,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845450867931524e-05,
+      "loss": 1.0456,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843766757324312e-05,
+      "loss": 1.0225,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784208206831112e-05,
+      "loss": 0.2658,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840396801016187e-05,
+      "loss": 0.915,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783871095556378e-05,
+      "loss": 0.9625,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837024532078214e-05,
+      "loss": 1.0162,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783533753068385e-05,
+      "loss": 0.9586,
+      "step": 2827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783364995150509e-05,
+      "loss": 0.9164,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831961794666373e-05,
+      "loss": 0.8391,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783027306029219e-05,
+      "loss": 0.2676,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782858374850707e-05,
+      "loss": 1.0164,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782689385943557e-05,
+      "loss": 0.8803,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825203393202325e-05,
+      "loss": 0.999,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823512349931983e-05,
+      "loss": 0.874,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821820729749236e-05,
+      "loss": 0.9333,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820128532778834e-05,
+      "loss": 0.9274,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818435759145562e-05,
+      "loss": 1.0779,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816742408974236e-05,
+      "loss": 0.7623,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781504848238974e-05,
+      "loss": 1.0013,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781335397951697e-05,
+      "loss": 0.9986,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781165890048089e-05,
+      "loss": 0.9147,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809963245406498e-05,
+      "loss": 0.9676,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780826701441882e-05,
+      "loss": 0.8926,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806570207642956e-05,
+      "loss": 0.9899,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804872825204017e-05,
+      "loss": 0.9495,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803174867227176e-05,
+      "loss": 0.9708,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801476333837636e-05,
+      "loss": 0.9871,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799777225160652e-05,
+      "loss": 0.8561,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798077541321513e-05,
+      "loss": 0.9282,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796377282445567e-05,
+      "loss": 0.9521,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794676448658182e-05,
+      "loss": 0.975,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779297504008478e-05,
+      "loss": 0.989,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791273056850825e-05,
+      "loss": 0.9453,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789570499081822e-05,
+      "loss": 0.9196,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787867366903325e-05,
+      "loss": 0.9345,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786163660440914e-05,
+      "loss": 0.9934,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784459379820236e-05,
+      "loss": 0.942,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782754525166952e-05,
+      "loss": 0.9683,
+      "step": 2858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778104909660679e-05,
+      "loss": 0.9143,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77793430942655e-05,
+      "loss": 1.0316,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777763651826889e-05,
+      "loss": 0.8986,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775929368742803e-05,
+      "loss": 0.9372,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777422164581313e-05,
+      "loss": 0.8554,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777251334960579e-05,
+      "loss": 0.2985,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770804480246763e-05,
+      "loss": 0.9863,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776909503786206e-05,
+      "loss": 1.0895,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767385022577735e-05,
+      "loss": 0.993,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776567443451989e-05,
+      "loss": 0.7879,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776396327381466e-05,
+      "loss": 0.9308,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762251540588228e-05,
+      "loss": 0.8124,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760539234966825e-05,
+      "loss": 0.9181,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775882635707671e-05,
+      "loss": 1.0676,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.974,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775539888499564e-05,
+      "loss": 0.2595,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753684291057423e-05,
+      "loss": 0.8775,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751969125355988e-05,
+      "loss": 0.9055,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775025338801781e-05,
+      "loss": 0.8802,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748537079169414e-05,
+      "loss": 1.031,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746820198937355e-05,
+      "loss": 0.9084,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745102747448242e-05,
+      "loss": 0.9506,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743384724828722e-05,
+      "loss": 0.9154,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741666131205478e-05,
+      "loss": 0.9791,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739946966705247e-05,
+      "loss": 0.9788,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738227231454795e-05,
+      "loss": 0.9275,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736506925580947e-05,
+      "loss": 0.9654,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734786049210548e-05,
+      "loss": 0.9559,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733064602470503e-05,
+      "loss": 0.921,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731342585487753e-05,
+      "loss": 0.914,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729619998389277e-05,
+      "loss": 1.073,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727896841302108e-05,
+      "loss": 0.8541,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726173114353303e-05,
+      "loss": 0.936,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724448817669978e-05,
+      "loss": 0.9446,
+      "step": 2892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722723951379283e-05,
+      "loss": 0.9973,
+      "step": 2893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720998515608413e-05,
+      "loss": 0.9737,
+      "step": 2894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719272510484597e-05,
+      "loss": 0.9089,
+      "step": 2895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717545936135115e-05,
+      "loss": 0.9467,
+      "step": 2896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771581879268729e-05,
+      "loss": 1.0439,
+      "step": 2897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714091080268477e-05,
+      "loss": 1.0388,
+      "step": 2898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712362799006084e-05,
+      "loss": 0.9099,
+      "step": 2899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771063394902755e-05,
+      "loss": 0.9322,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708904530460363e-05,
+      "loss": 0.9886,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770717454343206e-05,
+      "loss": 0.8946,
+      "step": 2902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705443988070204e-05,
+      "loss": 1.0714,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770371286450241e-05,
+      "loss": 0.9255,
+      "step": 2904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770198117285633e-05,
+      "loss": 0.9842,
+      "step": 2905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700248913259663e-05,
+      "loss": 0.8788,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698516085840146e-05,
+      "loss": 0.9115,
+      "step": 2907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696782690725564e-05,
+      "loss": 1.0333,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695048728043728e-05,
+      "loss": 0.9373,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693314197922513e-05,
+      "loss": 0.9754,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691579100489818e-05,
+      "loss": 0.9245,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689843435873596e-05,
+      "loss": 0.8878,
+      "step": 2912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688107204201832e-05,
+      "loss": 0.8734,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768637040560256e-05,
+      "loss": 0.9384,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684633040203847e-05,
+      "loss": 0.8393,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682895108133818e-05,
+      "loss": 0.2616,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768115660952062e-05,
+      "loss": 0.8592,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679417544492455e-05,
+      "loss": 0.9937,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677677913177566e-05,
+      "loss": 0.9795,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767593771570423e-05,
+      "loss": 1.0457,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767419695220077e-05,
+      "loss": 1.1326,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7672455622795558e-05,
+      "loss": 0.9462,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670713727616998e-05,
+      "loss": 0.912,
+      "step": 2923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766897126679354e-05,
+      "loss": 0.9516,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766722824045367e-05,
+      "loss": 0.9833,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766548464872592e-05,
+      "loss": 0.8458,
+      "step": 2926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766374049173887e-05,
+      "loss": 1.0063,
+      "step": 2927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661995769621133e-05,
+      "loss": 0.8972,
+      "step": 2928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660250482501365e-05,
+      "loss": 0.9389,
+      "step": 2929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765850463050826e-05,
+      "loss": 0.8977,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656758213770572e-05,
+      "loss": 0.2792,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655011232417075e-05,
+      "loss": 0.9243,
+      "step": 2932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765326368657659e-05,
+      "loss": 0.9122,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651515576377985e-05,
+      "loss": 1.0428,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764976690195017e-05,
+      "loss": 0.9585,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648017663422094e-05,
+      "loss": 0.9017,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646267860922743e-05,
+      "loss": 1.0529,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764451749458115e-05,
+      "loss": 0.9057,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764276656452639e-05,
+      "loss": 0.9012,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641015070887578e-05,
+      "loss": 0.9189,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639263013793872e-05,
+      "loss": 0.8645,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637510393374463e-05,
+      "loss": 0.9494,
+      "step": 2942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76357572097586e-05,
+      "loss": 0.9997,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763400346307556e-05,
+      "loss": 1.0262,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632249153454663e-05,
+      "loss": 0.9924,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763049428102528e-05,
+      "loss": 0.9871,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628738845916812e-05,
+      "loss": 1.0224,
+      "step": 2947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626982848258705e-05,
+      "loss": 0.9092,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625226288180456e-05,
+      "loss": 0.9707,
+      "step": 2949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623469165811585e-05,
+      "loss": 0.9346,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621711481281665e-05,
+      "loss": 0.9783,
+      "step": 2951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619953234720316e-05,
+      "loss": 0.8384,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618194426257188e-05,
+      "loss": 0.2747,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761643505602198e-05,
+      "loss": 0.92,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614675124144423e-05,
+      "loss": 0.9738,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76129146307543e-05,
+      "loss": 1.0814,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611153575981435e-05,
+      "loss": 0.9486,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609391959955683e-05,
+      "loss": 0.2552,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607629782806952e-05,
+      "loss": 1.0235,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605867044665186e-05,
+      "loss": 0.9761,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604103745660364e-05,
+      "loss": 1.0223,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602339885922523e-05,
+      "loss": 1.0288,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600575465581723e-05,
+      "loss": 0.8741,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759881048476808e-05,
+      "loss": 0.9253,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597044943611744e-05,
+      "loss": 0.9689,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595278842242905e-05,
+      "loss": 0.9323,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75935121807918e-05,
+      "loss": 0.8848,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75917449593887e-05,
+      "loss": 0.8107,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758997717816392e-05,
+      "loss": 0.8538,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588208837247827e-05,
+      "loss": 1.0162,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586439936770814e-05,
+      "loss": 1.0122,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758467047686332e-05,
+      "loss": 0.8607,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582900457655826e-05,
+      "loss": 0.9346,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758112987927886e-05,
+      "loss": 0.919,
+      "step": 2974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579358741862983e-05,
+      "loss": 0.9862,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75775870455388e-05,
+      "loss": 0.9552,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575814790436954e-05,
+      "loss": 0.9106,
+      "step": 2977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574041976688135e-05,
+      "loss": 0.9658,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572268604423077e-05,
+      "loss": 0.9434,
+      "step": 2979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570494673772545e-05,
+      "loss": 0.9153,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568720184867348e-05,
+      "loss": 0.9525,
+      "step": 2981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756694513783834e-05,
+      "loss": 0.9818,
+      "step": 2982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756516953281642e-05,
+      "loss": 0.8403,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563393369932518e-05,
+      "loss": 1.0314,
+      "step": 2984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561616649317607e-05,
+      "loss": 0.8054,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559839371102706e-05,
+      "loss": 0.2861,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558061535418874e-05,
+      "loss": 0.9655,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556283142397213e-05,
+      "loss": 0.9371,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755450419216886e-05,
+      "loss": 0.8851,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755272468486499e-05,
+      "loss": 0.8434,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550944620616838e-05,
+      "loss": 0.9279,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754916399955566e-05,
+      "loss": 0.9551,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547382821812756e-05,
+      "loss": 1.0276,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545601087519482e-05,
+      "loss": 0.9642,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543818796807223e-05,
+      "loss": 0.8521,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542035949807397e-05,
+      "loss": 1.0174,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754025254665148e-05,
+      "loss": 0.8516,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538468587470984e-05,
+      "loss": 0.9488,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536684072397455e-05,
+      "loss": 0.9379,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534899001562485e-05,
+      "loss": 0.9569,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533113375097708e-05,
+      "loss": 0.9463,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531327193134798e-05,
+      "loss": 0.908,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529540455805466e-05,
+      "loss": 0.9634,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527753163241474e-05,
+      "loss": 0.9891,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525965315574612e-05,
+      "loss": 0.2727,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752417691293672e-05,
+      "loss": 0.9005,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522387955459678e-05,
+      "loss": 1.04,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520598443275402e-05,
+      "loss": 0.9805,
+      "step": 3008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518808376515856e-05,
+      "loss": 0.97,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751701775531304e-05,
+      "loss": 0.2268,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751522657979899e-05,
+      "loss": 1.0111,
+      "step": 3011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513434850105797e-05,
+      "loss": 0.9093,
+      "step": 3012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751164256636558e-05,
+      "loss": 0.922,
+      "step": 3013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509849728710505e-05,
+      "loss": 0.2682,
+      "step": 3014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508056337272777e-05,
+      "loss": 0.9785,
+      "step": 3015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506262392184642e-05,
+      "loss": 0.2863,
+      "step": 3016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504467893578388e-05,
+      "loss": 0.8821,
+      "step": 3017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750267284158634e-05,
+      "loss": 0.93,
+      "step": 3018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500877236340872e-05,
+      "loss": 1.0288,
+      "step": 3019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749908107797439e-05,
+      "loss": 0.9618,
+      "step": 3020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749728436661934e-05,
+      "loss": 0.2919,
+      "step": 3021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749548710240822e-05,
+      "loss": 0.9587,
+      "step": 3022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749368928547356e-05,
+      "loss": 0.9848,
+      "step": 3023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491890915947934e-05,
+      "loss": 0.8784,
+      "step": 3024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749009199396395e-05,
+      "loss": 0.983,
+      "step": 3025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488292519654267e-05,
+      "loss": 0.8611,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486492493151578e-05,
+      "loss": 0.932,
+      "step": 3027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484691914588613e-05,
+      "loss": 0.9482,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748289078409816e-05,
+      "loss": 0.9641,
+      "step": 3029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481089101813026e-05,
+      "loss": 0.2559,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479286867866074e-05,
+      "loss": 1.117,
+      "step": 3031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74774840823902e-05,
+      "loss": 0.9301,
+      "step": 3032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747568074551834e-05,
+      "loss": 0.8305,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747387685738348e-05,
+      "loss": 0.8923,
+      "step": 3034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472072418118636e-05,
+      "loss": 0.9285,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747026742785687e-05,
+      "loss": 0.9753,
+      "step": 3036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746846188673128e-05,
+      "loss": 0.9079,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466655794875017e-05,
+      "loss": 0.9619,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464849152421258e-05,
+      "loss": 0.9377,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463041959503223e-05,
+      "loss": 1.064,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746123421625418e-05,
+      "loss": 0.8098,
+      "step": 3041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459425922807433e-05,
+      "loss": 0.8337,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457617079296328e-05,
+      "loss": 1.0461,
+      "step": 3043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455807685854252e-05,
+      "loss": 0.9413,
+      "step": 3044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453997742614626e-05,
+      "loss": 0.9365,
+      "step": 3045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745218724971092e-05,
+      "loss": 0.9901,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745037620727664e-05,
+      "loss": 0.9387,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448564615445336e-05,
+      "loss": 0.9242,
+      "step": 3048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446752474350593e-05,
+      "loss": 0.8818,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444939784126043e-05,
+      "loss": 0.9988,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443126544905355e-05,
+      "loss": 0.9797,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744131275682224e-05,
+      "loss": 0.8603,
+      "step": 3052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439498420010444e-05,
+      "loss": 0.9583,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437683534603758e-05,
+      "loss": 0.3011,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435868100736015e-05,
+      "loss": 0.9371,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743405211854109e-05,
+      "loss": 0.9531,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432235588152887e-05,
+      "loss": 0.8637,
+      "step": 3057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743041850970537e-05,
+      "loss": 0.8918,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742860088333252e-05,
+      "loss": 0.9561,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742678270916838e-05,
+      "loss": 0.7654,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424963987347013e-05,
+      "loss": 0.9847,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423144718002544e-05,
+      "loss": 0.921,
+      "step": 3062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421324901269122e-05,
+      "loss": 0.932,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741950453728094e-05,
+      "loss": 1.0184,
+      "step": 3064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417683626172238e-05,
+      "loss": 1.0131,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741586216807729e-05,
+      "loss": 0.922,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741404016313041e-05,
+      "loss": 0.8976,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741221761146595e-05,
+      "loss": 0.9584,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741039451321832e-05,
+      "loss": 0.8789,
+      "step": 3069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408570868521945e-05,
+      "loss": 0.879,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406746677511307e-05,
+      "loss": 0.9507,
+      "step": 3071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404921940320922e-05,
+      "loss": 0.9264,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403096657085347e-05,
+      "loss": 0.8719,
+      "step": 3073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401270827939178e-05,
+      "loss": 0.8927,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399444453017058e-05,
+      "loss": 0.9098,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397617532453663e-05,
+      "loss": 0.962,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395790066383713e-05,
+      "loss": 0.8435,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739396205494196e-05,
+      "loss": 0.9801,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392133498263217e-05,
+      "loss": 0.8434,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390304396482315e-05,
+      "loss": 0.896,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388474749734127e-05,
+      "loss": 0.9108,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386644558153585e-05,
+      "loss": 0.9347,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738481382187564e-05,
+      "loss": 0.8932,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73829825410353e-05,
+      "loss": 0.8486,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381150715767593e-05,
+      "loss": 0.8764,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737931834620761e-05,
+      "loss": 0.8879,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737748543249047e-05,
+      "loss": 0.8549,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375651974751333e-05,
+      "loss": 0.8273,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73738179731254e-05,
+      "loss": 0.8335,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371983427747904e-05,
+      "loss": 0.9309,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370148338754137e-05,
+      "loss": 0.9486,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736831270627941e-05,
+      "loss": 0.9728,
+      "step": 3092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366476530459094e-05,
+      "loss": 0.9099,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364639811428588e-05,
+      "loss": 0.9762,
+      "step": 3094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362802549323324e-05,
+      "loss": 0.9129,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360964744278798e-05,
+      "loss": 0.9586,
+      "step": 3096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735912639643052e-05,
+      "loss": 1.0793,
+      "step": 3097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357287505914055e-05,
+      "loss": 0.8862,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355448072865e-05,
+      "loss": 0.8611,
+      "step": 3099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353608097419005e-05,
+      "loss": 0.9294,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351767579711743e-05,
+      "loss": 0.8566,
+      "step": 3101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349926519878946e-05,
+      "loss": 0.8325,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348084918056362e-05,
+      "loss": 0.8867,
+      "step": 3103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73462427743798e-05,
+      "loss": 1.0251,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344400088985103e-05,
+      "loss": 0.971,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342556862008143e-05,
+      "loss": 0.8873,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340713093584852e-05,
+      "loss": 0.8915,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338868783851184e-05,
+      "loss": 1.0692,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337023932943142e-05,
+      "loss": 0.8827,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335178540996768e-05,
+      "loss": 0.9313,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333332608148145e-05,
+      "loss": 0.9727,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331486134533386e-05,
+      "loss": 0.9601,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329639120288657e-05,
+      "loss": 0.9749,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732779156555016e-05,
+      "loss": 0.9543,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325943470454127e-05,
+      "loss": 0.9204,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732409483513685e-05,
+      "loss": 0.8741,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322245659734638e-05,
+      "loss": 0.8945,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320395944383857e-05,
+      "loss": 1.0061,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318545689220904e-05,
+      "loss": 0.9025,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731669489438222e-05,
+      "loss": 0.9279,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314843560004283e-05,
+      "loss": 1.0223,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312991686223614e-05,
+      "loss": 0.9352,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311139273176766e-05,
+      "loss": 0.9775,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309286321000346e-05,
+      "loss": 1.0048,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730743282983099e-05,
+      "loss": 0.8426,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305578799805367e-05,
+      "loss": 0.9093,
+      "step": 3126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303724231060207e-05,
+      "loss": 0.9373,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730186912373226e-05,
+      "loss": 0.9127,
+      "step": 3128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300013477958323e-05,
+      "loss": 0.9704,
+      "step": 3129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298157293875237e-05,
+      "loss": 0.9368,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729630057161988e-05,
+      "loss": 0.2563,
+      "step": 3131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294443311329165e-05,
+      "loss": 0.9481,
+      "step": 3132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292585513140044e-05,
+      "loss": 0.8894,
+      "step": 3133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729072717718952e-05,
+      "loss": 0.8475,
+      "step": 3134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288868303614622e-05,
+      "loss": 0.8846,
+      "step": 3135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287008892552428e-05,
+      "loss": 0.9209,
+      "step": 3136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285148944140053e-05,
+      "loss": 0.9786,
+      "step": 3137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728328845851465e-05,
+      "loss": 0.8979,
+      "step": 3138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728142743581341e-05,
+      "loss": 0.2416,
+      "step": 3139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279565876173575e-05,
+      "loss": 0.8957,
+      "step": 3140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277703779732412e-05,
+      "loss": 0.9894,
+      "step": 3141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727584114662723e-05,
+      "loss": 1.0101,
+      "step": 3142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273977976995392e-05,
+      "loss": 1.0265,
+      "step": 3143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272114270974278e-05,
+      "loss": 0.2352,
+      "step": 3144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270250028701325e-05,
+      "loss": 1.0056,
+      "step": 3145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268385250314005e-05,
+      "loss": 1.0322,
+      "step": 3146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266519935949824e-05,
+      "loss": 0.9265,
+      "step": 3147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264654085746336e-05,
+      "loss": 1.0164,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726278769984113e-05,
+      "loss": 0.8581,
+      "step": 3149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726092077837183e-05,
+      "loss": 0.8903,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725905332147611e-05,
+      "loss": 0.7858,
+      "step": 3151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257185329291672e-05,
+      "loss": 1.0144,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725531680195627e-05,
+      "loss": 0.3153,
+      "step": 3153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253447739607687e-05,
+      "loss": 0.7781,
+      "step": 3154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725157814238375e-05,
+      "loss": 0.8756,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249708010422324e-05,
+      "loss": 0.933,
+      "step": 3156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724783734386132e-05,
+      "loss": 0.9565,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724596614283867e-05,
+      "loss": 0.8814,
+      "step": 3158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724409440749237e-05,
+      "loss": 0.9751,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724222213796044e-05,
+      "loss": 0.9729,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240349334380936e-05,
+      "loss": 1.0026,
+      "step": 3161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238475996891967e-05,
+      "loss": 0.9255,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236602125631675e-05,
+      "loss": 0.9843,
+      "step": 3163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234727720738238e-05,
+      "loss": 0.9128,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232852782349877e-05,
+      "loss": 0.944,
+      "step": 3165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723097731060485e-05,
+      "loss": 0.9561,
+      "step": 3166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722910130564146e-05,
+      "loss": 0.9841,
+      "step": 3167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227224767598037e-05,
+      "loss": 0.9126,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225347696612972e-05,
+      "loss": 0.2391,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722347009282467e-05,
+      "loss": 0.9784,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221591956371592e-05,
+      "loss": 0.9679,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721971328739223e-05,
+      "loss": 0.9225,
+      "step": 3172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217834086025123e-05,
+      "loss": 0.9256,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721595435240884e-05,
+      "loss": 0.8962,
+      "step": 3174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214074086682e-05,
+      "loss": 0.8598,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212193288983252e-05,
+      "loss": 0.8931,
+      "step": 3176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210311959451286e-05,
+      "loss": 0.8026,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720843009822484e-05,
+      "loss": 0.9476,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206547705442673e-05,
+      "loss": 0.8897,
+      "step": 3179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204664781243603e-05,
+      "loss": 0.2753,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202781325766475e-05,
+      "loss": 0.9127,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200897339150174e-05,
+      "loss": 1.0189,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199012821533636e-05,
+      "loss": 0.8588,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197127773055816e-05,
+      "loss": 1.0496,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195242193855727e-05,
+      "loss": 0.2992,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193356084072408e-05,
+      "loss": 0.8699,
+      "step": 3186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191469443844944e-05,
+      "loss": 0.951,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718958227331246e-05,
+      "loss": 0.9473,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718769457261411e-05,
+      "loss": 0.9888,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185806341889105e-05,
+      "loss": 0.9957,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183917581276678e-05,
+      "loss": 0.9324,
+      "step": 3191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182028290916114e-05,
+      "loss": 1.0419,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180138470946723e-05,
+      "loss": 0.9088,
+      "step": 3193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178248121507865e-05,
+      "loss": 0.964,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176357242738934e-05,
+      "loss": 0.8939,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717446583477937e-05,
+      "loss": 1.0195,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7172573897768644e-05,
+      "loss": 0.9895,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717068143184627e-05,
+      "loss": 0.9593,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168788437151798e-05,
+      "loss": 0.2349,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166894913824822e-05,
+      "loss": 0.9069,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716500086200497e-05,
+      "loss": 0.7879,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716310628183191e-05,
+      "loss": 0.8748,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716121117344535e-05,
+      "loss": 0.9121,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715931553698504e-05,
+      "loss": 0.7954,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157419372590764e-05,
+      "loss": 0.9663,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155522680402348e-05,
+      "loss": 0.9768,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153625460559653e-05,
+      "loss": 0.9399,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151727713202583e-05,
+      "loss": 0.9732,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714982943847108e-05,
+      "loss": 0.9175,
+      "step": 3209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147930636505125e-05,
+      "loss": 0.9775,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146031307444733e-05,
+      "loss": 0.939,
+      "step": 3211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714413145142997e-05,
+      "loss": 1.0237,
+      "step": 3212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142231068600922e-05,
+      "loss": 0.9439,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140330159097736e-05,
+      "loss": 0.9174,
+      "step": 3214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713842872306058e-05,
+      "loss": 1.0041,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713652676062967e-05,
+      "loss": 0.9093,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134624271945256e-05,
+      "loss": 0.9267,
+      "step": 3217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132721257147636e-05,
+      "loss": 0.9075,
+      "step": 3218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130817716377128e-05,
+      "loss": 0.9579,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712891364977411e-05,
+      "loss": 0.791,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712700905747899e-05,
+      "loss": 0.8998,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125103939632206e-05,
+      "loss": 1.0181,
+      "step": 3222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712319829637425e-05,
+      "loss": 0.9169,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121292127845644e-05,
+      "loss": 0.9842,
+      "step": 3224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711938543418695e-05,
+      "loss": 0.9568,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117478215538768e-05,
+      "loss": 0.9451,
+      "step": 3226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711557047204174e-05,
+      "loss": 0.8737,
+      "step": 3227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711366220383655e-05,
+      "loss": 0.9223,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111753411063903e-05,
+      "loss": 1.013,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109844093864563e-05,
+      "loss": 0.941,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710793425237932e-05,
+      "loss": 0.9263,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106023886749016e-05,
+      "loss": 0.9856,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710411299711451e-05,
+      "loss": 0.9074,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102201583616725e-05,
+      "loss": 0.9377,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100289646396605e-05,
+      "loss": 0.992,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098377185595137e-05,
+      "loss": 0.9862,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096464201353348e-05,
+      "loss": 1.038,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094550693812304e-05,
+      "loss": 0.8179,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709263666311311e-05,
+      "loss": 1.0282,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090722109396903e-05,
+      "loss": 0.8855,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088807032804872e-05,
+      "loss": 0.9578,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086891433478228e-05,
+      "loss": 0.8437,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084975311558232e-05,
+      "loss": 0.902,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708305866718618e-05,
+      "loss": 0.9675,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081141500503414e-05,
+      "loss": 0.9868,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079223811651294e-05,
+      "loss": 0.9465,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077305600771247e-05,
+      "loss": 0.8901,
+      "step": 3247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075386868004708e-05,
+      "loss": 0.8857,
+      "step": 3248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707346761349318e-05,
+      "loss": 0.9076,
+      "step": 3249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707154783737818e-05,
+      "loss": 0.9722,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706962753980128e-05,
+      "loss": 0.9507,
+      "step": 3251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706770672090408e-05,
+      "loss": 0.9297,
+      "step": 3252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706578538082823e-05,
+      "loss": 1.0042,
+      "step": 3253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70638635197154e-05,
+      "loss": 0.856,
+      "step": 3254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061941137707326e-05,
+      "loss": 0.9318,
+      "step": 3255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706001823494575e-05,
+      "loss": 0.2878,
+      "step": 3256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058094811572478e-05,
+      "loss": 0.9349,
+      "step": 3257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056170867729336e-05,
+      "loss": 0.9475,
+      "step": 3258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054246403558208e-05,
+      "loss": 0.9567,
+      "step": 3259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052321419201e-05,
+      "loss": 0.934,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050395914799664e-05,
+      "loss": 0.9373,
+      "step": 3261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048469890496183e-05,
+      "loss": 0.7805,
+      "step": 3262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046543346432587e-05,
+      "loss": 0.8928,
+      "step": 3263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044616282750944e-05,
+      "loss": 0.9214,
+      "step": 3264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042688699593353e-05,
+      "loss": 0.9677,
+      "step": 3265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704076059710196e-05,
+      "loss": 0.8996,
+      "step": 3266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703883197541894e-05,
+      "loss": 0.8686,
+      "step": 3267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036902834686515e-05,
+      "loss": 0.9096,
+      "step": 3268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703497317504694e-05,
+      "loss": 0.8583,
+      "step": 3269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033042996642506e-05,
+      "loss": 0.8308,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031112299615557e-05,
+      "loss": 0.9635,
+      "step": 3271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702918108410845e-05,
+      "loss": 1.0165,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027249350263603e-05,
+      "loss": 0.9513,
+      "step": 3273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025317098223463e-05,
+      "loss": 0.9361,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023384328130514e-05,
+      "loss": 0.8803,
+      "step": 3275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702145104012728e-05,
+      "loss": 0.8873,
+      "step": 3276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701951723435633e-05,
+      "loss": 1.0379,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017582910960254e-05,
+      "loss": 0.929,
+      "step": 3278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015648070081695e-05,
+      "loss": 0.9333,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701371271186333e-05,
+      "loss": 1.0151,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011776836447874e-05,
+      "loss": 0.8932,
+      "step": 3281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009840443978086e-05,
+      "loss": 0.8918,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007903534596745e-05,
+      "loss": 0.8629,
+      "step": 3283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700596610844669e-05,
+      "loss": 0.8743,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004028165670786e-05,
+      "loss": 0.8652,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002089706411933e-05,
+      "loss": 0.8568,
+      "step": 3286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700015073081308e-05,
+      "loss": 0.8547,
+      "step": 3287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998211239017213e-05,
+      "loss": 0.8976,
+      "step": 3288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996271231167345e-05,
+      "loss": 0.884,
+      "step": 3289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994330707406535e-05,
+      "loss": 0.291,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699238966787788e-05,
+      "loss": 0.935,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990448112724514e-05,
+      "loss": 0.9648,
+      "step": 3292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988506042089604e-05,
+      "loss": 1.0532,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698656345611637e-05,
+      "loss": 0.8876,
+      "step": 3294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698462035494805e-05,
+      "loss": 0.9649,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982676738727933e-05,
+      "loss": 0.9474,
+      "step": 3296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698073260759935e-05,
+      "loss": 0.9461,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978787961705652e-05,
+      "loss": 0.9312,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976842801190245e-05,
+      "loss": 0.9827,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974897126196567e-05,
+      "loss": 0.9026,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972950936868087e-05,
+      "loss": 0.9057,
+      "step": 3301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697100423334833e-05,
+      "loss": 0.2382,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696905701578084e-05,
+      "loss": 0.9733,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967109284309204e-05,
+      "loss": 0.9374,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965161039077057e-05,
+      "loss": 0.267,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963212280228057e-05,
+      "loss": 0.9564,
+      "step": 3306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961263007905916e-05,
+      "loss": 0.9442,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959313222254365e-05,
+      "loss": 0.963,
+      "step": 3308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695736292341719e-05,
+      "loss": 0.9095,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955412111538203e-05,
+      "loss": 0.9622,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695346078676126e-05,
+      "loss": 0.9661,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951508949230254e-05,
+      "loss": 0.9005,
+      "step": 3312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694955659908912e-05,
+      "loss": 0.9998,
+      "step": 3313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694760373648182e-05,
+      "loss": 0.9168,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945650361552357e-05,
+      "loss": 0.9424,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943696474444778e-05,
+      "loss": 1.0186,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941742075303166e-05,
+      "loss": 0.2646,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693978716427164e-05,
+      "loss": 0.9618,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937831741494354e-05,
+      "loss": 0.9094,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935875807115506e-05,
+      "loss": 0.8747,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693391936127932e-05,
+      "loss": 1.0334,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931962404130077e-05,
+      "loss": 0.8678,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693000493581208e-05,
+      "loss": 0.2681,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928046956469676e-05,
+      "loss": 0.8701,
+      "step": 3324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692608846624724e-05,
+      "loss": 0.9197,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924129465289203e-05,
+      "loss": 1.0067,
+      "step": 3326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922169953740017e-05,
+      "loss": 0.8758,
+      "step": 3327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920209931744183e-05,
+      "loss": 0.9022,
+      "step": 3328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691824939944623e-05,
+      "loss": 0.941,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916288356990733e-05,
+      "loss": 0.905,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69143268045223e-05,
+      "loss": 0.9056,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912364742185575e-05,
+      "loss": 0.7964,
+      "step": 3332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910402170125245e-05,
+      "loss": 0.9301,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908439088486028e-05,
+      "loss": 0.9774,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906475497412688e-05,
+      "loss": 0.8705,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690451139705002e-05,
+      "loss": 0.9872,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902546787542857e-05,
+      "loss": 0.9631,
+      "step": 3337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900581669036075e-05,
+      "loss": 0.8486,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898616041674578e-05,
+      "loss": 0.9686,
+      "step": 3339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896649905603318e-05,
+      "loss": 0.9213,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894683260967273e-05,
+      "loss": 0.9989,
+      "step": 3341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892716107911475e-05,
+      "loss": 0.9437,
+      "step": 3342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890748446580973e-05,
+      "loss": 1.0096,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888780277120868e-05,
+      "loss": 0.9426,
+      "step": 3344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886811599676297e-05,
+      "loss": 0.8918,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884842414392434e-05,
+      "loss": 0.9112,
+      "step": 3346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688287272141448e-05,
+      "loss": 0.9339,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880902520887687e-05,
+      "loss": 0.9902,
+      "step": 3348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687893181295734e-05,
+      "loss": 0.9829,
+      "step": 3349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876960597768757e-05,
+      "loss": 0.8686,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874988875467302e-05,
+      "loss": 0.9723,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687301664619837e-05,
+      "loss": 0.8984,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687104391010739e-05,
+      "loss": 0.9613,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686907066733984e-05,
+      "loss": 0.9431,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867096918041222e-05,
+      "loss": 1.0161,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865122662357087e-05,
+      "loss": 0.841,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863147900433018e-05,
+      "loss": 0.892,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861172632414633e-05,
+      "loss": 0.9214,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859196858447596e-05,
+      "loss": 0.2606,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685722057867759e-05,
+      "loss": 0.9057,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685524379325036e-05,
+      "loss": 0.9824,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853266502311667e-05,
+      "loss": 0.9023,
+      "step": 3362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851288706007325e-05,
+      "loss": 0.8363,
+      "step": 3363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849310404483175e-05,
+      "loss": 0.902,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68473315978851e-05,
+      "loss": 0.957,
+      "step": 3365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845352286359015e-05,
+      "loss": 0.9124,
+      "step": 3366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684337247005088e-05,
+      "loss": 1.0511,
+      "step": 3367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684139214910669e-05,
+      "loss": 1.0488,
+      "step": 3368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683941132367247e-05,
+      "loss": 0.2662,
+      "step": 3369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837429993894286e-05,
+      "loss": 0.9956,
+      "step": 3370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835448159918255e-05,
+      "loss": 0.9777,
+      "step": 3371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683346582189051e-05,
+      "loss": 1.0285,
+      "step": 3372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683148297995723e-05,
+      "loss": 0.9569,
+      "step": 3373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829499634264633e-05,
+      "loss": 1.0029,
+      "step": 3374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827515784958973e-05,
+      "loss": 0.8605,
+      "step": 3375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825531432186545e-05,
+      "loss": 0.9453,
+      "step": 3376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682354657609367e-05,
+      "loss": 0.876,
+      "step": 3377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682156121682671e-05,
+      "loss": 0.9255,
+      "step": 3378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819575354532076e-05,
+      "loss": 0.8584,
+      "step": 3379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817588989356207e-05,
+      "loss": 1.0178,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815602121445572e-05,
+      "loss": 0.9343,
+      "step": 3381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681361475094669e-05,
+      "loss": 0.9309,
+      "step": 3382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811626878006108e-05,
+      "loss": 0.8655,
+      "step": 3383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809638502770416e-05,
+      "loss": 0.8939,
+      "step": 3384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807649625386237e-05,
+      "loss": 0.9366,
+      "step": 3385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805660246000233e-05,
+      "loss": 0.8399,
+      "step": 3386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803670364759104e-05,
+      "loss": 0.943,
+      "step": 3387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680167998180958e-05,
+      "loss": 0.8424,
+      "step": 3388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799689097298442e-05,
+      "loss": 0.9001,
+      "step": 3389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679769771137249e-05,
+      "loss": 0.8067,
+      "step": 3390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679570582417858e-05,
+      "loss": 0.9385,
+      "step": 3391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793713435863592e-05,
+      "loss": 1.0017,
+      "step": 3392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679172054657444e-05,
+      "loss": 0.9199,
+      "step": 3393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678972715645809e-05,
+      "loss": 0.9198,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787733265661535e-05,
+      "loss": 0.9926,
+      "step": 3395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67857388743318e-05,
+      "loss": 0.863,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783743982615957e-05,
+      "loss": 1.0075,
+      "step": 3397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781748590661116e-05,
+      "loss": 0.9851,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6779752698614408e-05,
+      "loss": 0.8145,
+      "step": 3399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677775630662302e-05,
+      "loss": 0.9096,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775759414834166e-05,
+      "loss": 0.8937,
+      "step": 3401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67737620233951e-05,
+      "loss": 0.9666,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677176413245311e-05,
+      "loss": 0.9379,
+      "step": 3403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769765742155517e-05,
+      "loss": 0.9271,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676776685264969e-05,
+      "loss": 0.9086,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765767464083027e-05,
+      "loss": 0.972,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763767576602968e-05,
+      "loss": 0.9899,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761767190356983e-05,
+      "loss": 0.3016,
+      "step": 3408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675976630549258e-05,
+      "loss": 0.9184,
+      "step": 3409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757764922157305e-05,
+      "loss": 1.0133,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755763040498748e-05,
+      "loss": 0.9854,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675376066066453e-05,
+      "loss": 0.8982,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67517577828023e-05,
+      "loss": 0.9501,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749754407059756e-05,
+      "loss": 0.8783,
+      "step": 3414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747750533584633e-05,
+      "loss": 0.8321,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745746162524695e-05,
+      "loss": 1.0179,
+      "step": 3416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674374129402774e-05,
+      "loss": 0.9634,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674173592824162e-05,
+      "loss": 0.9274,
+      "step": 3418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739730065314203e-05,
+      "loss": 0.9631,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737723705393406e-05,
+      "loss": 0.8991,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735716848627185e-05,
+      "loss": 0.9529,
+      "step": 3421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733709495163518e-05,
+      "loss": 0.9229,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731701645150434e-05,
+      "loss": 1.0149,
+      "step": 3423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729693298735996e-05,
+      "loss": 0.962,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727684456068297e-05,
+      "loss": 0.8329,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725675117295472e-05,
+      "loss": 0.9767,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672366528256569e-05,
+      "loss": 0.9953,
+      "step": 3427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672165495202716e-05,
+      "loss": 0.9175,
+      "step": 3428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719644125828125e-05,
+      "loss": 0.9987,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717632804116862e-05,
+      "loss": 0.8985,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715620987041697e-05,
+      "loss": 0.9088,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671360867475097e-05,
+      "loss": 0.9452,
+      "step": 3432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671159586739308e-05,
+      "loss": 0.8798,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709582565116446e-05,
+      "loss": 0.8762,
+      "step": 3434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670756876806954e-05,
+      "loss": 0.9666,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705554476400853e-05,
+      "loss": 0.9933,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703539690258923e-05,
+      "loss": 0.8564,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701524409792322e-05,
+      "loss": 0.8999,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699508635149658e-05,
+      "loss": 0.8831,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669749236647958e-05,
+      "loss": 0.9933,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695475603930766e-05,
+      "loss": 0.8502,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693458347651934e-05,
+      "loss": 0.9439,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691440597791836e-05,
+      "loss": 0.9321,
+      "step": 3443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689422354499263e-05,
+      "loss": 0.9153,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687403617923048e-05,
+      "loss": 0.9069,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685384388212046e-05,
+      "loss": 0.2489,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683364665515164e-05,
+      "loss": 1.0373,
+      "step": 3447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681344449981333e-05,
+      "loss": 0.9054,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679323741759526e-05,
+      "loss": 0.9058,
+      "step": 3449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677302540998757e-05,
+      "loss": 0.8606,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675280847848062e-05,
+      "loss": 0.9564,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667325866245653e-05,
+      "loss": 0.7916,
+      "step": 3452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667123598497328e-05,
+      "loss": 0.9286,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669212815547454e-05,
+      "loss": 0.9702,
+      "step": 3454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667189154328253e-05,
+      "loss": 0.9363,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66651650014649e-05,
+      "loss": 0.8958,
+      "step": 3456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666314035710666e-05,
+      "loss": 1.0059,
+      "step": 3457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666111522140283e-05,
+      "loss": 1.01,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659089594502743e-05,
+      "loss": 0.9751,
+      "step": 3459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657063476555775e-05,
+      "loss": 0.8554,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655036867711333e-05,
+      "loss": 1.0056,
+      "step": 3461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653009768118858e-05,
+      "loss": 0.9375,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665098217792783e-05,
+      "loss": 0.9004,
+      "step": 3463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664895409728777e-05,
+      "loss": 0.9072,
+      "step": 3464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664692552634822e-05,
+      "loss": 0.9554,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644896465258782e-05,
+      "loss": 0.9828,
+      "step": 3466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642866914169073e-05,
+      "loss": 0.8167,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640836873228752e-05,
+      "loss": 0.8413,
+      "step": 3468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638806342587524e-05,
+      "loss": 0.2412,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636775322395117e-05,
+      "loss": 0.9742,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634743812801296e-05,
+      "loss": 1.0526,
+      "step": 3471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663271181395587e-05,
+      "loss": 0.9116,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630679326008686e-05,
+      "loss": 0.9246,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662864634910961e-05,
+      "loss": 0.8441,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626612883408568e-05,
+      "loss": 0.882,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624578929055498e-05,
+      "loss": 0.9127,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662254448620039e-05,
+      "loss": 0.8524,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620509554993272e-05,
+      "loss": 0.9726,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661847413558419e-05,
+      "loss": 0.8547,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616438228123245e-05,
+      "loss": 0.9615,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661440183276056e-05,
+      "loss": 0.9558,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612364949646315e-05,
+      "loss": 0.9169,
+      "step": 3482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610327578930695e-05,
+      "loss": 0.8989,
+      "step": 3483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608289720763945e-05,
+      "loss": 0.8856,
+      "step": 3484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606251375296335e-05,
+      "loss": 0.9785,
+      "step": 3485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660421254267818e-05,
+      "loss": 0.9834,
+      "step": 3486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602173223059823e-05,
+      "loss": 0.9289,
+      "step": 3487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600133416591645e-05,
+      "loss": 0.9028,
+      "step": 3488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598093123424058e-05,
+      "loss": 0.8629,
+      "step": 3489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659605234370752e-05,
+      "loss": 0.859,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659401107759252e-05,
+      "loss": 0.9505,
+      "step": 3491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659196932522958e-05,
+      "loss": 1.0297,
+      "step": 3492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589927086769265e-05,
+      "loss": 0.9434,
+      "step": 3493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658788436236217e-05,
+      "loss": 1.0251,
+      "step": 3494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658584115215892e-05,
+      "loss": 0.9325,
+      "step": 3495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583797456310195e-05,
+      "loss": 1.086,
+      "step": 3496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581753274966684e-05,
+      "loss": 0.9438,
+      "step": 3497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579708608279145e-05,
+      "loss": 0.9202,
+      "step": 3498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577663456398338e-05,
+      "loss": 0.8877,
+      "step": 3499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575617819475084e-05,
+      "loss": 0.2297,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573571697660223e-05,
+      "loss": 0.9323,
+      "step": 3501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571525091104642e-05,
+      "loss": 0.9445,
+      "step": 3502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569477999959255e-05,
+      "loss": 0.8609,
+      "step": 3503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567430424375025e-05,
+      "loss": 0.9543,
+      "step": 3504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565382364502932e-05,
+      "loss": 0.8786,
+      "step": 3505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563333820494007e-05,
+      "loss": 0.9102,
+      "step": 3506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561284792499312e-05,
+      "loss": 1.0258,
+      "step": 3507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559235280669942e-05,
+      "loss": 1.0423,
+      "step": 3508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655718528515703e-05,
+      "loss": 0.8653,
+      "step": 3509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555134806111745e-05,
+      "loss": 0.979,
+      "step": 3510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655308384368529e-05,
+      "loss": 0.8576,
+      "step": 3511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551032398028905e-05,
+      "loss": 0.9066,
+      "step": 3512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548980469293867e-05,
+      "loss": 0.9127,
+      "step": 3513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546928057631487e-05,
+      "loss": 0.8581,
+      "step": 3514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544875163193104e-05,
+      "loss": 0.8073,
+      "step": 3515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542821786130114e-05,
+      "loss": 0.8893,
+      "step": 3516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540767926593924e-05,
+      "loss": 0.7951,
+      "step": 3517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538713584735992e-05,
+      "loss": 0.89,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536658760707804e-05,
+      "loss": 0.9117,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534603454660883e-05,
+      "loss": 0.9604,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532547666746796e-05,
+      "loss": 0.8339,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530491397117132e-05,
+      "loss": 0.8844,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528434645923526e-05,
+      "loss": 0.9758,
+      "step": 3523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526377413317643e-05,
+      "loss": 0.8901,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524319699451184e-05,
+      "loss": 0.9676,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652226150447589e-05,
+      "loss": 1.0202,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652020282854353e-05,
+      "loss": 0.9147,
+      "step": 3527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518143671805915e-05,
+      "loss": 0.9568,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516084034414883e-05,
+      "loss": 0.9423,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514023916522325e-05,
+      "loss": 0.9283,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511963318280147e-05,
+      "loss": 0.9093,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509902239840298e-05,
+      "loss": 0.8813,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650784068135477e-05,
+      "loss": 0.9351,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505778642975585e-05,
+      "loss": 0.8566,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503716124854794e-05,
+      "loss": 0.8524,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650165312714449e-05,
+      "loss": 0.9332,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64995896499968e-05,
+      "loss": 0.9879,
+      "step": 3537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497525693563887e-05,
+      "loss": 1.0349,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649546125799795e-05,
+      "loss": 0.9217,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493396343451222e-05,
+      "loss": 0.9728,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649133095007597e-05,
+      "loss": 0.9285,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64892650780245e-05,
+      "loss": 0.9458,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648719872744915e-05,
+      "loss": 0.8876,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485131898502295e-05,
+      "loss": 0.9012,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483064591336343e-05,
+      "loss": 0.9491,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648099680610374e-05,
+      "loss": 0.9151,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478928542956965e-05,
+      "loss": 0.9178,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476859802048534e-05,
+      "loss": 0.9193,
+      "step": 3548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474790583531002e-05,
+      "loss": 0.9918,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647272088755695e-05,
+      "loss": 0.3014,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470650714279003e-05,
+      "loss": 0.2222,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646858006384981e-05,
+      "loss": 0.854,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466508936422073e-05,
+      "loss": 1.0364,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646443733214851e-05,
+      "loss": 0.9684,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462365251181885e-05,
+      "loss": 0.968,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460292693674997e-05,
+      "loss": 0.9449,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458219659780678e-05,
+      "loss": 0.8772,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645614614965179e-05,
+      "loss": 0.9668,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454072163441244e-05,
+      "loss": 0.7951,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451997701301975e-05,
+      "loss": 1.0085,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644992276338695e-05,
+      "loss": 0.9857,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644784734984918e-05,
+      "loss": 0.2641,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644577146084171e-05,
+      "loss": 0.9051,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443695096517615e-05,
+      "loss": 0.8764,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644161825703001e-05,
+      "loss": 0.9431,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439540942532043e-05,
+      "loss": 0.9708,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643746315317689e-05,
+      "loss": 0.9713,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643538488911778e-05,
+      "loss": 0.8608,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643330615050796e-05,
+      "loss": 0.9193,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643122693750072e-05,
+      "loss": 0.9919,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642914725024938e-05,
+      "loss": 0.8559,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427067088907292e-05,
+      "loss": 0.8792,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424986453627866e-05,
+      "loss": 0.7516,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422905344564518e-05,
+      "loss": 0.981,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420823761870706e-05,
+      "loss": 0.2448,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418741705699945e-05,
+      "loss": 1.0602,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641665917620575e-05,
+      "loss": 0.9073,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414576173541697e-05,
+      "loss": 0.9117,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412492697861386e-05,
+      "loss": 1.0089,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410408749318457e-05,
+      "loss": 0.89,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640832432806658e-05,
+      "loss": 1.0221,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640623943425946e-05,
+      "loss": 0.9637,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404154068050843e-05,
+      "loss": 0.8147,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64020682295945e-05,
+      "loss": 0.8783,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399981919044246e-05,
+      "loss": 0.866,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397895136553933e-05,
+      "loss": 0.9233,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395807882277438e-05,
+      "loss": 1.0093,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639372015636867e-05,
+      "loss": 0.8441,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391631958981588e-05,
+      "loss": 1.0071,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638954329027017e-05,
+      "loss": 0.2619,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387454150388443e-05,
+      "loss": 1.0167,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385364539490462e-05,
+      "loss": 0.8801,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383274457730315e-05,
+      "loss": 1.0188,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638118390526212e-05,
+      "loss": 0.9725,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379092882240046e-05,
+      "loss": 1.0421,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377001388818283e-05,
+      "loss": 0.9993,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374909425151057e-05,
+      "loss": 0.8687,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372816991392634e-05,
+      "loss": 0.9512,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637072408769731e-05,
+      "loss": 0.942,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636863071421942e-05,
+      "loss": 0.9227,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366536871113333e-05,
+      "loss": 0.9747,
+      "step": 3601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364442558533444e-05,
+      "loss": 0.2868,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362347776634192e-05,
+      "loss": 0.9272,
+      "step": 3603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636025252557005e-05,
+      "loss": 0.956,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358156805495524e-05,
+      "loss": 0.9153,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635606061656515e-05,
+      "loss": 0.9597,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353963958933506e-05,
+      "loss": 0.8823,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351866832755203e-05,
+      "loss": 0.9691,
+      "step": 3608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634976923818488e-05,
+      "loss": 0.9106,
+      "step": 3609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347671175377216e-05,
+      "loss": 0.8875,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634557264448693e-05,
+      "loss": 0.9373,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343473645668763e-05,
+      "loss": 0.9144,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63413741790775e-05,
+      "loss": 0.9365,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339274244867955e-05,
+      "loss": 0.9161,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337173843194983e-05,
+      "loss": 0.9203,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335072974213466e-05,
+      "loss": 0.9552,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332971638078327e-05,
+      "loss": 0.857,
+      "step": 3617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330869834944515e-05,
+      "loss": 0.9159,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632876756496702e-05,
+      "loss": 0.8596,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632666482830087e-05,
+      "loss": 0.9662,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324561625101116e-05,
+      "loss": 0.9013,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632245795552286e-05,
+      "loss": 0.9103,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632035381972122e-05,
+      "loss": 1.0219,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318249217851355e-05,
+      "loss": 0.8185,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316144150068466e-05,
+      "loss": 1.0142,
+      "step": 3625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314038616527778e-05,
+      "loss": 0.9785,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631193261738456e-05,
+      "loss": 0.2666,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309826152794107e-05,
+      "loss": 0.8362,
+      "step": 3628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307719222911748e-05,
+      "loss": 0.9363,
+      "step": 3629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305611827892857e-05,
+      "loss": 1.0649,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630350396789283e-05,
+      "loss": 0.9278,
+      "step": 3631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301395643067108e-05,
+      "loss": 1.0063,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299286853571154e-05,
+      "loss": 0.9816,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297177599560472e-05,
+      "loss": 0.8792,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295067881190603e-05,
+      "loss": 0.8716,
+      "step": 3635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292957698617122e-05,
+      "loss": 0.9499,
+      "step": 3636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629084705199563e-05,
+      "loss": 0.8892,
+      "step": 3637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288735941481774e-05,
+      "loss": 0.8698,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286624367231226e-05,
+      "loss": 0.8721,
+      "step": 3639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284512329399688e-05,
+      "loss": 0.2679,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282399828142916e-05,
+      "loss": 0.8394,
+      "step": 3641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280286863616682e-05,
+      "loss": 0.8615,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278173435976795e-05,
+      "loss": 0.9443,
+      "step": 3643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276059545379107e-05,
+      "loss": 0.8633,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273945191979496e-05,
+      "loss": 0.2359,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271830375933867e-05,
+      "loss": 0.968,
+      "step": 3646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269715097398183e-05,
+      "loss": 0.9406,
+      "step": 3647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626759935652842e-05,
+      "loss": 0.9042,
+      "step": 3648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265483153480594e-05,
+      "loss": 0.8804,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263366488410752e-05,
+      "loss": 0.9992,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261249361474983e-05,
+      "loss": 0.8963,
+      "step": 3651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259131772829408e-05,
+      "loss": 0.9472,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257013722630176e-05,
+      "loss": 0.9346,
+      "step": 3653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254895211033477e-05,
+      "loss": 0.9673,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252776238195527e-05,
+      "loss": 0.872,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250656804272586e-05,
+      "loss": 0.9119,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248536909420936e-05,
+      "loss": 0.8973,
+      "step": 3657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246416553796905e-05,
+      "loss": 0.9021,
+      "step": 3658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244295737556855e-05,
+      "loss": 0.9821,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242174460857163e-05,
+      "loss": 0.8359,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240052723854268e-05,
+      "loss": 0.2327,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237930526704617e-05,
+      "loss": 1.0012,
+      "step": 3662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623580786956471e-05,
+      "loss": 0.8948,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623368475259107e-05,
+      "loss": 0.9329,
+      "step": 3664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623156117594026e-05,
+      "loss": 0.962,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229437139768875e-05,
+      "loss": 0.2791,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622731264423354e-05,
+      "loss": 0.9124,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225187689490914e-05,
+      "loss": 0.9246,
+      "step": 3668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223062275697705e-05,
+      "loss": 0.8155,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622093640301063e-05,
+      "loss": 0.8625,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621881007158646e-05,
+      "loss": 1.009,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621668328158199e-05,
+      "loss": 0.9113,
+      "step": 3672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621455603315405e-05,
+      "loss": 0.9389,
+      "step": 3673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621242832645951e-05,
+      "loss": 0.9644,
+      "step": 3674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210300161655262e-05,
+      "loss": 0.8752,
+      "step": 3675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208171538898244e-05,
+      "loss": 0.9534,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620604245834542e-05,
+      "loss": 0.9042,
+      "step": 3677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203912920153787e-05,
+      "loss": 0.9773,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620178292448039e-05,
+      "loss": 0.8833,
+      "step": 3679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199652471482285e-05,
+      "loss": 0.7689,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197521561316577e-05,
+      "loss": 0.8658,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195390194140406e-05,
+      "loss": 0.8867,
+      "step": 3682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193258370110933e-05,
+      "loss": 0.9088,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191126089385367e-05,
+      "loss": 0.8299,
+      "step": 3684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188993352120942e-05,
+      "loss": 0.9106,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186860158474924e-05,
+      "loss": 0.9593,
+      "step": 3686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618472650860462e-05,
+      "loss": 0.9083,
+      "step": 3687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618259240266737e-05,
+      "loss": 0.9182,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618045784082054e-05,
+      "loss": 0.2672,
+      "step": 3689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178322823221537e-05,
+      "loss": 0.9118,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176187350027797e-05,
+      "loss": 0.935,
+      "step": 3691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174051421396795e-05,
+      "loss": 0.2599,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617191503748603e-05,
+      "loss": 0.9021,
+      "step": 3693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616977819845305e-05,
+      "loss": 0.8478,
+      "step": 3694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616764090445542e-05,
+      "loss": 1.0403,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165503155650745e-05,
+      "loss": 0.904,
+      "step": 3696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616336495219667e-05,
+      "loss": 0.9876,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161226294250862e-05,
+      "loss": 0.8777,
+      "step": 3698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159087181971033e-05,
+      "loss": 0.9276,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615694761551492e-05,
+      "loss": 0.946,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154807595040295e-05,
+      "loss": 0.9388,
+      "step": 3701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615266712070497e-05,
+      "loss": 0.8928,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615052619266678e-05,
+      "loss": 0.9847,
+      "step": 3703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61483848110836e-05,
+      "loss": 0.9513,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146242976113345e-05,
+      "loss": 1.0428,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144100687913942e-05,
+      "loss": 0.9875,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141957946643376e-05,
+      "loss": 0.9638,
+      "step": 3707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613981475245965e-05,
+      "loss": 0.9121,
+      "step": 3708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137671105520807e-05,
+      "loss": 0.9522,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135527005984917e-05,
+      "loss": 0.8806,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133382454010095e-05,
+      "loss": 0.8574,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131237449754475e-05,
+      "loss": 0.8819,
+      "step": 3712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129091993376232e-05,
+      "loss": 0.895,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612694608503358e-05,
+      "loss": 0.8535,
+      "step": 3714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124799724884754e-05,
+      "loss": 0.2646,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122652913088032e-05,
+      "loss": 0.8655,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120505649801718e-05,
+      "loss": 0.9046,
+      "step": 3717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118357935184158e-05,
+      "loss": 0.9771,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611620976939372e-05,
+      "loss": 0.8962,
+      "step": 3719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611406115258882e-05,
+      "loss": 0.864,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611191208492789e-05,
+      "loss": 0.8618,
+      "step": 3721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610976256656941e-05,
+      "loss": 0.9137,
+      "step": 3722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107612597671883e-05,
+      "loss": 0.8738,
+      "step": 3723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105462178393853e-05,
+      "loss": 0.9266,
+      "step": 3724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103311308893894e-05,
+      "loss": 1.0378,
+      "step": 3725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101159989330612e-05,
+      "loss": 0.9682,
+      "step": 3726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099008219862643e-05,
+      "loss": 0.9,
+      "step": 3727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096856000648666e-05,
+      "loss": 1.0005,
+      "step": 3728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609470333184738e-05,
+      "loss": 0.953,
+      "step": 3729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609255021361754e-05,
+      "loss": 0.9515,
+      "step": 3730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090396646117904e-05,
+      "loss": 0.872,
+      "step": 3731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608824262950728e-05,
+      "loss": 0.9221,
+      "step": 3732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608608816394451e-05,
+      "loss": 0.9041,
+      "step": 3733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608393324958847e-05,
+      "loss": 0.8261,
+      "step": 3734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608177788659806e-05,
+      "loss": 0.9872,
+      "step": 3735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079622075132217e-05,
+      "loss": 0.8357,
+      "step": 3736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077465815349914e-05,
+      "loss": 0.9515,
+      "step": 3737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075309107410157e-05,
+      "loss": 0.9584,
+      "step": 3738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607315195147198e-05,
+      "loss": 0.9357,
+      "step": 3739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607099434769446e-05,
+      "loss": 0.8645,
+      "step": 3740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068836296236695e-05,
+      "loss": 1.0251,
+      "step": 3741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066677797257822e-05,
+      "loss": 0.8956,
+      "step": 3742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606451885091701e-05,
+      "loss": 0.31,
+      "step": 3743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062359457373467e-05,
+      "loss": 0.9836,
+      "step": 3744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060199616786424e-05,
+      "loss": 0.2801,
+      "step": 3745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605803932931515e-05,
+      "loss": 0.9146,
+      "step": 3746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055878595118945e-05,
+      "loss": 0.9596,
+      "step": 3747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053717414357142e-05,
+      "loss": 0.9539,
+      "step": 3748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051555787189115e-05,
+      "loss": 0.8728,
+      "step": 3749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049393713774258e-05,
+      "loss": 1.0429,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604723119427201e-05,
+      "loss": 0.7966,
+      "step": 3751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604506822884183e-05,
+      "loss": 0.9406,
+      "step": 3752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604290481764322e-05,
+      "loss": 0.8995,
+      "step": 3753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040740960835717e-05,
+      "loss": 0.9045,
+      "step": 3754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6038576658578874e-05,
+      "loss": 0.8669,
+      "step": 3755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036411911032298e-05,
+      "loss": 1.0345,
+      "step": 3756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603424671835562e-05,
+      "loss": 0.8117,
+      "step": 3757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032081080708494e-05,
+      "loss": 0.2732,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029914998250624e-05,
+      "loss": 0.8849,
+      "step": 3759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602774847114174e-05,
+      "loss": 0.9108,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025581499541593e-05,
+      "loss": 0.276,
+      "step": 3761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023414083609987e-05,
+      "loss": 0.967,
+      "step": 3762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021246223506747e-05,
+      "loss": 0.8847,
+      "step": 3763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601907791939173e-05,
+      "loss": 0.9929,
+      "step": 3764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601690917142483e-05,
+      "loss": 0.918,
+      "step": 3765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014739979765976e-05,
+      "loss": 0.8683,
+      "step": 3766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012570344575125e-05,
+      "loss": 0.8679,
+      "step": 3767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601040026601226e-05,
+      "loss": 0.995,
+      "step": 3768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008229744237415e-05,
+      "loss": 0.9188,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600605877941064e-05,
+      "loss": 0.9854,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003887371692022e-05,
+      "loss": 0.9435,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001715521241688e-05,
+      "loss": 0.8768,
+      "step": 3772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599954322821979e-05,
+      "loss": 0.963,
+      "step": 3773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599737049278651e-05,
+      "loss": 0.9921,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995197315102075e-05,
+      "loss": 0.2856,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599302369532673e-05,
+      "loss": 0.8646,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599084963362077e-05,
+      "loss": 0.9152,
+      "step": 3777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988675130144505e-05,
+      "loss": 0.8592,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986500185058278e-05,
+      "loss": 0.2781,
+      "step": 3779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984324798522484e-05,
+      "loss": 0.8373,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598214897069753e-05,
+      "loss": 0.8793,
+      "step": 3781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979972701743865e-05,
+      "loss": 0.2549,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597779599182197e-05,
+      "loss": 0.9155,
+      "step": 3783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975618841092355e-05,
+      "loss": 0.8797,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597344124971557e-05,
+      "loss": 0.914,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971263217852188e-05,
+      "loss": 0.9178,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969084745662818e-05,
+      "loss": 0.9127,
+      "step": 3787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966905833308105e-05,
+      "loss": 0.8423,
+      "step": 3788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964726480948724e-05,
+      "loss": 0.2846,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596254668874538e-05,
+      "loss": 0.9524,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596036645685882e-05,
+      "loss": 0.9972,
+      "step": 3791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958185785449805e-05,
+      "loss": 0.9711,
+      "step": 3792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595600467467915e-05,
+      "loss": 0.2875,
+      "step": 3793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953823124707677e-05,
+      "loss": 0.8939,
+      "step": 3794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951641135696274e-05,
+      "loss": 0.9393,
+      "step": 3795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949458707805832e-05,
+      "loss": 0.949,
+      "step": 3796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947275841197284e-05,
+      "loss": 0.8317,
+      "step": 3797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945092536031603e-05,
+      "loss": 0.9091,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942908792469783e-05,
+      "loss": 0.944,
+      "step": 3799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594072461067286e-05,
+      "loss": 0.9522,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938539990801886e-05,
+      "loss": 0.9709,
+      "step": 3801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593635493301797e-05,
+      "loss": 0.8799,
+      "step": 3802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934169437482233e-05,
+      "loss": 0.8647,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931983504355837e-05,
+      "loss": 0.8588,
+      "step": 3804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929797133799974e-05,
+      "loss": 0.9287,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592761032597587e-05,
+      "loss": 0.9576,
+      "step": 3806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925423081044778e-05,
+      "loss": 0.9814,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923235399167993e-05,
+      "loss": 0.9407,
+      "step": 3808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921047280506833e-05,
+      "loss": 0.9517,
+      "step": 3809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5918858725222653e-05,
+      "loss": 0.946,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591666973347684e-05,
+      "loss": 0.9905,
+      "step": 3811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914480305430808e-05,
+      "loss": 0.9397,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591229044124601e-05,
+      "loss": 0.9498,
+      "step": 3813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591010014108393e-05,
+      "loss": 0.9314,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907909405106082e-05,
+      "loss": 0.8331,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590571823347401e-05,
+      "loss": 0.9765,
+      "step": 3816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903526626349293e-05,
+      "loss": 0.9562,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901334583893546e-05,
+      "loss": 0.7818,
+      "step": 3818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899142106268408e-05,
+      "loss": 0.9625,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896949193635554e-05,
+      "loss": 0.9113,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894755846156698e-05,
+      "loss": 0.988,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589256206399357e-05,
+      "loss": 0.9019,
+      "step": 3822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890367847307948e-05,
+      "loss": 0.8585,
+      "step": 3823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588817319626163e-05,
+      "loss": 1.0211,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885978111016454e-05,
+      "loss": 0.9009,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883782591734292e-05,
+      "loss": 0.9437,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881586638577037e-05,
+      "loss": 0.2403,
+      "step": 3827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587939025170662e-05,
+      "loss": 0.9718,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587719343128501e-05,
+      "loss": 1.0525,
+      "step": 3829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58749961774742e-05,
+      "loss": 0.9296,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587279849043622e-05,
+      "loss": 0.899,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870600370333118e-05,
+      "loss": 0.944,
+      "step": 3832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868401817327e-05,
+      "loss": 0.9311,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586620283157998e-05,
+      "loss": 0.8001,
+      "step": 3834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864003413254213e-05,
+      "loss": 1.0139,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861803562511893e-05,
+      "loss": 1.0224,
+      "step": 3836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585960327951523e-05,
+      "loss": 0.9816,
+      "step": 3837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585740256442648e-05,
+      "loss": 0.9382,
+      "step": 3838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585520141740793e-05,
+      "loss": 0.9633,
+      "step": 3839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852999838621882e-05,
+      "loss": 0.9341,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850797828230695e-05,
+      "loss": 0.8971,
+      "step": 3841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848595386396738e-05,
+      "loss": 0.9442,
+      "step": 3842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846392513282426e-05,
+      "loss": 1.0377,
+      "step": 3843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844189209050198e-05,
+      "loss": 0.9382,
+      "step": 3844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584198547386253e-05,
+      "loss": 0.9223,
+      "step": 3845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839781307881923e-05,
+      "loss": 0.9091,
+      "step": 3846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583757671127092e-05,
+      "loss": 0.9597,
+      "step": 3847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835371684192084e-05,
+      "loss": 0.9642,
+      "step": 3848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583316622680802e-05,
+      "loss": 0.8593,
+      "step": 3849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830960339281363e-05,
+      "loss": 0.8629,
+      "step": 3850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828754021774765e-05,
+      "loss": 1.0184,
+      "step": 3851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826547274450933e-05,
+      "loss": 0.9811,
+      "step": 3852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824340097472594e-05,
+      "loss": 0.9828,
+      "step": 3853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58221324910025e-05,
+      "loss": 0.9196,
+      "step": 3854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819924455203444e-05,
+      "loss": 0.9777,
+      "step": 3855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581771599023825e-05,
+      "loss": 0.8661,
+      "step": 3856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815507096269777e-05,
+      "loss": 1.0026,
+      "step": 3857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813297773460904e-05,
+      "loss": 0.9424,
+      "step": 3858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811088021974548e-05,
+      "loss": 0.8484,
+      "step": 3859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580887784197366e-05,
+      "loss": 0.9912,
+      "step": 3860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806667233621223e-05,
+      "loss": 0.8507,
+      "step": 3861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804456197080247e-05,
+      "loss": 0.8816,
+      "step": 3862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580224473251377e-05,
+      "loss": 0.9707,
+      "step": 3863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580003284008488e-05,
+      "loss": 0.95,
+      "step": 3864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797820519956675e-05,
+      "loss": 0.9005,
+      "step": 3865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795607772292296e-05,
+      "loss": 0.9123,
+      "step": 3866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579339459725491e-05,
+      "loss": 0.9487,
+      "step": 3867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791180995007717e-05,
+      "loss": 0.8762,
+      "step": 3868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788966965713953e-05,
+      "loss": 0.916,
+      "step": 3869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786752509536886e-05,
+      "loss": 0.951,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784537626639808e-05,
+      "loss": 0.8879,
+      "step": 3871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782322317186042e-05,
+      "loss": 0.9641,
+      "step": 3872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780106581338957e-05,
+      "loss": 0.9528,
+      "step": 3873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777890419261935e-05,
+      "loss": 0.9026,
+      "step": 3874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775673831118403e-05,
+      "loss": 0.8688,
+      "step": 3875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577345681707181e-05,
+      "loss": 0.9546,
+      "step": 3876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577123937728564e-05,
+      "loss": 0.9163,
+      "step": 3877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576902151192341e-05,
+      "loss": 0.8955,
+      "step": 3878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766803221148676e-05,
+      "loss": 0.9815,
+      "step": 3879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764584505125e-05,
+      "loss": 0.9954,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762365364016005e-05,
+      "loss": 0.9592,
+      "step": 3881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760145797985326e-05,
+      "loss": 0.873,
+      "step": 3882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757925807196637e-05,
+      "loss": 0.3176,
+      "step": 3883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755705391813646e-05,
+      "loss": 0.9001,
+      "step": 3884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753484552000083e-05,
+      "loss": 0.8542,
+      "step": 3885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751263287919717e-05,
+      "loss": 0.9444,
+      "step": 3886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749041599736342e-05,
+      "loss": 0.9244,
+      "step": 3887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746819487613793e-05,
+      "loss": 0.2521,
+      "step": 3888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574459695171593e-05,
+      "loss": 0.971,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574237399220664e-05,
+      "loss": 1.0727,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574015060924985e-05,
+      "loss": 0.9002,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737926803009513e-05,
+      "loss": 0.9074,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735702573649612e-05,
+      "loss": 0.9075,
+      "step": 3893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733477921334167e-05,
+      "loss": 0.897,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731252846227224e-05,
+      "loss": 1.0172,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729027348492863e-05,
+      "loss": 0.2688,
+      "step": 3896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726801428295196e-05,
+      "loss": 0.2937,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724575085798356e-05,
+      "loss": 1.0134,
+      "step": 3898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722348321166526e-05,
+      "loss": 0.2784,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720121134563902e-05,
+      "loss": 0.9415,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571789352615472e-05,
+      "loss": 0.8348,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571566549610325e-05,
+      "loss": 0.9328,
+      "step": 3902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571343704457378e-05,
+      "loss": 0.2514,
+      "step": 3903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571120817173065e-05,
+      "loss": 0.9718,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708978877738208e-05,
+      "loss": 0.8979,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706749162760854e-05,
+      "loss": 0.9262,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704519026962997e-05,
+      "loss": 0.9661,
+      "step": 3907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57022884705091e-05,
+      "loss": 0.9175,
+      "step": 3908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570005749356364e-05,
+      "loss": 0.8935,
+      "step": 3909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569782609629113e-05,
+      "loss": 0.2543,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695594278856124e-05,
+      "loss": 0.9569,
+      "step": 3911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569336204142319e-05,
+      "loss": 0.8127,
+      "step": 3912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691129384156935e-05,
+      "loss": 0.8621,
+      "step": 3913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688896307222e-05,
+      "loss": 0.9445,
+      "step": 3914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686662810783057e-05,
+      "loss": 0.9116,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56844288950048e-05,
+      "loss": 0.9139,
+      "step": 3916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568219456005196e-05,
+      "loss": 0.8805,
+      "step": 3917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56799598060893e-05,
+      "loss": 0.8651,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677724633281614e-05,
+      "loss": 0.9884,
+      "step": 3919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675489041793726e-05,
+      "loss": 0.962,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567325303179049e-05,
+      "loss": 0.8254,
+      "step": 3921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671016603436792e-05,
+      "loss": 0.8482,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668779756897543e-05,
+      "loss": 0.9555,
+      "step": 3923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666542492337695e-05,
+      "loss": 0.8723,
+      "step": 3924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664304809922224e-05,
+      "loss": 0.946,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662066709816138e-05,
+      "loss": 0.9272,
+      "step": 3926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565982819218448e-05,
+      "loss": 0.9232,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657589257192316e-05,
+      "loss": 0.8637,
+      "step": 3928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565534990500475e-05,
+      "loss": 0.8522,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653110135786908e-05,
+      "loss": 1.0024,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565086994970396e-05,
+      "loss": 1.0164,
+      "step": 3931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648629346921097e-05,
+      "loss": 0.9901,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564638832760354e-05,
+      "loss": 1.01,
+      "step": 3933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644146891916546e-05,
+      "loss": 0.2693,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641905040025397e-05,
+      "loss": 0.94,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639662772095415e-05,
+      "loss": 0.912,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637420088291945e-05,
+      "loss": 0.7969,
+      "step": 3937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563517698878036e-05,
+      "loss": 0.9911,
+      "step": 3938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563293347372607e-05,
+      "loss": 0.8713,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630689543294518e-05,
+      "loss": 0.9461,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628445197651168e-05,
+      "loss": 0.9756,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626200436961525e-05,
+      "loss": 0.9855,
+      "step": 3942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623955261391113e-05,
+      "loss": 0.9321,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621709671105496e-05,
+      "loss": 0.9376,
+      "step": 3944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561946366627027e-05,
+      "loss": 0.9063,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617217247051054e-05,
+      "loss": 1.0143,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614970413613496e-05,
+      "loss": 0.8873,
+      "step": 3947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612723166123286e-05,
+      "loss": 0.9695,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610475504746138e-05,
+      "loss": 0.9308,
+      "step": 3949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560822742964779e-05,
+      "loss": 0.978,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605978940994027e-05,
+      "loss": 0.2874,
+      "step": 3951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603730038950644e-05,
+      "loss": 0.912,
+      "step": 3952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5601480723683488e-05,
+      "loss": 0.9146,
+      "step": 3953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559923099535841e-05,
+      "loss": 0.9124,
+      "step": 3954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559698085414132e-05,
+      "loss": 0.9841,
+      "step": 3955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594730300198143e-05,
+      "loss": 0.9218,
+      "step": 3956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592479333694833e-05,
+      "loss": 0.2783,
+      "step": 3957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590227954797384e-05,
+      "loss": 0.9381,
+      "step": 3958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587976163671805e-05,
+      "loss": 0.9824,
+      "step": 3959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585723960484152e-05,
+      "loss": 0.9447,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583471345400505e-05,
+      "loss": 0.8187,
+      "step": 3961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558121831858697e-05,
+      "loss": 0.9055,
+      "step": 3962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557896488020969e-05,
+      "loss": 0.891,
+      "step": 3963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576711030434835e-05,
+      "loss": 0.8323,
+      "step": 3964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574456769428606e-05,
+      "loss": 0.8311,
+      "step": 3965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557220209735723e-05,
+      "loss": 0.8896,
+      "step": 3966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569947014386973e-05,
+      "loss": 0.9177,
+      "step": 3967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567691520684128e-05,
+      "loss": 1.0157,
+      "step": 3968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556543561641501e-05,
+      "loss": 0.8386,
+      "step": 3969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556317930174598e-05,
+      "loss": 0.97,
+      "step": 3970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560922576843412e-05,
+      "loss": 0.8768,
+      "step": 3971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558665441873726e-05,
+      "loss": 1.0019,
+      "step": 3972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555640789700336e-05,
+      "loss": 0.8495,
+      "step": 3973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554149942398792e-05,
+      "loss": 0.883,
+      "step": 3974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551891578226518e-05,
+      "loss": 0.9967,
+      "step": 3975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549632804653082e-05,
+      "loss": 0.9347,
+      "step": 3976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554737362184504e-05,
+      "loss": 0.7975,
+      "step": 3977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545114029968987e-05,
+      "loss": 0.8804,
+      "step": 3978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554285402919155e-05,
+      "loss": 0.9774,
+      "step": 3979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554059361967938e-05,
+      "loss": 0.9068,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538332801599168e-05,
+      "loss": 0.9013,
+      "step": 3981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553607157511762e-05,
+      "loss": 0.9261,
+      "step": 3982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553380994040149e-05,
+      "loss": 0.2565,
+      "step": 3983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531547897617542e-05,
+      "loss": 0.8997,
+      "step": 3984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529285446932587e-05,
+      "loss": 0.9387,
+      "step": 3985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527022588513463e-05,
+      "loss": 0.9998,
+      "step": 3986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524759322527028e-05,
+      "loss": 0.9419,
+      "step": 3987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552249564914018e-05,
+      "loss": 0.8633,
+      "step": 3988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552023156851985e-05,
+      "loss": 0.8884,
+      "step": 3989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517967080832987e-05,
+      "loss": 0.8461,
+      "step": 3990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515702186246573e-05,
+      "loss": 0.933,
+      "step": 3991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513436884927634e-05,
+      "loss": 0.9195,
+      "step": 3992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511171177043207e-05,
+      "loss": 0.8555,
+      "step": 3993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550890506276037e-05,
+      "loss": 1.004,
+      "step": 3994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506638542246226e-05,
+      "loss": 0.9206,
+      "step": 3995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550437161566791e-05,
+      "loss": 0.9689,
+      "step": 3996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550210428319259e-05,
+      "loss": 0.8787,
+      "step": 3997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549983654498746e-05,
+      "loss": 0.8864,
+      "step": 3998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497568401219745e-05,
+      "loss": 0.9486,
+      "step": 3999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495299852056697e-05,
+      "loss": 0.91,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493030897665606e-05,
+      "loss": 0.8807,
+      "step": 4001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490761538213782e-05,
+      "loss": 0.9399,
+      "step": 4002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548849177386857e-05,
+      "loss": 0.9035,
+      "step": 4003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548622160479735e-05,
+      "loss": 0.9072,
+      "step": 4004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483951031167518e-05,
+      "loss": 0.9611,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548168005314651e-05,
+      "loss": 0.8338,
+      "step": 4006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479408670901795e-05,
+      "loss": 0.8882,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477136884600864e-05,
+      "loss": 0.8639,
+      "step": 4008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474864694411235e-05,
+      "loss": 0.9126,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472592100500472e-05,
+      "loss": 0.9375,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470319103036146e-05,
+      "loss": 0.8673,
+      "step": 4011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546804570218588e-05,
+      "loss": 0.8808,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546577189811731e-05,
+      "loss": 0.9699,
+      "step": 4013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546349769099811e-05,
+      "loss": 0.9493,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546122308099598e-05,
+      "loss": 0.8969,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545894806827866e-05,
+      "loss": 0.8882,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54566726530139e-05,
+      "loss": 1.0225,
+      "step": 4017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454396835369498e-05,
+      "loss": 0.8892,
+      "step": 4018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452120615513275e-05,
+      "loss": 0.2759,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449843993613076e-05,
+      "loss": 0.9671,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447566969836788e-05,
+      "loss": 0.9375,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445289544352316e-05,
+      "loss": 0.9075,
+      "step": 4022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443011717327596e-05,
+      "loss": 0.9166,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440733488930605e-05,
+      "loss": 0.8525,
+      "step": 4024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543845485932934e-05,
+      "loss": 1.0155,
+      "step": 4025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436175828691822e-05,
+      "loss": 0.9387,
+      "step": 4026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433896397186116e-05,
+      "loss": 0.9642,
+      "step": 4027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431616564980307e-05,
+      "loss": 0.9001,
+      "step": 4028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542933633224251e-05,
+      "loss": 0.9434,
+      "step": 4029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427055699140876e-05,
+      "loss": 0.9782,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424774665843575e-05,
+      "loss": 0.9415,
+      "step": 4031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422493232518812e-05,
+      "loss": 1.0016,
+      "step": 4032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420211399334827e-05,
+      "loss": 0.8716,
+      "step": 4033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417929166459883e-05,
+      "loss": 0.8311,
+      "step": 4034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541564653406227e-05,
+      "loss": 0.9417,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541336350231032e-05,
+      "loss": 0.9698,
+      "step": 4036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411080071372376e-05,
+      "loss": 0.9643,
+      "step": 4037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540879624141682e-05,
+      "loss": 0.9592,
+      "step": 4038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540651201261207e-05,
+      "loss": 0.9574,
+      "step": 4039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404227385126567e-05,
+      "loss": 0.8176,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401942359128777e-05,
+      "loss": 0.2962,
+      "step": 4041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53996569347872e-05,
+      "loss": 0.9976,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397371112270368e-05,
+      "loss": 0.8825,
+      "step": 4043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395084891746838e-05,
+      "loss": 0.9255,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392798273385194e-05,
+      "loss": 0.8803,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539051125735406e-05,
+      "loss": 0.8964,
+      "step": 4046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538822384382208e-05,
+      "loss": 0.8878,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385936032957926e-05,
+      "loss": 0.968,
+      "step": 4048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538364782493031e-05,
+      "loss": 0.9093,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381359219907964e-05,
+      "loss": 0.8617,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379070218059645e-05,
+      "loss": 0.8612,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376780819554155e-05,
+      "loss": 0.8767,
+      "step": 4052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374491024560315e-05,
+      "loss": 0.8218,
+      "step": 4053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537220083324697e-05,
+      "loss": 0.857,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369910245783006e-05,
+      "loss": 0.9762,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536761926233733e-05,
+      "loss": 1.0042,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365327883078882e-05,
+      "loss": 0.8922,
+      "step": 4057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536303610817663e-05,
+      "loss": 0.95,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360743937799576e-05,
+      "loss": 0.9875,
+      "step": 4059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535845137211674e-05,
+      "loss": 0.9564,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356158411297185e-05,
+      "loss": 0.9489,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353865055509985e-05,
+      "loss": 0.882,
+      "step": 4062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351571304924263e-05,
+      "loss": 0.9612,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349277159709156e-05,
+      "loss": 0.9091,
+      "step": 4064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346982620033845e-05,
+      "loss": 0.9363,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344687686067522e-05,
+      "loss": 0.9893,
+      "step": 4066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5342392357979426e-05,
+      "loss": 0.8738,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340096635938806e-05,
+      "loss": 0.8535,
+      "step": 4068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337800520114958e-05,
+      "loss": 0.8523,
+      "step": 4069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53355040106772e-05,
+      "loss": 0.9228,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333207107794873e-05,
+      "loss": 0.9244,
+      "step": 4071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533090981163736e-05,
+      "loss": 0.9397,
+      "step": 4072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328612122374055e-05,
+      "loss": 0.9037,
+      "step": 4073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326314040174404e-05,
+      "loss": 0.8888,
+      "step": 4074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532401556520786e-05,
+      "loss": 0.9127,
+      "step": 4075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321716697643918e-05,
+      "loss": 0.9121,
+      "step": 4076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53194174376521e-05,
+      "loss": 0.8683,
+      "step": 4077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317117785401952e-05,
+      "loss": 0.985,
+      "step": 4078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531481774106305e-05,
+      "loss": 0.8355,
+      "step": 4079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312517304805012e-05,
+      "loss": 0.8739,
+      "step": 4080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310216476797467e-05,
+      "loss": 0.9365,
+      "step": 4081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307915257210076e-05,
+      "loss": 0.8849,
+      "step": 4082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305613646212542e-05,
+      "loss": 0.9213,
+      "step": 4083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530331164397458e-05,
+      "loss": 0.9732,
+      "step": 4084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301009250665942e-05,
+      "loss": 0.873,
+      "step": 4085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298706466456415e-05,
+      "loss": 0.892,
+      "step": 4086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296403291515804e-05,
+      "loss": 0.9404,
+      "step": 4087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294099726013945e-05,
+      "loss": 0.903,
+      "step": 4088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291795770120707e-05,
+      "loss": 0.9058,
+      "step": 4089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528949142400599e-05,
+      "loss": 0.8026,
+      "step": 4090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5287186687839705e-05,
+      "loss": 0.9081,
+      "step": 4091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284881561791825e-05,
+      "loss": 0.8362,
+      "step": 4092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282576046032317e-05,
+      "loss": 0.9985,
+      "step": 4093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528027014073119e-05,
+      "loss": 0.9685,
+      "step": 4094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277963846058497e-05,
+      "loss": 0.848,
+      "step": 4095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275657162184298e-05,
+      "loss": 0.9461,
+      "step": 4096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527335008927869e-05,
+      "loss": 0.8765,
+      "step": 4097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271042627511797e-05,
+      "loss": 0.9046,
+      "step": 4098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268734777053777e-05,
+      "loss": 0.9753,
+      "step": 4099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266426538074804e-05,
+      "loss": 0.9753,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264117910745104e-05,
+      "loss": 0.9619,
+      "step": 4101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261808895234908e-05,
+      "loss": 0.8311,
+      "step": 4102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259499491714483e-05,
+      "loss": 0.886,
+      "step": 4103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257189700354134e-05,
+      "loss": 0.9194,
+      "step": 4104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254879521324179e-05,
+      "loss": 1.0474,
+      "step": 4105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252568954794977e-05,
+      "loss": 0.9606,
+      "step": 4106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525025800093691e-05,
+      "loss": 0.8781,
+      "step": 4107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247946659920392e-05,
+      "loss": 0.3085,
+      "step": 4108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245634931915855e-05,
+      "loss": 0.8503,
+      "step": 4109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524332281709378e-05,
+      "loss": 0.9193,
+      "step": 4110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241010315624655e-05,
+      "loss": 0.9579,
+      "step": 4111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238697427679013e-05,
+      "loss": 0.9277,
+      "step": 4112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236384153427401e-05,
+      "loss": 0.9595,
+      "step": 4113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234070493040408e-05,
+      "loss": 0.9473,
+      "step": 4114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231756446688639e-05,
+      "loss": 0.9308,
+      "step": 4115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229442014542738e-05,
+      "loss": 0.9077,
+      "step": 4116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227127196773371e-05,
+      "loss": 0.8758,
+      "step": 4117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224811993551236e-05,
+      "loss": 0.981,
+      "step": 4118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522249640504706e-05,
+      "loss": 0.9758,
+      "step": 4119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220180431431594e-05,
+      "loss": 0.9481,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217864072875622e-05,
+      "loss": 0.983,
+      "step": 4121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215547329549947e-05,
+      "loss": 0.8782,
+      "step": 4122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213230201625418e-05,
+      "loss": 0.8991,
+      "step": 4123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210912689272895e-05,
+      "loss": 1.0179,
+      "step": 4124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208594792663274e-05,
+      "loss": 0.8864,
+      "step": 4125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206276511967482e-05,
+      "loss": 0.9536,
+      "step": 4126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203957847356468e-05,
+      "loss": 0.8901,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201638799001215e-05,
+      "loss": 0.9795,
+      "step": 4128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199319367072726e-05,
+      "loss": 0.9744,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196999551742045e-05,
+      "loss": 0.8423,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194679353180233e-05,
+      "loss": 0.947,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192358771558384e-05,
+      "loss": 0.9374,
+      "step": 4132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190037807047622e-05,
+      "loss": 0.8008,
+      "step": 4133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518771645981909e-05,
+      "loss": 0.8299,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185394730043974e-05,
+      "loss": 0.86,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183072617893478e-05,
+      "loss": 0.8956,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180750123538832e-05,
+      "loss": 0.9232,
+      "step": 4137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178427247151306e-05,
+      "loss": 0.993,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176103988902187e-05,
+      "loss": 0.9252,
+      "step": 4139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173780348962792e-05,
+      "loss": 0.9418,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171456327504472e-05,
+      "loss": 0.9684,
+      "step": 4141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51691319246986e-05,
+      "loss": 0.2488,
+      "step": 4142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166807140716582e-05,
+      "loss": 0.9005,
+      "step": 4143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164481975729846e-05,
+      "loss": 0.9781,
+      "step": 4144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162156429909853e-05,
+      "loss": 0.9252,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515983050342809e-05,
+      "loss": 0.9378,
+      "step": 4146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157504196456076e-05,
+      "loss": 1.0052,
+      "step": 4147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155177509165352e-05,
+      "loss": 0.8374,
+      "step": 4148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515285044172749e-05,
+      "loss": 0.8891,
+      "step": 4149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150522994314091e-05,
+      "loss": 0.9834,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148195167096785e-05,
+      "loss": 0.9382,
+      "step": 4151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145866960247219e-05,
+      "loss": 0.8164,
+      "step": 4152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143538373937088e-05,
+      "loss": 0.9068,
+      "step": 4153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51412094083381e-05,
+      "loss": 0.2594,
+      "step": 4154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513888006362199e-05,
+      "loss": 0.8442,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136550339960532e-05,
+      "loss": 0.8901,
+      "step": 4156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134220237525519e-05,
+      "loss": 0.9138,
+      "step": 4157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131889756488777e-05,
+      "loss": 0.873,
+      "step": 4158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129558897022157e-05,
+      "loss": 0.8975,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127227659297534e-05,
+      "loss": 0.8213,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512489604348682e-05,
+      "loss": 0.8083,
+      "step": 4161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512256404976195e-05,
+      "loss": 0.9335,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120231678294888e-05,
+      "loss": 0.8828,
+      "step": 4163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117898929257621e-05,
+      "loss": 0.8693,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115565802822172e-05,
+      "loss": 0.9064,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113232299160587e-05,
+      "loss": 0.8778,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110898418444938e-05,
+      "loss": 0.962,
+      "step": 4167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108564160847326e-05,
+      "loss": 0.926,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510622952653989e-05,
+      "loss": 0.9934,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510389451569478e-05,
+      "loss": 0.7968,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101559128484182e-05,
+      "loss": 0.8442,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099223365080313e-05,
+      "loss": 0.9535,
+      "step": 4172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509688722565541e-05,
+      "loss": 0.8666,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094550710381748e-05,
+      "loss": 0.9786,
+      "step": 4174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092213819431618e-05,
+      "loss": 1.0129,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089876552977345e-05,
+      "loss": 1.019,
+      "step": 4176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087538911191281e-05,
+      "loss": 0.7843,
+      "step": 4177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508520089424581e-05,
+      "loss": 1.0435,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082862502313335e-05,
+      "loss": 0.8213,
+      "step": 4179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508052373556629e-05,
+      "loss": 1.0022,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078184594177143e-05,
+      "loss": 0.9425,
+      "step": 4181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075845078318381e-05,
+      "loss": 0.7841,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073505188162515e-05,
+      "loss": 0.7998,
+      "step": 4183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071164923882107e-05,
+      "loss": 0.8872,
+      "step": 4184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068824285649716e-05,
+      "loss": 0.8595,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066483273637947e-05,
+      "loss": 1.0702,
+      "step": 4186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064141888019427e-05,
+      "loss": 0.917,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061800128966818e-05,
+      "loss": 0.9507,
+      "step": 4188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505945799665279e-05,
+      "loss": 0.9452,
+      "step": 4189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057115491250071e-05,
+      "loss": 0.9973,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054772612931386e-05,
+      "loss": 0.2708,
+      "step": 4191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052429361869503e-05,
+      "loss": 0.9246,
+      "step": 4192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505008573823722e-05,
+      "loss": 0.9501,
+      "step": 4193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047741742207355e-05,
+      "loss": 0.8705,
+      "step": 4194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504539737395276e-05,
+      "loss": 0.9915,
+      "step": 4195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043052633646302e-05,
+      "loss": 0.9301,
+      "step": 4196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040707521460893e-05,
+      "loss": 0.8801,
+      "step": 4197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038362037569458e-05,
+      "loss": 0.8702,
+      "step": 4198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036016182144957e-05,
+      "loss": 0.94,
+      "step": 4199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033669955360379e-05,
+      "loss": 0.9676,
+      "step": 4200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031323357388727e-05,
+      "loss": 0.7595,
+      "step": 4201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028976388403052e-05,
+      "loss": 0.9403,
+      "step": 4202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026629048576413e-05,
+      "loss": 0.2981,
+      "step": 4203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024281338081909e-05,
+      "loss": 0.8596,
+      "step": 4204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021933257092665e-05,
+      "loss": 0.8961,
+      "step": 4205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019584805781823e-05,
+      "loss": 0.8924,
+      "step": 4206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017235984322565e-05,
+      "loss": 1.0944,
+      "step": 4207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014886792888095e-05,
+      "loss": 0.9435,
+      "step": 4208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012537231651643e-05,
+      "loss": 0.8864,
+      "step": 4209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010187300786467e-05,
+      "loss": 0.9608,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007837000465855e-05,
+      "loss": 0.8404,
+      "step": 4211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005486330863116e-05,
+      "loss": 0.9839,
+      "step": 4212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003135292151595e-05,
+      "loss": 0.8329,
+      "step": 4213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000783884504659e-05,
+      "loss": 0.9986,
+      "step": 4214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998432108095701e-05,
+      "loss": 0.908,
+      "step": 4215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996079963098145e-05,
+      "loss": 0.8728,
+      "step": 4216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993727449685435e-05,
+      "loss": 0.875,
+      "step": 4217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991374568031054e-05,
+      "loss": 0.949,
+      "step": 4218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989021318308503e-05,
+      "loss": 0.9698,
+      "step": 4219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986667700691309e-05,
+      "loss": 0.9896,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984313715353036e-05,
+      "loss": 0.8582,
+      "step": 4221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981959362467265e-05,
+      "loss": 0.8991,
+      "step": 4222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979604642207608e-05,
+      "loss": 0.8822,
+      "step": 4223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977249554747705e-05,
+      "loss": 0.9874,
+      "step": 4224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974894100261223e-05,
+      "loss": 0.8946,
+      "step": 4225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972538278921853e-05,
+      "loss": 0.886,
+      "step": 4226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970182090903315e-05,
+      "loss": 0.9389,
+      "step": 4227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967825536379361e-05,
+      "loss": 0.8649,
+      "step": 4228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965468615523759e-05,
+      "loss": 0.9858,
+      "step": 4229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963111328510317e-05,
+      "loss": 0.922,
+      "step": 4230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960753675512855e-05,
+      "loss": 0.9733,
+      "step": 4231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958395656705236e-05,
+      "loss": 0.86,
+      "step": 4232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495603727226134e-05,
+      "loss": 0.8153,
+      "step": 4233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953678522355075e-05,
+      "loss": 0.9772,
+      "step": 4234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951319407160375e-05,
+      "loss": 0.9856,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494895992685121e-05,
+      "loss": 0.861,
+      "step": 4236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946600081601564e-05,
+      "loss": 0.9741,
+      "step": 4237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944239871585453e-05,
+      "loss": 0.9175,
+      "step": 4238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941879296976927e-05,
+      "loss": 0.9695,
+      "step": 4239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939518357950053e-05,
+      "loss": 0.9465,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493715705467893e-05,
+      "loss": 0.8894,
+      "step": 4241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934795387337682e-05,
+      "loss": 0.942,
+      "step": 4242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932433356100457e-05,
+      "loss": 0.9517,
+      "step": 4243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930070961141439e-05,
+      "loss": 0.8114,
+      "step": 4244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927708202634828e-05,
+      "loss": 0.9536,
+      "step": 4245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492534508075486e-05,
+      "loss": 0.9968,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922981595675792e-05,
+      "loss": 0.9525,
+      "step": 4247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920617747571908e-05,
+      "loss": 0.9967,
+      "step": 4248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918253536617523e-05,
+      "loss": 0.8749,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915888962986971e-05,
+      "loss": 0.9241,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913524026854622e-05,
+      "loss": 0.9092,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911158728394867e-05,
+      "loss": 0.9048,
+      "step": 4252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908793067782125e-05,
+      "loss": 0.8871,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906427045190843e-05,
+      "loss": 0.8954,
+      "step": 4254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904060660795492e-05,
+      "loss": 0.8812,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490169391477057e-05,
+      "loss": 0.9432,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899326807290608e-05,
+      "loss": 1.0436,
+      "step": 4257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896959338530154e-05,
+      "loss": 0.8004,
+      "step": 4258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894591508663786e-05,
+      "loss": 0.8484,
+      "step": 4259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892223317866114e-05,
+      "loss": 0.8813,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889854766311768e-05,
+      "loss": 0.9992,
+      "step": 4261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488748585417541e-05,
+      "loss": 0.8147,
+      "step": 4262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488511658163172e-05,
+      "loss": 0.8891,
+      "step": 4263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882746948855413e-05,
+      "loss": 0.8838,
+      "step": 4264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880376956021226e-05,
+      "loss": 0.9115,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878006603303926e-05,
+      "loss": 0.9121,
+      "step": 4266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875635890878313e-05,
+      "loss": 0.8631,
+      "step": 4267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873264818919189e-05,
+      "loss": 1.071,
+      "step": 4268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870893387601409e-05,
+      "loss": 0.8927,
+      "step": 4269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868521597099843e-05,
+      "loss": 0.8839,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486614944758939e-05,
+      "loss": 1.0228,
+      "step": 4271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486377693924497e-05,
+      "loss": 0.9084,
+      "step": 4272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861404072241537e-05,
+      "loss": 0.8676,
+      "step": 4273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859030846754063e-05,
+      "loss": 0.9689,
+      "step": 4274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485665726295756e-05,
+      "loss": 0.9368,
+      "step": 4275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854283321027057e-05,
+      "loss": 0.9188,
+      "step": 4276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851909021137604e-05,
+      "loss": 0.2782,
+      "step": 4277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849534363464293e-05,
+      "loss": 0.8944,
+      "step": 4278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847159348182222e-05,
+      "loss": 0.9459,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844783975466536e-05,
+      "loss": 0.8557,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484240824549239e-05,
+      "loss": 1.0214,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840032158434978e-05,
+      "loss": 0.9607,
+      "step": 4282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837655714469512e-05,
+      "loss": 0.9644,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835278913771234e-05,
+      "loss": 0.8851,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832901756515413e-05,
+      "loss": 0.9365,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830524242877335e-05,
+      "loss": 0.9571,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828146373032328e-05,
+      "loss": 0.8823,
+      "step": 4287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825768147155737e-05,
+      "loss": 0.9332,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482338956542293e-05,
+      "loss": 0.8432,
+      "step": 4289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821010628009309e-05,
+      "loss": 1.0034,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818631335090299e-05,
+      "loss": 0.9525,
+      "step": 4291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481625168684135e-05,
+      "loss": 0.9079,
+      "step": 4292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813871683437939e-05,
+      "loss": 0.9064,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481149132505557e-05,
+      "loss": 0.8978,
+      "step": 4294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809110611869773e-05,
+      "loss": 0.8784,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806729544056102e-05,
+      "loss": 0.8691,
+      "step": 4296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804348121790142e-05,
+      "loss": 0.9034,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48019663452475e-05,
+      "loss": 0.9422,
+      "step": 4298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799584214603808e-05,
+      "loss": 0.8501,
+      "step": 4299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479720173003473e-05,
+      "loss": 0.8546,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794818891715948e-05,
+      "loss": 0.9592,
+      "step": 4301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792435699823176e-05,
+      "loss": 0.8595,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790052154532157e-05,
+      "loss": 0.871,
+      "step": 4303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478766825601865e-05,
+      "loss": 0.9156,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478528400445845e-05,
+      "loss": 0.976,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782899400027368e-05,
+      "loss": 0.825,
+      "step": 4306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780514442901253e-05,
+      "loss": 0.3,
+      "step": 4307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778129133255968e-05,
+      "loss": 0.9496,
+      "step": 4308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775743471267415e-05,
+      "loss": 0.9747,
+      "step": 4309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773357457111509e-05,
+      "loss": 0.8548,
+      "step": 4310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770971090964195e-05,
+      "loss": 0.9441,
+      "step": 4311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476858437300145e-05,
+      "loss": 0.9279,
+      "step": 4312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766197303399275e-05,
+      "loss": 0.8984,
+      "step": 4313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763809882333688e-05,
+      "loss": 0.9746,
+      "step": 4314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761422109980746e-05,
+      "loss": 0.8853,
+      "step": 4315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759033986516518e-05,
+      "loss": 0.9672,
+      "step": 4316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756645512117108e-05,
+      "loss": 0.9435,
+      "step": 4317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475425668695865e-05,
+      "loss": 0.9502,
+      "step": 4318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751867511217294e-05,
+      "loss": 0.76,
+      "step": 4319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749477985069218e-05,
+      "loss": 0.8594,
+      "step": 4320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747088108690631e-05,
+      "loss": 0.844,
+      "step": 4321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744697882257764e-05,
+      "loss": 0.9077,
+      "step": 4322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742307305946871e-05,
+      "loss": 0.9062,
+      "step": 4323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739916379934238e-05,
+      "loss": 0.9663,
+      "step": 4324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737525104396175e-05,
+      "loss": 0.9343,
+      "step": 4325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735133479509011e-05,
+      "loss": 0.9241,
+      "step": 4326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732741505449116e-05,
+      "loss": 0.946,
+      "step": 4327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473034918239287e-05,
+      "loss": 0.9472,
+      "step": 4328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727956510516683e-05,
+      "loss": 0.8682,
+      "step": 4329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725563489996997e-05,
+      "loss": 0.8011,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723170121010272e-05,
+      "loss": 0.8994,
+      "step": 4331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720776403732998e-05,
+      "loss": 0.9352,
+      "step": 4332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718382338341692e-05,
+      "loss": 0.7891,
+      "step": 4333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715987925012896e-05,
+      "loss": 0.8937,
+      "step": 4334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713593163923168e-05,
+      "loss": 0.9886,
+      "step": 4335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711198055249107e-05,
+      "loss": 0.9094,
+      "step": 4336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708802599167327e-05,
+      "loss": 0.8652,
+      "step": 4337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706406795854471e-05,
+      "loss": 1.002,
+      "step": 4338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470401064548721e-05,
+      "loss": 0.9375,
+      "step": 4339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701614148242236e-05,
+      "loss": 0.8517,
+      "step": 4340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699217304296269e-05,
+      "loss": 0.2354,
+      "step": 4341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696820113826054e-05,
+      "loss": 0.9523,
+      "step": 4342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694422577008366e-05,
+      "loss": 0.8167,
+      "step": 4343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692024694019993e-05,
+      "loss": 0.9039,
+      "step": 4344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689626465037765e-05,
+      "loss": 0.9454,
+      "step": 4345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687227890238525e-05,
+      "loss": 0.9328,
+      "step": 4346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684828969799147e-05,
+      "loss": 0.8463,
+      "step": 4347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682429703896529e-05,
+      "loss": 0.8918,
+      "step": 4348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680030092707596e-05,
+      "loss": 0.8193,
+      "step": 4349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677630136409298e-05,
+      "loss": 0.8911,
+      "step": 4350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675229835178607e-05,
+      "loss": 0.8617,
+      "step": 4351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672829189192526e-05,
+      "loss": 0.9874,
+      "step": 4352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670428198628077e-05,
+      "loss": 0.8415,
+      "step": 4353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668026863662315e-05,
+      "loss": 0.9525,
+      "step": 4354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665625184472316e-05,
+      "loss": 0.9497,
+      "step": 4355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663223161235181e-05,
+      "loss": 0.9128,
+      "step": 4356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660820794128035e-05,
+      "loss": 0.9384,
+      "step": 4357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658418083328033e-05,
+      "loss": 0.9452,
+      "step": 4358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656015029012353e-05,
+      "loss": 0.9133,
+      "step": 4359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653611631358198e-05,
+      "loss": 0.9227,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651207890542796e-05,
+      "loss": 0.8806,
+      "step": 4361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648803806743404e-05,
+      "loss": 0.9468,
+      "step": 4362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646399380137292e-05,
+      "loss": 0.8883,
+      "step": 4363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643994610901775e-05,
+      "loss": 0.8959,
+      "step": 4364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641589499214182e-05,
+      "loss": 0.9378,
+      "step": 4365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463918404525186e-05,
+      "loss": 0.9125,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636778249192194e-05,
+      "loss": 0.9952,
+      "step": 4367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634372111212592e-05,
+      "loss": 0.8854,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631965631490476e-05,
+      "loss": 1.022,
+      "step": 4369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629558810203314e-05,
+      "loss": 0.9663,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627151647528579e-05,
+      "loss": 0.905,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624744143643773e-05,
+      "loss": 0.934,
+      "step": 4372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462233629872644e-05,
+      "loss": 0.8394,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619928112954129e-05,
+      "loss": 0.9023,
+      "step": 4374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617519586504419e-05,
+      "loss": 0.8774,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615110719554927e-05,
+      "loss": 0.8321,
+      "step": 4376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612701512283271e-05,
+      "loss": 0.8686,
+      "step": 4377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610291964867118e-05,
+      "loss": 0.8418,
+      "step": 4378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607882077484145e-05,
+      "loss": 0.9172,
+      "step": 4379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605471850312066e-05,
+      "loss": 0.9255,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603061283528606e-05,
+      "loss": 0.2583,
+      "step": 4381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600650377311523e-05,
+      "loss": 0.9733,
+      "step": 4382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598239131838603e-05,
+      "loss": 0.8929,
+      "step": 4383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459582754728765e-05,
+      "loss": 0.8378,
+      "step": 4384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593415623836495e-05,
+      "loss": 0.967,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591003361663004e-05,
+      "loss": 1.0194,
+      "step": 4386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588590760945047e-05,
+      "loss": 0.9363,
+      "step": 4387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458617782186054e-05,
+      "loss": 0.921,
+      "step": 4388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583764544587413e-05,
+      "loss": 0.9241,
+      "step": 4389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581350929303622e-05,
+      "loss": 0.8857,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578936976187147e-05,
+      "loss": 0.9056,
+      "step": 4391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576522685416002e-05,
+      "loss": 0.9745,
+      "step": 4392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574108057168209e-05,
+      "loss": 0.8665,
+      "step": 4393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571693091621832e-05,
+      "loss": 0.8799,
+      "step": 4394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569277788954952e-05,
+      "loss": 0.9142,
+      "step": 4395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566862149345675e-05,
+      "loss": 0.8684,
+      "step": 4396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564446172972129e-05,
+      "loss": 0.9273,
+      "step": 4397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562029860012475e-05,
+      "loss": 0.9477,
+      "step": 4398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559613210644887e-05,
+      "loss": 0.8922,
+      "step": 4399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557196225047578e-05,
+      "loss": 0.8901,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554778903398777e-05,
+      "loss": 0.8604,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552361245876734e-05,
+      "loss": 0.8781,
+      "step": 4402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549943252659737e-05,
+      "loss": 0.988,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547524923926086e-05,
+      "loss": 0.8914,
+      "step": 4404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545106259854111e-05,
+      "loss": 1.0809,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542687260622167e-05,
+      "loss": 0.9003,
+      "step": 4406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540267926408635e-05,
+      "loss": 0.9668,
+      "step": 4407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537848257391916e-05,
+      "loss": 1.052,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535428253750437e-05,
+      "loss": 0.9166,
+      "step": 4409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533007915662656e-05,
+      "loss": 0.2876,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530587243307047e-05,
+      "loss": 0.9452,
+      "step": 4411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528166236862111e-05,
+      "loss": 0.9115,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525744896506382e-05,
+      "loss": 0.8512,
+      "step": 4413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523323222418406e-05,
+      "loss": 0.8797,
+      "step": 4414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520901214776757e-05,
+      "loss": 0.9609,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518478873760042e-05,
+      "loss": 0.8841,
+      "step": 4416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516056199546886e-05,
+      "loss": 0.9056,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513633192315935e-05,
+      "loss": 0.8939,
+      "step": 4418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511209852245865e-05,
+      "loss": 0.8969,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508786179515374e-05,
+      "loss": 0.9109,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450636217430319e-05,
+      "loss": 0.9011,
+      "step": 4421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450393783678806e-05,
+      "loss": 0.9274,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501513167148753e-05,
+      "loss": 0.8731,
+      "step": 4423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499088165564069e-05,
+      "loss": 0.2755,
+      "step": 4424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496662832212826e-05,
+      "loss": 0.2797,
+      "step": 4425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494237167273876e-05,
+      "loss": 0.9272,
+      "step": 4426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449181117092609e-05,
+      "loss": 0.8054,
+      "step": 4427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489384843348356e-05,
+      "loss": 0.9481,
+      "step": 4428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486958184719597e-05,
+      "loss": 0.9285,
+      "step": 4429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484531195218756e-05,
+      "loss": 0.8632,
+      "step": 4430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482103875024804e-05,
+      "loss": 0.9699,
+      "step": 4431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479676224316734e-05,
+      "loss": 0.9501,
+      "step": 4432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477248243273558e-05,
+      "loss": 0.821,
+      "step": 4433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474819932074323e-05,
+      "loss": 0.9223,
+      "step": 4434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447239129089809e-05,
+      "loss": 0.9325,
+      "step": 4435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446996231992395e-05,
+      "loss": 0.9987,
+      "step": 4436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446753301933102e-05,
+      "loss": 1.032,
+      "step": 4437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465103389298441e-05,
+      "loss": 0.8994,
+      "step": 4438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462673430005365e-05,
+      "loss": 0.8529,
+      "step": 4439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460243141630992e-05,
+      "loss": 0.8524,
+      "step": 4440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457812524354525e-05,
+      "loss": 0.9614,
+      "step": 4441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455381578355203e-05,
+      "loss": 0.9307,
+      "step": 4442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452950303812285e-05,
+      "loss": 0.9291,
+      "step": 4443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450518700905059e-05,
+      "loss": 0.8886,
+      "step": 4444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448086769812826e-05,
+      "loss": 0.9255,
+      "step": 4445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445654510714926e-05,
+      "loss": 0.7972,
+      "step": 4446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444322192379071e-05,
+      "loss": 0.9017,
+      "step": 4447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440789009219566e-05,
+      "loss": 0.2652,
+      "step": 4448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438355767180894e-05,
+      "loss": 0.9064,
+      "step": 4449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435922197854124e-05,
+      "loss": 0.8851,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443348830141871e-05,
+      "loss": 0.8964,
+      "step": 4451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431054078054126e-05,
+      "loss": 0.9168,
+      "step": 4452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428619527939883e-05,
+      "loss": 0.2751,
+      "step": 4453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426184651255495e-05,
+      "loss": 0.9571,
+      "step": 4454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423749448180519e-05,
+      "loss": 0.8734,
+      "step": 4455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442131391889453e-05,
+      "loss": 0.955,
+      "step": 4456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418878063577122e-05,
+      "loss": 0.905,
+      "step": 4457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416441882407922e-05,
+      "loss": 0.2777,
+      "step": 4458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414005375566569e-05,
+      "loss": 0.8512,
+      "step": 4459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411568543232736e-05,
+      "loss": 0.9485,
+      "step": 4460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409131385586118e-05,
+      "loss": 0.8748,
+      "step": 4461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406693902806434e-05,
+      "loss": 0.8893,
+      "step": 4462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404256095073428e-05,
+      "loss": 0.9059,
+      "step": 4463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440181796256686e-05,
+      "loss": 0.909,
+      "step": 4464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399379505466522e-05,
+      "loss": 0.867,
+      "step": 4465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439694072395223e-05,
+      "loss": 0.8705,
+      "step": 4466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439450161820382e-05,
+      "loss": 0.9207,
+      "step": 4467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392062188401154e-05,
+      "loss": 0.9137,
+      "step": 4468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438962243472412e-05,
+      "loss": 0.9972,
+      "step": 4469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387182357352622e-05,
+      "loss": 0.9304,
+      "step": 4470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384741956466599e-05,
+      "loss": 0.9297,
+      "step": 4471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382301232246006e-05,
+      "loss": 1.0016,
+      "step": 4472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379860184870822e-05,
+      "loss": 1.0022,
+      "step": 4473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377418814521058e-05,
+      "loss": 0.2941,
+      "step": 4474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374977121376736e-05,
+      "loss": 0.8637,
+      "step": 4475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372535105617908e-05,
+      "loss": 0.9637,
+      "step": 4476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370092767424659e-05,
+      "loss": 0.8911,
+      "step": 4477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436765010697708e-05,
+      "loss": 1.0427,
+      "step": 4478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43652071244553e-05,
+      "loss": 0.9661,
+      "step": 4479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362763820039467e-05,
+      "loss": 1.0258,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360320193909747e-05,
+      "loss": 0.2196,
+      "step": 4481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357876246246338e-05,
+      "loss": 0.9131,
+      "step": 4482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355431977229461e-05,
+      "loss": 0.872,
+      "step": 4483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352987387039356e-05,
+      "loss": 0.9553,
+      "step": 4484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350542475856289e-05,
+      "loss": 0.9446,
+      "step": 4485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348097243860553e-05,
+      "loss": 0.9611,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4345651691232456e-05,
+      "loss": 0.8672,
+      "step": 4487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343205818152336e-05,
+      "loss": 0.8936,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340759624800561e-05,
+      "loss": 0.984,
+      "step": 4489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433831311135751e-05,
+      "loss": 0.9158,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335866278003587e-05,
+      "loss": 0.9269,
+      "step": 4491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433341912491923e-05,
+      "loss": 0.978,
+      "step": 4492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330971652284893e-05,
+      "loss": 0.9808,
+      "step": 4493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328523860281052e-05,
+      "loss": 1.042,
+      "step": 4494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326075749088213e-05,
+      "loss": 0.2846,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323627318886897e-05,
+      "loss": 0.8586,
+      "step": 4496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321178569857654e-05,
+      "loss": 0.8116,
+      "step": 4497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318729502181064e-05,
+      "loss": 0.846,
+      "step": 4498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316280116037716e-05,
+      "loss": 0.9742,
+      "step": 4499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313830411608231e-05,
+      "loss": 0.8615,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311380389073253e-05,
+      "loss": 0.8923,
+      "step": 4501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430893004861345e-05,
+      "loss": 0.2775,
+      "step": 4502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306479390409508e-05,
+      "loss": 0.8756,
+      "step": 4503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304028414642146e-05,
+      "loss": 0.8744,
+      "step": 4504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301577121492101e-05,
+      "loss": 0.8496,
+      "step": 4505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299125511140126e-05,
+      "loss": 1.0112,
+      "step": 4506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296673583767014e-05,
+      "loss": 0.8776,
+      "step": 4507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294221339553567e-05,
+      "loss": 1.0002,
+      "step": 4508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291768778680614e-05,
+      "loss": 0.9062,
+      "step": 4509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289315901329014e-05,
+      "loss": 0.8972,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286862707679642e-05,
+      "loss": 0.9518,
+      "step": 4511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284409197913398e-05,
+      "loss": 0.9915,
+      "step": 4512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281955372211205e-05,
+      "loss": 0.8311,
+      "step": 4513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279501230754012e-05,
+      "loss": 0.9913,
+      "step": 4514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427704677372279e-05,
+      "loss": 0.9127,
+      "step": 4515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427459200129853e-05,
+      "loss": 0.8839,
+      "step": 4516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272136913662252e-05,
+      "loss": 0.8645,
+      "step": 4517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269681510994993e-05,
+      "loss": 0.8954,
+      "step": 4518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267225793477817e-05,
+      "loss": 0.9305,
+      "step": 4519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264769761291816e-05,
+      "loss": 0.8521,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262313414618096e-05,
+      "loss": 0.9011,
+      "step": 4521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259856753637787e-05,
+      "loss": 0.8579,
+      "step": 4522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257399778532048e-05,
+      "loss": 0.8885,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254942489482059e-05,
+      "loss": 0.3001,
+      "step": 4524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252484886669025e-05,
+      "loss": 0.2617,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250026970274166e-05,
+      "loss": 0.9431,
+      "step": 4526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247568740478735e-05,
+      "loss": 0.9909,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245110197464001e-05,
+      "loss": 0.236,
+      "step": 4528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424265134141126e-05,
+      "loss": 0.7979,
+      "step": 4529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240192172501834e-05,
+      "loss": 0.9568,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237732690917059e-05,
+      "loss": 0.9182,
+      "step": 4531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42352728968383e-05,
+      "loss": 0.8546,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232812790446948e-05,
+      "loss": 0.9474,
+      "step": 4533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230352371924407e-05,
+      "loss": 0.8639,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227891641452118e-05,
+      "loss": 0.8275,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225430599211535e-05,
+      "loss": 0.9833,
+      "step": 4536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222969245384133e-05,
+      "loss": 0.9286,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220507580151415e-05,
+      "loss": 0.8598,
+      "step": 4538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218045603694911e-05,
+      "loss": 0.8984,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215583316196164e-05,
+      "loss": 0.9821,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213120717836754e-05,
+      "loss": 0.9781,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210657808798265e-05,
+      "loss": 0.7928,
+      "step": 4542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208194589262316e-05,
+      "loss": 0.9727,
+      "step": 4543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420573105941055e-05,
+      "loss": 0.3141,
+      "step": 4544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420326721942463e-05,
+      "loss": 0.896,
+      "step": 4545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420080306948624e-05,
+      "loss": 1.0049,
+      "step": 4546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198338609777087e-05,
+      "loss": 0.9088,
+      "step": 4547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195873840478906e-05,
+      "loss": 0.8227,
+      "step": 4548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193408761773446e-05,
+      "loss": 0.2348,
+      "step": 4549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419094337384249e-05,
+      "loss": 0.9535,
+      "step": 4550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188477676867835e-05,
+      "loss": 0.9629,
+      "step": 4551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186011671031301e-05,
+      "loss": 0.9258,
+      "step": 4552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183545356514738e-05,
+      "loss": 0.7954,
+      "step": 4553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418107873350001e-05,
+      "loss": 0.2818,
+      "step": 4554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178611802169014e-05,
+      "loss": 0.8375,
+      "step": 4555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176144562703655e-05,
+      "loss": 0.8248,
+      "step": 4556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173677015285873e-05,
+      "loss": 0.9097,
+      "step": 4557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417120916009763e-05,
+      "loss": 0.9106,
+      "step": 4558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168740997320907e-05,
+      "loss": 0.9383,
+      "step": 4559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166272527137704e-05,
+      "loss": 0.8995,
+      "step": 4560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163803749730052e-05,
+      "loss": 0.2613,
+      "step": 4561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416133466528e-05,
+      "loss": 0.9154,
+      "step": 4562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158865273969622e-05,
+      "loss": 0.9537,
+      "step": 4563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156395575981008e-05,
+      "loss": 0.9807,
+      "step": 4564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153925571496277e-05,
+      "loss": 0.9522,
+      "step": 4565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151455260697575e-05,
+      "loss": 0.9196,
+      "step": 4566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148984643767058e-05,
+      "loss": 0.9377,
+      "step": 4567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146513720886917e-05,
+      "loss": 0.2583,
+      "step": 4568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144042492239354e-05,
+      "loss": 0.9346,
+      "step": 4569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141570958006603e-05,
+      "loss": 0.91,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139099118370915e-05,
+      "loss": 0.894,
+      "step": 4571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136626973514574e-05,
+      "loss": 0.907,
+      "step": 4572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134154523619863e-05,
+      "loss": 0.8575,
+      "step": 4573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131681768869115e-05,
+      "loss": 0.9135,
+      "step": 4574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129208709444666e-05,
+      "loss": 0.2534,
+      "step": 4575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126735345528885e-05,
+      "loss": 0.893,
+      "step": 4576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124261677304163e-05,
+      "loss": 0.8949,
+      "step": 4577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121787704952903e-05,
+      "loss": 0.8986,
+      "step": 4578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119313428657541e-05,
+      "loss": 0.9644,
+      "step": 4579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116838848600532e-05,
+      "loss": 0.914,
+      "step": 4580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114363964964356e-05,
+      "loss": 0.9532,
+      "step": 4581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111888777931508e-05,
+      "loss": 0.9836,
+      "step": 4582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109413287684515e-05,
+      "loss": 0.8888,
+      "step": 4583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106937494405921e-05,
+      "loss": 0.9259,
+      "step": 4584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104461398278294e-05,
+      "loss": 0.9401,
+      "step": 4585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101984999484217e-05,
+      "loss": 0.9148,
+      "step": 4586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099508298206312e-05,
+      "loss": 1.0562,
+      "step": 4587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097031294627204e-05,
+      "loss": 0.9378,
+      "step": 4588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094553988929552e-05,
+      "loss": 0.9127,
+      "step": 4589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092076381296038e-05,
+      "loss": 0.8829,
+      "step": 4590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089598471909357e-05,
+      "loss": 0.9018,
+      "step": 4591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408712026095224e-05,
+      "loss": 0.9455,
+      "step": 4592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084641748607427e-05,
+      "loss": 0.8706,
+      "step": 4593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082162935057683e-05,
+      "loss": 0.8599,
+      "step": 4594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079683820485803e-05,
+      "loss": 0.9157,
+      "step": 4595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077204405074598e-05,
+      "loss": 0.9961,
+      "step": 4596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40747246890069e-05,
+      "loss": 0.8831,
+      "step": 4597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072244672465567e-05,
+      "loss": 0.8817,
+      "step": 4598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406976435563348e-05,
+      "loss": 0.8232,
+      "step": 4599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067283738693534e-05,
+      "loss": 0.9443,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4064802821828654e-05,
+      "loss": 0.2771,
+      "step": 4601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062321605221787e-05,
+      "loss": 0.8889,
+      "step": 4602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40598400890559e-05,
+      "loss": 0.8752,
+      "step": 4603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057358273513976e-05,
+      "loss": 0.8005,
+      "step": 4604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054876158779036e-05,
+      "loss": 0.7604,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052393745034106e-05,
+      "loss": 0.9836,
+      "step": 4606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049911032462242e-05,
+      "loss": 0.8704,
+      "step": 4607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047428021246528e-05,
+      "loss": 0.7937,
+      "step": 4608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044944711570055e-05,
+      "loss": 0.9073,
+      "step": 4609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042461103615943e-05,
+      "loss": 0.8512,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039977197567345e-05,
+      "loss": 0.9138,
+      "step": 4611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403749299360742e-05,
+      "loss": 0.9011,
+      "step": 4612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4035008491919355e-05,
+      "loss": 0.9417,
+      "step": 4613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032523692686363e-05,
+      "loss": 0.8632,
+      "step": 4614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030038596091671e-05,
+      "loss": 0.9183,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027553202318531e-05,
+      "loss": 0.9464,
+      "step": 4616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025067511550226e-05,
+      "loss": 0.9154,
+      "step": 4617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022581523970046e-05,
+      "loss": 0.85,
+      "step": 4618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020095239761312e-05,
+      "loss": 0.9088,
+      "step": 4619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017608659107364e-05,
+      "loss": 1.0599,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015121782191567e-05,
+      "loss": 0.8948,
+      "step": 4621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012634609197302e-05,
+      "loss": 1.0236,
+      "step": 4622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010147140307977e-05,
+      "loss": 0.24,
+      "step": 4623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400765937570702e-05,
+      "loss": 0.9011,
+      "step": 4624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005171315577874e-05,
+      "loss": 0.8426,
+      "step": 4625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4002682960104024e-05,
+      "loss": 0.8308,
+      "step": 4626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000194309468955e-05,
+      "loss": 0.9224,
+      "step": 4627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997705363856184e-05,
+      "loss": 0.9438,
+      "step": 4628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399521612344925e-05,
+      "loss": 0.8672,
+      "step": 4629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992726588431707e-05,
+      "loss": 0.7995,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990236758987137e-05,
+      "loss": 1.0255,
+      "step": 4631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987746635299141e-05,
+      "loss": 0.907,
+      "step": 4632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985256217551349e-05,
+      "loss": 0.7612,
+      "step": 4633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982765505927401e-05,
+      "loss": 0.8725,
+      "step": 4634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980274500610966e-05,
+      "loss": 0.9368,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977783201785732e-05,
+      "loss": 0.2473,
+      "step": 4636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975291609635408e-05,
+      "loss": 0.9044,
+      "step": 4637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972799724343728e-05,
+      "loss": 1.0411,
+      "step": 4638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397030754609445e-05,
+      "loss": 0.8679,
+      "step": 4639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967815075071342e-05,
+      "loss": 0.8513,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965322311458203e-05,
+      "loss": 1.0392,
+      "step": 4641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962829255438852e-05,
+      "loss": 0.9823,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960335907197132e-05,
+      "loss": 0.8988,
+      "step": 4643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957842266916899e-05,
+      "loss": 0.877,
+      "step": 4644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955348334782042e-05,
+      "loss": 0.9721,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952854110976462e-05,
+      "loss": 0.9351,
+      "step": 4646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950359595684085e-05,
+      "loss": 0.917,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947864789088859e-05,
+      "loss": 0.9261,
+      "step": 4648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945369691374757e-05,
+      "loss": 0.8646,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942874302725763e-05,
+      "loss": 0.9379,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940378623325892e-05,
+      "loss": 0.9977,
+      "step": 4651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393788265335918e-05,
+      "loss": 0.9434,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935386393009678e-05,
+      "loss": 0.8981,
+      "step": 4653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932889842461468e-05,
+      "loss": 0.8482,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930393001898642e-05,
+      "loss": 0.9098,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927895871505317e-05,
+      "loss": 0.9566,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925398451465642e-05,
+      "loss": 0.8853,
+      "step": 4657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922900741963774e-05,
+      "loss": 0.8973,
+      "step": 4658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920402743183896e-05,
+      "loss": 0.2561,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917904455310213e-05,
+      "loss": 0.9417,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915405878526953e-05,
+      "loss": 0.8825,
+      "step": 4661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912907013018357e-05,
+      "loss": 0.9096,
+      "step": 4662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910407858968703e-05,
+      "loss": 0.9201,
+      "step": 4663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390790841656227e-05,
+      "loss": 0.9558,
+      "step": 4664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905408685983379e-05,
+      "loss": 0.9623,
+      "step": 4665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902908667416355e-05,
+      "loss": 0.9755,
+      "step": 4666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900408361045554e-05,
+      "loss": 0.8776,
+      "step": 4667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897907767055351e-05,
+      "loss": 0.8831,
+      "step": 4668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895406885630142e-05,
+      "loss": 0.7755,
+      "step": 4669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892905716954346e-05,
+      "loss": 0.9524,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890404261212395e-05,
+      "loss": 0.8549,
+      "step": 4671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887902518588756e-05,
+      "loss": 1.0182,
+      "step": 4672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885400489267907e-05,
+      "loss": 0.9783,
+      "step": 4673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882898173434345e-05,
+      "loss": 0.8627,
+      "step": 4674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880395571272601e-05,
+      "loss": 0.9262,
+      "step": 4675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877892682967212e-05,
+      "loss": 0.8946,
+      "step": 4676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875389508702744e-05,
+      "loss": 0.9215,
+      "step": 4677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872886048663787e-05,
+      "loss": 1.0639,
+      "step": 4678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870382303034945e-05,
+      "loss": 0.9214,
+      "step": 4679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386787827200085e-05,
+      "loss": 0.8978,
+      "step": 4680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386537395574615e-05,
+      "loss": 0.8913,
+      "step": 4681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3862869354455513e-05,
+      "loss": 0.9971,
+      "step": 4682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860364468313632e-05,
+      "loss": 0.92,
+      "step": 4683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857859297505216e-05,
+      "loss": 0.9792,
+      "step": 4684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855353842215006e-05,
+      "loss": 0.9413,
+      "step": 4685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852848102627749e-05,
+      "loss": 0.9326,
+      "step": 4686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850342078928226e-05,
+      "loss": 0.8793,
+      "step": 4687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847835771301228e-05,
+      "loss": 0.951,
+      "step": 4688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845329179931577e-05,
+      "loss": 1.0013,
+      "step": 4689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842822305004109e-05,
+      "loss": 0.8755,
+      "step": 4690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840315146703682e-05,
+      "loss": 0.8714,
+      "step": 4691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837807705215179e-05,
+      "loss": 0.8454,
+      "step": 4692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835299980723493e-05,
+      "loss": 0.9193,
+      "step": 4693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832791973413557e-05,
+      "loss": 0.9073,
+      "step": 4694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830283683470306e-05,
+      "loss": 0.9995,
+      "step": 4695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382777511107871e-05,
+      "loss": 0.9804,
+      "step": 4696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825266256423746e-05,
+      "loss": 0.8781,
+      "step": 4697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822757119690423e-05,
+      "loss": 0.9449,
+      "step": 4698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382024770106376e-05,
+      "loss": 0.2404,
+      "step": 4699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817738000728815e-05,
+      "loss": 0.9917,
+      "step": 4700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815228018870652e-05,
+      "loss": 0.1975,
+      "step": 4701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812717755674353e-05,
+      "loss": 0.8434,
+      "step": 4702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810207211325031e-05,
+      "loss": 0.9252,
+      "step": 4703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807696386007816e-05,
+      "loss": 0.9178,
+      "step": 4704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805185279907858e-05,
+      "loss": 0.9581,
+      "step": 4705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802673893210333e-05,
+      "loss": 0.8541,
+      "step": 4706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800162226100422e-05,
+      "loss": 0.924,
+      "step": 4707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797650278763344e-05,
+      "loss": 0.8753,
+      "step": 4708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379513805138433e-05,
+      "loss": 0.8778,
+      "step": 4709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792625544148642e-05,
+      "loss": 0.9236,
+      "step": 4710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790112757241542e-05,
+      "loss": 0.9245,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787599690848333e-05,
+      "loss": 0.9694,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785086345154326e-05,
+      "loss": 1.0434,
+      "step": 4713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378257272034486e-05,
+      "loss": 0.2777,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378005881660529e-05,
+      "loss": 0.8616,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777544634121001e-05,
+      "loss": 0.9637,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377503017307738e-05,
+      "loss": 0.7902,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772515433659849e-05,
+      "loss": 1.0276,
+      "step": 4718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770000416053851e-05,
+      "loss": 0.9991,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376748512044484e-05,
+      "loss": 0.8724,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37649695470183e-05,
+      "loss": 0.8958,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376245369595973e-05,
+      "loss": 1.0301,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375993756745465e-05,
+      "loss": 0.8258,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757421161688604e-05,
+      "loss": 0.9373,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754904478847153e-05,
+      "loss": 0.775,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752387519115877e-05,
+      "loss": 0.803,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749870282680382e-05,
+      "loss": 0.8721,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747352769726292e-05,
+      "loss": 0.8585,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744834980439241e-05,
+      "loss": 0.9925,
+      "step": 4729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742316915004908e-05,
+      "loss": 0.9832,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739798573608965e-05,
+      "loss": 0.8406,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737279956437121e-05,
+      "loss": 0.2448,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734761063675101e-05,
+      "loss": 1.0189,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732241895508654e-05,
+      "loss": 0.9538,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729722452123538e-05,
+      "loss": 0.9893,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372720273370554e-05,
+      "loss": 0.8594,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724682740440476e-05,
+      "loss": 0.9734,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372216247251416e-05,
+      "loss": 0.9235,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719641930112446e-05,
+      "loss": 0.9857,
+      "step": 4739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717121113421197e-05,
+      "loss": 0.9602,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714600022626306e-05,
+      "loss": 0.9311,
+      "step": 4741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712078657913672e-05,
+      "loss": 0.9272,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370955701946923e-05,
+      "loss": 0.8833,
+      "step": 4743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707035107478923e-05,
+      "loss": 0.25,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704512922128718e-05,
+      "loss": 0.9736,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701990463604611e-05,
+      "loss": 0.2643,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699467732092602e-05,
+      "loss": 1.0091,
+      "step": 4747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696944727778723e-05,
+      "loss": 0.944,
+      "step": 4748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694421450849022e-05,
+      "loss": 0.8239,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691897901489566e-05,
+      "loss": 0.9819,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689374079886444e-05,
+      "loss": 0.8982,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368684998622577e-05,
+      "loss": 0.8961,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684325620693666e-05,
+      "loss": 0.8673,
+      "step": 4753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368180098347628e-05,
+      "loss": 0.2616,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367927607475979e-05,
+      "loss": 0.8818,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676750894730378e-05,
+      "loss": 1.0067,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367422544357425e-05,
+      "loss": 0.9331,
+      "step": 4757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671699721477642e-05,
+      "loss": 0.9736,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669173728626798e-05,
+      "loss": 0.8931,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366664746520799e-05,
+      "loss": 0.8946,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664120931407505e-05,
+      "loss": 0.9739,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661594127411653e-05,
+      "loss": 1.0188,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365906705340676e-05,
+      "loss": 0.2398,
+      "step": 4763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656539709579177e-05,
+      "loss": 0.9109,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654012096115274e-05,
+      "loss": 0.9139,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651484213201433e-05,
+      "loss": 0.8409,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364895606102407e-05,
+      "loss": 0.9281,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364642763976961e-05,
+      "loss": 0.9,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643898949624496e-05,
+      "loss": 0.898,
+      "step": 4769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641369990775204e-05,
+      "loss": 0.9576,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638840763408217e-05,
+      "loss": 0.8968,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363631126771004e-05,
+      "loss": 0.9294,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363378150386721e-05,
+      "loss": 1.0224,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631251472066263e-05,
+      "loss": 0.9281,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628721172493767e-05,
+      "loss": 0.9777,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626190605336313e-05,
+      "loss": 0.9218,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623659770780506e-05,
+      "loss": 0.9409,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621128669012974e-05,
+      "loss": 0.9001,
+      "step": 4778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618597300220358e-05,
+      "loss": 0.2512,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616065664589324e-05,
+      "loss": 0.9753,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613533762306555e-05,
+      "loss": 0.8733,
+      "step": 4781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611001593558763e-05,
+      "loss": 0.9109,
+      "step": 4782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608469158532669e-05,
+      "loss": 0.987,
+      "step": 4783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605936457415011e-05,
+      "loss": 0.9393,
+      "step": 4784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603403490392561e-05,
+      "loss": 0.2069,
+      "step": 4785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36008702576521e-05,
+      "loss": 0.2404,
+      "step": 4786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598336759380426e-05,
+      "loss": 0.8926,
+      "step": 4787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595802995764364e-05,
+      "loss": 0.9264,
+      "step": 4788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593268966990758e-05,
+      "loss": 0.942,
+      "step": 4789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590734673246465e-05,
+      "loss": 0.8999,
+      "step": 4790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588200114718373e-05,
+      "loss": 0.9878,
+      "step": 4791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585665291593374e-05,
+      "loss": 0.9116,
+      "step": 4792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583130204058396e-05,
+      "loss": 1.0051,
+      "step": 4793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580594852300374e-05,
+      "loss": 0.8957,
+      "step": 4794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578059236506267e-05,
+      "loss": 0.8691,
+      "step": 4795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575523356863053e-05,
+      "loss": 0.9318,
+      "step": 4796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572987213557731e-05,
+      "loss": 0.9918,
+      "step": 4797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357045080677732e-05,
+      "loss": 0.8619,
+      "step": 4798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567914136708855e-05,
+      "loss": 0.8656,
+      "step": 4799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356537720353939e-05,
+      "loss": 0.8408,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562840007456007e-05,
+      "loss": 1.0453,
+      "step": 4801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560302548645795e-05,
+      "loss": 0.9442,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557764827295867e-05,
+      "loss": 0.8565,
+      "step": 4803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555226843593365e-05,
+      "loss": 0.9535,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552688597725437e-05,
+      "loss": 0.8543,
+      "step": 4805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355015008987925e-05,
+      "loss": 0.9084,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547611320242004e-05,
+      "loss": 0.9394,
+      "step": 4807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354507228900091e-05,
+      "loss": 0.9463,
+      "step": 4808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542532996343193e-05,
+      "loss": 0.8918,
+      "step": 4809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539993442456103e-05,
+      "loss": 0.9403,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537453627526914e-05,
+      "loss": 0.886,
+      "step": 4811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534913551742908e-05,
+      "loss": 0.9811,
+      "step": 4812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532373215291396e-05,
+      "loss": 0.9306,
+      "step": 4813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529832618359708e-05,
+      "loss": 0.9265,
+      "step": 4814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527291761135179e-05,
+      "loss": 1.0502,
+      "step": 4815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524750643805184e-05,
+      "loss": 0.8999,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522209266557105e-05,
+      "loss": 1.0131,
+      "step": 4817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519667629578341e-05,
+      "loss": 0.8475,
+      "step": 4818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517125733056322e-05,
+      "loss": 0.8893,
+      "step": 4819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514583577178482e-05,
+      "loss": 0.8956,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512041162132283e-05,
+      "loss": 0.9626,
+      "step": 4821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509498488105209e-05,
+      "loss": 0.885,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506955555284756e-05,
+      "loss": 0.9841,
+      "step": 4823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350441236385844e-05,
+      "loss": 0.9163,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501868914013807e-05,
+      "loss": 0.9516,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499325205938404e-05,
+      "loss": 0.2484,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496781239819805e-05,
+      "loss": 0.9082,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494237015845611e-05,
+      "loss": 0.2171,
+      "step": 4828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491692534203434e-05,
+      "loss": 0.8759,
+      "step": 4829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489147795080901e-05,
+      "loss": 0.8741,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348660279866567e-05,
+      "loss": 0.9097,
+      "step": 4831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484057545145407e-05,
+      "loss": 0.9166,
+      "step": 4832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34815120347078e-05,
+      "loss": 0.9776,
+      "step": 4833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347896626754056e-05,
+      "loss": 0.8334,
+      "step": 4834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476420243831419e-05,
+      "loss": 0.3017,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473873963768108e-05,
+      "loss": 0.9458,
+      "step": 4836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471327427538406e-05,
+      "loss": 0.9427,
+      "step": 4837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468780635330092e-05,
+      "loss": 0.8488,
+      "step": 4838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466233587330965e-05,
+      "loss": 0.9292,
+      "step": 4839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463686283728854e-05,
+      "loss": 0.8742,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461138724711592e-05,
+      "loss": 0.9497,
+      "step": 4841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458590910467044e-05,
+      "loss": 0.8931,
+      "step": 4842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456042841183083e-05,
+      "loss": 0.8215,
+      "step": 4843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345349451704761e-05,
+      "loss": 0.8721,
+      "step": 4844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450945938248536e-05,
+      "loss": 0.9305,
+      "step": 4845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448397104973801e-05,
+      "loss": 0.8834,
+      "step": 4846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445848017411352e-05,
+      "loss": 0.9684,
+      "step": 4847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443298675749163e-05,
+      "loss": 0.9212,
+      "step": 4848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344074908017523e-05,
+      "loss": 0.8807,
+      "step": 4849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438199230877558e-05,
+      "loss": 0.9199,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435649128044171e-05,
+      "loss": 1.0172,
+      "step": 4851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343309877186312e-05,
+      "loss": 0.9275,
+      "step": 4852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430548162522472e-05,
+      "loss": 0.8973,
+      "step": 4853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427997300210307e-05,
+      "loss": 0.8653,
+      "step": 4854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342544618511473e-05,
+      "loss": 0.2831,
+      "step": 4855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422894817423865e-05,
+      "loss": 0.8583,
+      "step": 4856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420343197325848e-05,
+      "loss": 1.1006,
+      "step": 4857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417791325008838e-05,
+      "loss": 0.9275,
+      "step": 4858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415239200661012e-05,
+      "loss": 0.9911,
+      "step": 4859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341268682447057e-05,
+      "loss": 0.983,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341013419662572e-05,
+      "loss": 0.9341,
+      "step": 4861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407581317314702e-05,
+      "loss": 0.9447,
+      "step": 4862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405028186725761e-05,
+      "loss": 0.8855,
+      "step": 4863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340247480504717e-05,
+      "loss": 0.9694,
+      "step": 4864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399921172467217e-05,
+      "loss": 0.9178,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339736728917421e-05,
+      "loss": 0.972,
+      "step": 4866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394813155356473e-05,
+      "loss": 0.9165,
+      "step": 4867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339225877120235e-05,
+      "loss": 0.8604,
+      "step": 4868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389704136900205e-05,
+      "loss": 0.8351,
+      "step": 4869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387149252638419e-05,
+      "loss": 0.8738,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338459411860539e-05,
+      "loss": 0.8545,
+      "step": 4871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382038734989535e-05,
+      "loss": 0.9614,
+      "step": 4872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379483101979292e-05,
+      "loss": 0.2921,
+      "step": 4873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376927219763114e-05,
+      "loss": 0.9204,
+      "step": 4874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374371088529474e-05,
+      "loss": 0.9286,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371814708466866e-05,
+      "loss": 0.8637,
+      "step": 4876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369258079763796e-05,
+      "loss": 0.9392,
+      "step": 4877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366701202608793e-05,
+      "loss": 0.946,
+      "step": 4878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364144077190404e-05,
+      "loss": 0.9036,
+      "step": 4879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361586703697196e-05,
+      "loss": 0.9572,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335902908231775e-05,
+      "loss": 0.9232,
+      "step": 4881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356471213240662e-05,
+      "loss": 0.9613,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353913096654557e-05,
+      "loss": 0.9551,
+      "step": 4883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351354732748072e-05,
+      "loss": 0.8712,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.8786,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346237263728603e-05,
+      "loss": 0.9542,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343678158992987e-05,
+      "loss": 0.8696,
+      "step": 4887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334111880769172e-05,
+      "loss": 0.9004,
+      "step": 4888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338559210013534e-05,
+      "loss": 0.9131,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335999366147179e-05,
+      "loss": 0.9351,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333439276281414e-05,
+      "loss": 0.8578,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333087894060503e-05,
+      "loss": 0.8959,
+      "step": 4892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328318359306823e-05,
+      "loss": 0.9142,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325757532575607e-05,
+      "loss": 0.9185,
+      "step": 4894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323196460600229e-05,
+      "loss": 0.9109,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320635143569542e-05,
+      "loss": 0.9268,
+      "step": 4896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318073581672419e-05,
+      "loss": 0.9209,
+      "step": 4897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315511775097751e-05,
+      "loss": 0.8541,
+      "step": 4898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312949724034448e-05,
+      "loss": 0.9099,
+      "step": 4899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310387428671438e-05,
+      "loss": 0.8651,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3307824889197667e-05,
+      "loss": 0.9697,
+      "step": 4901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305262105802102e-05,
+      "loss": 0.9482,
+      "step": 4902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302699078673716e-05,
+      "loss": 0.9735,
+      "step": 4903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300135808001517e-05,
+      "loss": 0.8314,
+      "step": 4904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329757229397452e-05,
+      "loss": 0.8875,
+      "step": 4905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329500853678176e-05,
+      "loss": 0.8669,
+      "step": 4906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292444536612296e-05,
+      "loss": 0.9008,
+      "step": 4907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289880293655191e-05,
+      "loss": 0.256,
+      "step": 4908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287315808099537e-05,
+      "loss": 0.8784,
+      "step": 4909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284751080134446e-05,
+      "loss": 0.9716,
+      "step": 4910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282186109949039e-05,
+      "loss": 0.8781,
+      "step": 4911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279620897732459e-05,
+      "loss": 0.9481,
+      "step": 4912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327705544367387e-05,
+      "loss": 0.8357,
+      "step": 4913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274489747962447e-05,
+      "loss": 0.9347,
+      "step": 4914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271923810787388e-05,
+      "loss": 0.9499,
+      "step": 4915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269357632337908e-05,
+      "loss": 0.8808,
+      "step": 4916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326679121280324e-05,
+      "loss": 0.9124,
+      "step": 4917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326422455237263e-05,
+      "loss": 0.2783,
+      "step": 4918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261657651235348e-05,
+      "loss": 0.8466,
+      "step": 4919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259090509580683e-05,
+      "loss": 0.9348,
+      "step": 4920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256523127597936e-05,
+      "loss": 0.959,
+      "step": 4921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325395550547642e-05,
+      "loss": 0.8516,
+      "step": 4922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251387643405489e-05,
+      "loss": 0.2386,
+      "step": 4923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248819541574487e-05,
+      "loss": 0.8255,
+      "step": 4924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246251200172791e-05,
+      "loss": 0.9274,
+      "step": 4925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243682619389796e-05,
+      "loss": 0.8152,
+      "step": 4926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241113799414906e-05,
+      "loss": 0.9734,
+      "step": 4927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238544740437553e-05,
+      "loss": 0.8948,
+      "step": 4928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323597544264718e-05,
+      "loss": 0.9364,
+      "step": 4929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233405906233247e-05,
+      "loss": 0.9526,
+      "step": 4930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230836131385234e-05,
+      "loss": 1.0149,
+      "step": 4931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228266118292642e-05,
+      "loss": 0.9749,
+      "step": 4932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225695867144985e-05,
+      "loss": 0.9684,
+      "step": 4933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223125378131792e-05,
+      "loss": 0.8864,
+      "step": 4934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220554651442614e-05,
+      "loss": 0.9443,
+      "step": 4935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217983687267022e-05,
+      "loss": 0.8348,
+      "step": 4936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215412485794597e-05,
+      "loss": 0.9204,
+      "step": 4937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212841047214947e-05,
+      "loss": 0.9706,
+      "step": 4938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210269371717686e-05,
+      "loss": 0.8934,
+      "step": 4939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320769745949245e-05,
+      "loss": 0.9963,
+      "step": 4940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205125310728904e-05,
+      "loss": 0.8341,
+      "step": 4941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202552925616712e-05,
+      "loss": 0.8886,
+      "step": 4942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3199980304345566e-05,
+      "loss": 0.8982,
+      "step": 4943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197407447105173e-05,
+      "loss": 0.8958,
+      "step": 4944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319483435408526e-05,
+      "loss": 0.9402,
+      "step": 4945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192261025475564e-05,
+      "loss": 0.9089,
+      "step": 4946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318968746146585e-05,
+      "loss": 0.9561,
+      "step": 4947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187113662245895e-05,
+      "loss": 0.9162,
+      "step": 4948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3184539628005484e-05,
+      "loss": 0.8484,
+      "step": 4949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318196535893444e-05,
+      "loss": 0.9408,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179390855222584e-05,
+      "loss": 0.8778,
+      "step": 4951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176816117059763e-05,
+      "loss": 0.9135,
+      "step": 4952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174241144635847e-05,
+      "loss": 0.8995,
+      "step": 4953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171665938140711e-05,
+      "loss": 0.9499,
+      "step": 4954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316909049776425e-05,
+      "loss": 0.9713,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166514823696387e-05,
+      "loss": 0.9175,
+      "step": 4956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163938916127048e-05,
+      "loss": 0.9171,
+      "step": 4957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161362775246185e-05,
+      "loss": 0.958,
+      "step": 4958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158786401243766e-05,
+      "loss": 0.9807,
+      "step": 4959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156209794309774e-05,
+      "loss": 0.9056,
+      "step": 4960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153632954634205e-05,
+      "loss": 0.9113,
+      "step": 4961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151055882407085e-05,
+      "loss": 0.9122,
+      "step": 4962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148478577818446e-05,
+      "loss": 0.9259,
+      "step": 4963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145901041058342e-05,
+      "loss": 0.8405,
+      "step": 4964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143323272316843e-05,
+      "loss": 0.903,
+      "step": 4965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140745271784034e-05,
+      "loss": 0.8506,
+      "step": 4966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138167039650021e-05,
+      "loss": 0.8828,
+      "step": 4967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135588576104921e-05,
+      "loss": 1.0095,
+      "step": 4968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133009881338877e-05,
+      "loss": 0.3066,
+      "step": 4969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313043095554204e-05,
+      "loss": 0.81,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127851798904585e-05,
+      "loss": 0.9222,
+      "step": 4971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31252724116167e-05,
+      "loss": 0.9787,
+      "step": 4972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122692793868593e-05,
+      "loss": 0.9116,
+      "step": 4973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120112945850484e-05,
+      "loss": 0.8572,
+      "step": 4974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117532867752613e-05,
+      "loss": 0.8366,
+      "step": 4975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114952559765244e-05,
+      "loss": 0.8178,
+      "step": 4976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311237202207864e-05,
+      "loss": 0.9161,
+      "step": 4977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109791254883101e-05,
+      "loss": 0.8681,
+      "step": 4978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107210258368935e-05,
+      "loss": 0.9839,
+      "step": 4979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104629032726459e-05,
+      "loss": 0.8704,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102047578146024e-05,
+      "loss": 0.7789,
+      "step": 4981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099465894817982e-05,
+      "loss": 0.9109,
+      "step": 4982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096883982932709e-05,
+      "loss": 1.0607,
+      "step": 4983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094301842680601e-05,
+      "loss": 0.3003,
+      "step": 4984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091719474252067e-05,
+      "loss": 0.2227,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089136877837528e-05,
+      "loss": 0.9546,
+      "step": 4986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086554053627433e-05,
+      "loss": 0.8437,
+      "step": 4987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083971001812237e-05,
+      "loss": 0.8966,
+      "step": 4988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081387722582417e-05,
+      "loss": 1.0103,
+      "step": 4989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078804216128473e-05,
+      "loss": 0.9258,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076220482640905e-05,
+      "loss": 0.2764,
+      "step": 4991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073636522310244e-05,
+      "loss": 0.8936,
+      "step": 4992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071052335327035e-05,
+      "loss": 0.8898,
+      "step": 4993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068467921881835e-05,
+      "loss": 0.9397,
+      "step": 4994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065883282165223e-05,
+      "loss": 0.9534,
+      "step": 4995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063298416367794e-05,
+      "loss": 0.9318,
+      "step": 4996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060713324680154e-05,
+      "loss": 0.8118,
+      "step": 4997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058128007292931e-05,
+      "loss": 0.2843,
+      "step": 4998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055542464396771e-05,
+      "loss": 0.8634,
+      "step": 4999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052956696182336e-05,
+      "loss": 0.9245,
+      "step": 5000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050370702840293e-05,
+      "loss": 0.9178,
+      "step": 5001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047784484561346e-05,
+      "loss": 0.9914,
+      "step": 5002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045198041536198e-05,
+      "loss": 1.0237,
+      "step": 5003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042611373955578e-05,
+      "loss": 0.7921,
+      "step": 5004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040024482010232e-05,
+      "loss": 0.903,
+      "step": 5005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037437365890914e-05,
+      "loss": 0.9703,
+      "step": 5006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034850025788404e-05,
+      "loss": 0.9272,
+      "step": 5007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032262461893492e-05,
+      "loss": 0.8742,
+      "step": 5008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029674674396991e-05,
+      "loss": 0.9775,
+      "step": 5009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027086663489723e-05,
+      "loss": 0.8658,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302449842936253e-05,
+      "loss": 0.2616,
+      "step": 5011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021909972206275e-05,
+      "loss": 0.9062,
+      "step": 5012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019321292211827e-05,
+      "loss": 0.9608,
+      "step": 5013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016732389570081e-05,
+      "loss": 0.9482,
+      "step": 5014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014143264471945e-05,
+      "loss": 0.8535,
+      "step": 5015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011553917108341e-05,
+      "loss": 0.9365,
+      "step": 5016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008964347670215e-05,
+      "loss": 0.9877,
+      "step": 5017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006374556348518e-05,
+      "loss": 1.0685,
+      "step": 5018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003784543334224e-05,
+      "loss": 0.2706,
+      "step": 5019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001194308818326e-05,
+      "loss": 0.9594,
+      "step": 5020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998603852991829e-05,
+      "loss": 0.8538,
+      "step": 5021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996013176045752e-05,
+      "loss": 0.8512,
+      "step": 5022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299342227817114e-05,
+      "loss": 0.9812,
+      "step": 5023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990831159559041e-05,
+      "loss": 0.8977,
+      "step": 5024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988239820400533e-05,
+      "loss": 0.9595,
+      "step": 5025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985648260886698e-05,
+      "loss": 0.8128,
+      "step": 5026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983056481208642e-05,
+      "loss": 0.9427,
+      "step": 5027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980464481557489e-05,
+      "loss": 0.884,
+      "step": 5028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977872262124364e-05,
+      "loss": 0.2249,
+      "step": 5029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2975279823100433e-05,
+      "loss": 0.9529,
+      "step": 5030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972687164676855e-05,
+      "loss": 0.9528,
+      "step": 5031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970094287044824e-05,
+      "loss": 0.9857,
+      "step": 5032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2967501190395529e-05,
+      "loss": 1.0229,
+      "step": 5033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964907874920197e-05,
+      "loss": 0.87,
+      "step": 5034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962314340810053e-05,
+      "loss": 0.9045,
+      "step": 5035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959720588256357e-05,
+      "loss": 0.9483,
+      "step": 5036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957126617450367e-05,
+      "loss": 0.9153,
+      "step": 5037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954532428583364e-05,
+      "loss": 0.7944,
+      "step": 5038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951938021846648e-05,
+      "loss": 0.7877,
+      "step": 5039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949343397431532e-05,
+      "loss": 0.2393,
+      "step": 5040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946748555529349e-05,
+      "loss": 0.9217,
+      "step": 5041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944153496331443e-05,
+      "loss": 0.9127,
+      "step": 5042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941558220029171e-05,
+      "loss": 0.2666,
+      "step": 5043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938962726813915e-05,
+      "loss": 0.9049,
+      "step": 5044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936367016877072e-05,
+      "loss": 0.8524,
+      "step": 5045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293377109041005e-05,
+      "loss": 0.7842,
+      "step": 5046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931174947604269e-05,
+      "loss": 0.8167,
+      "step": 5047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928578588651174e-05,
+      "loss": 0.878,
+      "step": 5048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925982013742229e-05,
+      "loss": 0.9619,
+      "step": 5049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923385223068898e-05,
+      "loss": 0.9827,
+      "step": 5050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920788216822678e-05,
+      "loss": 0.979,
+      "step": 5051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918190995195072e-05,
+      "loss": 0.8612,
+      "step": 5052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915593558377596e-05,
+      "loss": 0.9279,
+      "step": 5053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912995906561795e-05,
+      "loss": 0.9667,
+      "step": 5054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910398039939222e-05,
+      "loss": 0.908,
+      "step": 5055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907799958701439e-05,
+      "loss": 0.8873,
+      "step": 5056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905201663040036e-05,
+      "loss": 0.8958,
+      "step": 5057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902603153146614e-05,
+      "loss": 0.9723,
+      "step": 5058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900004429212783e-05,
+      "loss": 0.9478,
+      "step": 5059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897405491430181e-05,
+      "loss": 1.0268,
+      "step": 5060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894806339990455e-05,
+      "loss": 0.2341,
+      "step": 5061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892206975085269e-05,
+      "loss": 0.8466,
+      "step": 5062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28896073969063e-05,
+      "loss": 0.8974,
+      "step": 5063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887007605645245e-05,
+      "loss": 0.8875,
+      "step": 5064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884407601493814e-05,
+      "loss": 0.946,
+      "step": 5065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881807384643731e-05,
+      "loss": 0.8678,
+      "step": 5066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879206955286748e-05,
+      "loss": 0.8745,
+      "step": 5067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876606313614609e-05,
+      "loss": 0.8489,
+      "step": 5068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874005459819097e-05,
+      "loss": 0.2737,
+      "step": 5069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871404394091997e-05,
+      "loss": 0.8697,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868803116625116e-05,
+      "loss": 1.0408,
+      "step": 5071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866201627610275e-05,
+      "loss": 0.9597,
+      "step": 5072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863599927239307e-05,
+      "loss": 0.2694,
+      "step": 5073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860998015704063e-05,
+      "loss": 0.8764,
+      "step": 5074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858395893196416e-05,
+      "loss": 0.8744,
+      "step": 5075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855793559908244e-05,
+      "loss": 0.8728,
+      "step": 5076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853191016031445e-05,
+      "loss": 0.8778,
+      "step": 5077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850588261757935e-05,
+      "loss": 0.2764,
+      "step": 5078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847985297279644e-05,
+      "loss": 0.9239,
+      "step": 5079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845382122788515e-05,
+      "loss": 0.9229,
+      "step": 5080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842778738476502e-05,
+      "loss": 0.8715,
+      "step": 5081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840175144535594e-05,
+      "loss": 0.2365,
+      "step": 5082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837571341157772e-05,
+      "loss": 0.893,
+      "step": 5083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283496732853505e-05,
+      "loss": 0.9742,
+      "step": 5084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832363106859442e-05,
+      "loss": 0.8934,
+      "step": 5085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829758676322995e-05,
+      "loss": 0.8695,
+      "step": 5086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827154037117748e-05,
+      "loss": 0.9221,
+      "step": 5087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824549189435786e-05,
+      "loss": 0.958,
+      "step": 5088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821944133469182e-05,
+      "loss": 0.8644,
+      "step": 5089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819338869410035e-05,
+      "loss": 0.9098,
+      "step": 5090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816733397450465e-05,
+      "loss": 0.9337,
+      "step": 5091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2814127717782596e-05,
+      "loss": 0.9146,
+      "step": 5092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811521830598577e-05,
+      "loss": 0.9419,
+      "step": 5093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808915736090569e-05,
+      "loss": 0.9077,
+      "step": 5094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806309434450742e-05,
+      "loss": 0.7529,
+      "step": 5095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803702925871289e-05,
+      "loss": 0.911,
+      "step": 5096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280109621054442e-05,
+      "loss": 0.9479,
+      "step": 5097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798489288662355e-05,
+      "loss": 0.7997,
+      "step": 5098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279588216041733e-05,
+      "loss": 0.8301,
+      "step": 5099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793274826001596e-05,
+      "loss": 0.994,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790667285607418e-05,
+      "loss": 0.8337,
+      "step": 5101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788059539427083e-05,
+      "loss": 0.7768,
+      "step": 5102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785451587652886e-05,
+      "loss": 0.9832,
+      "step": 5103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782843430477141e-05,
+      "loss": 0.2412,
+      "step": 5104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278023506809217e-05,
+      "loss": 0.9367,
+      "step": 5105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277762650069032e-05,
+      "loss": 0.9056,
+      "step": 5106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775017728463954e-05,
+      "loss": 0.9595,
+      "step": 5107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772408751605438e-05,
+      "loss": 0.8798,
+      "step": 5108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2769799570307161e-05,
+      "loss": 0.9277,
+      "step": 5109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767190184761531e-05,
+      "loss": 0.8604,
+      "step": 5110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764580595160958e-05,
+      "loss": 0.2507,
+      "step": 5111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761970801697884e-05,
+      "loss": 0.8511,
+      "step": 5112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759360804564754e-05,
+      "loss": 0.9033,
+      "step": 5113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756750603954029e-05,
+      "loss": 0.2625,
+      "step": 5114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275414020005819e-05,
+      "loss": 0.9922,
+      "step": 5115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275152959306973e-05,
+      "loss": 0.9572,
+      "step": 5116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748918783181154e-05,
+      "loss": 0.9032,
+      "step": 5117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746307770584991e-05,
+      "loss": 0.8259,
+      "step": 5118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743696555473779e-05,
+      "loss": 0.8548,
+      "step": 5119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741085138040066e-05,
+      "loss": 1.0143,
+      "step": 5120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738473518476424e-05,
+      "loss": 0.9203,
+      "step": 5121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735861696975436e-05,
+      "loss": 0.8597,
+      "step": 5122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733249673729696e-05,
+      "loss": 0.797,
+      "step": 5123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730637448931824e-05,
+      "loss": 0.8745,
+      "step": 5124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728025022774442e-05,
+      "loss": 0.9392,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725412395450189e-05,
+      "loss": 0.9097,
+      "step": 5126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722799567151733e-05,
+      "loss": 0.8802,
+      "step": 5127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272018653807174e-05,
+      "loss": 0.8619,
+      "step": 5128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717573308402895e-05,
+      "loss": 0.9714,
+      "step": 5129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714959878337905e-05,
+      "loss": 0.8417,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712346248069483e-05,
+      "loss": 0.2635,
+      "step": 5131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709732417790357e-05,
+      "loss": 0.9114,
+      "step": 5132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707118387693282e-05,
+      "loss": 0.8999,
+      "step": 5133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704504157971014e-05,
+      "loss": 0.9029,
+      "step": 5134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701889728816326e-05,
+      "loss": 0.9525,
+      "step": 5135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699275100422011e-05,
+      "loss": 0.9462,
+      "step": 5136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696660272980874e-05,
+      "loss": 0.9532,
+      "step": 5137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694045246685734e-05,
+      "loss": 0.9161,
+      "step": 5138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691430021729425e-05,
+      "loss": 0.9308,
+      "step": 5139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688814598304797e-05,
+      "loss": 0.812,
+      "step": 5140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268619897660471e-05,
+      "loss": 0.8365,
+      "step": 5141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683583156822045e-05,
+      "loss": 0.8539,
+      "step": 5142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680967139149698e-05,
+      "loss": 0.9064,
+      "step": 5143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678350923780572e-05,
+      "loss": 0.8691,
+      "step": 5144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675734510907591e-05,
+      "loss": 0.959,
+      "step": 5145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673117900723688e-05,
+      "loss": 0.802,
+      "step": 5146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670501093421818e-05,
+      "loss": 0.8953,
+      "step": 5147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667884089194942e-05,
+      "loss": 0.9844,
+      "step": 5148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665266888236048e-05,
+      "loss": 0.9548,
+      "step": 5149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662649490738122e-05,
+      "loss": 0.8906,
+      "step": 5150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660031896894178e-05,
+      "loss": 0.9102,
+      "step": 5151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657414106897239e-05,
+      "loss": 0.9473,
+      "step": 5152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654796120940342e-05,
+      "loss": 0.9584,
+      "step": 5153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652177939216537e-05,
+      "loss": 0.2536,
+      "step": 5154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26495595619189e-05,
+      "loss": 1.006,
+      "step": 5155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646940989240504e-05,
+      "loss": 1.0132,
+      "step": 5156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644322221374443e-05,
+      "loss": 0.8689,
+      "step": 5157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641703258513833e-05,
+      "loss": 0.9343,
+      "step": 5158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26390841008518e-05,
+      "loss": 0.2402,
+      "step": 5159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636464748581476e-05,
+      "loss": 0.9191,
+      "step": 5160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263384520189602e-05,
+      "loss": 0.8369,
+      "step": 5161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631225460988597e-05,
+      "loss": 0.9413,
+      "step": 5162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2628605526052387e-05,
+      "loss": 0.9203,
+      "step": 5163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625985397280588e-05,
+      "loss": 0.946,
+      "step": 5164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623365074866418e-05,
+      "loss": 0.2885,
+      "step": 5165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620744559003088e-05,
+      "loss": 0.9153,
+      "step": 5166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618123849883845e-05,
+      "loss": 0.9316,
+      "step": 5167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615502947701943e-05,
+      "loss": 0.9004,
+      "step": 5168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612881852650645e-05,
+      "loss": 0.9432,
+      "step": 5169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261026056492324e-05,
+      "loss": 0.9694,
+      "step": 5170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607639084713016e-05,
+      "loss": 0.9283,
+      "step": 5171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605017412213285e-05,
+      "loss": 0.2792,
+      "step": 5172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602395547617376e-05,
+      "loss": 0.8708,
+      "step": 5173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599773491118622e-05,
+      "loss": 0.7863,
+      "step": 5174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597151242910382e-05,
+      "loss": 0.9149,
+      "step": 5175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594528803186017e-05,
+      "loss": 0.8798,
+      "step": 5176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591906172138908e-05,
+      "loss": 0.9494,
+      "step": 5177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589283349962452e-05,
+      "loss": 0.9597,
+      "step": 5178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586660336850058e-05,
+      "loss": 0.8915,
+      "step": 5179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258403713299515e-05,
+      "loss": 0.8789,
+      "step": 5180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581413738591166e-05,
+      "loss": 0.961,
+      "step": 5181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578790153831554e-05,
+      "loss": 0.8676,
+      "step": 5182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576166378909783e-05,
+      "loss": 0.8318,
+      "step": 5183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573542414019326e-05,
+      "loss": 0.9734,
+      "step": 5184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570918259353683e-05,
+      "loss": 0.8734,
+      "step": 5185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568293915106362e-05,
+      "loss": 0.9006,
+      "step": 5186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565669381470877e-05,
+      "loss": 0.8836,
+      "step": 5187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256304465864077e-05,
+      "loss": 0.9787,
+      "step": 5188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256041974680959e-05,
+      "loss": 0.8451,
+      "step": 5189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557794646170895e-05,
+      "loss": 0.871,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255516935691827e-05,
+      "loss": 0.2623,
+      "step": 5191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552543879245298e-05,
+      "loss": 0.8875,
+      "step": 5192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549918213345586e-05,
+      "loss": 0.9264,
+      "step": 5193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547292359412757e-05,
+      "loss": 0.9325,
+      "step": 5194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254466631764044e-05,
+      "loss": 0.8614,
+      "step": 5195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542040088222285e-05,
+      "loss": 0.8549,
+      "step": 5196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539413671351945e-05,
+      "loss": 0.3051,
+      "step": 5197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536787067223104e-05,
+      "loss": 0.8778,
+      "step": 5198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534160276029443e-05,
+      "loss": 1.0254,
+      "step": 5199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531533297964665e-05,
+      "loss": 0.8334,
+      "step": 5200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252890613322249e-05,
+      "loss": 0.2296,
+      "step": 5201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252627878199664e-05,
+      "loss": 0.8993,
+      "step": 5202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523651244480864e-05,
+      "loss": 0.8367,
+      "step": 5203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521023520868913e-05,
+      "loss": 0.3191,
+      "step": 5204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518395611354566e-05,
+      "loss": 0.9783,
+      "step": 5205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515767516131601e-05,
+      "loss": 0.8839,
+      "step": 5206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513139235393816e-05,
+      "loss": 0.8509,
+      "step": 5207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510510769335027e-05,
+      "loss": 0.8618,
+      "step": 5208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507882118149053e-05,
+      "loss": 0.9201,
+      "step": 5209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505253282029739e-05,
+      "loss": 0.8976,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502624261170934e-05,
+      "loss": 0.9466,
+      "step": 5211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499995055766507e-05,
+      "loss": 0.8751,
+      "step": 5212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497365666010335e-05,
+      "loss": 0.9158,
+      "step": 5213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494736092096311e-05,
+      "loss": 0.8486,
+      "step": 5214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492106334218344e-05,
+      "loss": 0.7827,
+      "step": 5215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489476392570354e-05,
+      "loss": 0.9349,
+      "step": 5216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486846267346278e-05,
+      "loss": 1.0209,
+      "step": 5217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484215958740058e-05,
+      "loss": 0.931,
+      "step": 5218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481585466945659e-05,
+      "loss": 0.9773,
+      "step": 5219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478954792157052e-05,
+      "loss": 0.8302,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476323934568228e-05,
+      "loss": 0.8242,
+      "step": 5221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473692894373195e-05,
+      "loss": 0.8883,
+      "step": 5222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471061671765955e-05,
+      "loss": 0.7988,
+      "step": 5223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468430266940546e-05,
+      "loss": 0.8927,
+      "step": 5224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465798680091007e-05,
+      "loss": 0.9133,
+      "step": 5225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463166911411393e-05,
+      "loss": 0.8049,
+      "step": 5226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460534961095774e-05,
+      "loss": 0.8657,
+      "step": 5227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457902829338231e-05,
+      "loss": 0.8434,
+      "step": 5228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245527051633286e-05,
+      "loss": 0.9616,
+      "step": 5229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245263802227377e-05,
+      "loss": 0.8746,
+      "step": 5230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450005347355086e-05,
+      "loss": 0.8974,
+      "step": 5231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447372491770941e-05,
+      "loss": 0.912,
+      "step": 5232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444739455715482e-05,
+      "loss": 0.8595,
+      "step": 5233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442106239382873e-05,
+      "loss": 0.9788,
+      "step": 5234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439472842967293e-05,
+      "loss": 0.9411,
+      "step": 5235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436839266662925e-05,
+      "loss": 0.9413,
+      "step": 5236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434205510663979e-05,
+      "loss": 0.944,
+      "step": 5237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431571575164665e-05,
+      "loss": 0.9711,
+      "step": 5238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428937460359207e-05,
+      "loss": 0.9577,
+      "step": 5239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426303166441855e-05,
+      "loss": 0.9449,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423668693606863e-05,
+      "loss": 0.2307,
+      "step": 5241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421034042048494e-05,
+      "loss": 0.8914,
+      "step": 5242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418399211961037e-05,
+      "loss": 0.8524,
+      "step": 5243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415764203538779e-05,
+      "loss": 0.865,
+      "step": 5244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413129016976032e-05,
+      "loss": 0.9706,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410493652467114e-05,
+      "loss": 0.9134,
+      "step": 5246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407858110206363e-05,
+      "loss": 1.0347,
+      "step": 5247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405222390388125e-05,
+      "loss": 0.9216,
+      "step": 5248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402586493206759e-05,
+      "loss": 0.8956,
+      "step": 5249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399950418856639e-05,
+      "loss": 0.9584,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239731416753215e-05,
+      "loss": 0.9128,
+      "step": 5251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239467773942769e-05,
+      "loss": 0.8879,
+      "step": 5252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239204113473768e-05,
+      "loss": 0.8969,
+      "step": 5253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389404353656533e-05,
+      "loss": 0.9059,
+      "step": 5254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386767396378696e-05,
+      "loss": 0.8613,
+      "step": 5255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384130263098619e-05,
+      "loss": 1.0333,
+      "step": 5256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381492954010765e-05,
+      "loss": 0.9696,
+      "step": 5257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378855469309612e-05,
+      "loss": 0.2753,
+      "step": 5258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237621780918965e-05,
+      "loss": 0.8021,
+      "step": 5259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373579973845384e-05,
+      "loss": 0.8899,
+      "step": 5260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370941963471325e-05,
+      "loss": 0.9,
+      "step": 5261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236830377826201e-05,
+      "loss": 0.9818,
+      "step": 5262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365665418411975e-05,
+      "loss": 0.9521,
+      "step": 5263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363026884115778e-05,
+      "loss": 0.8734,
+      "step": 5264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360388175567985e-05,
+      "loss": 0.8684,
+      "step": 5265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357749292963176e-05,
+      "loss": 0.8676,
+      "step": 5266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355110236495946e-05,
+      "loss": 0.8758,
+      "step": 5267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352471006360904e-05,
+      "loss": 0.8795,
+      "step": 5268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349831602752662e-05,
+      "loss": 0.8953,
+      "step": 5269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347192025865854e-05,
+      "loss": 0.8627,
+      "step": 5270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344552275895132e-05,
+      "loss": 0.9808,
+      "step": 5271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341912353035146e-05,
+      "loss": 0.957,
+      "step": 5272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339272257480563e-05,
+      "loss": 0.8458,
+      "step": 5273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336631989426078e-05,
+      "loss": 0.9548,
+      "step": 5274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333991549066378e-05,
+      "loss": 0.7704,
+      "step": 5275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331350936596166e-05,
+      "loss": 1.0139,
+      "step": 5276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328710152210177e-05,
+      "loss": 0.9623,
+      "step": 5277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326069196103134e-05,
+      "loss": 1.0492,
+      "step": 5278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323428068469788e-05,
+      "loss": 0.8739,
+      "step": 5279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320786769504895e-05,
+      "loss": 1.0233,
+      "step": 5280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231814529940323e-05,
+      "loss": 0.8947,
+      "step": 5281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315503658359573e-05,
+      "loss": 0.8714,
+      "step": 5282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312861846568724e-05,
+      "loss": 0.9,
+      "step": 5283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310219864225497e-05,
+      "loss": 0.8218,
+      "step": 5284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23075777115247e-05,
+      "loss": 0.8577,
+      "step": 5285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304935388661182e-05,
+      "loss": 0.9191,
+      "step": 5286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302292895829784e-05,
+      "loss": 0.9061,
+      "step": 5287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299650233225362e-05,
+      "loss": 0.814,
+      "step": 5288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297007401042799e-05,
+      "loss": 0.8506,
+      "step": 5289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229436439947697e-05,
+      "loss": 0.9055,
+      "step": 5290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229172122872277e-05,
+      "loss": 1.0147,
+      "step": 5291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289077888975121e-05,
+      "loss": 0.9673,
+      "step": 5292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286434380428935e-05,
+      "loss": 0.8205,
+      "step": 5293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283790703279149e-05,
+      "loss": 0.865,
+      "step": 5294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281146857720712e-05,
+      "loss": 0.9944,
+      "step": 5295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278502843948578e-05,
+      "loss": 0.9452,
+      "step": 5296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275858662157724e-05,
+      "loss": 0.803,
+      "step": 5297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273214312543134e-05,
+      "loss": 0.9631,
+      "step": 5298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270569795299804e-05,
+      "loss": 0.8539,
+      "step": 5299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226792511062274e-05,
+      "loss": 1.0252,
+      "step": 5300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265280258706969e-05,
+      "loss": 0.9445,
+      "step": 5301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262635239747518e-05,
+      "loss": 0.2342,
+      "step": 5302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225999005393944e-05,
+      "loss": 1.0111,
+      "step": 5303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257344701477785e-05,
+      "loss": 1.0403,
+      "step": 5304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225469918255763e-05,
+      "loss": 0.9311,
+      "step": 5305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252053497374052e-05,
+      "loss": 0.8925,
+      "step": 5306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249407646122154e-05,
+      "loss": 1.0563,
+      "step": 5307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246761628997037e-05,
+      "loss": 0.8946,
+      "step": 5308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244115446193822e-05,
+      "loss": 0.8851,
+      "step": 5309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241469097907644e-05,
+      "loss": 0.9507,
+      "step": 5310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238822584333643e-05,
+      "loss": 1.002,
+      "step": 5311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236175905666977e-05,
+      "loss": 0.9981,
+      "step": 5312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233529062102809e-05,
+      "loss": 0.7982,
+      "step": 5313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230882053836327e-05,
+      "loss": 0.951,
+      "step": 5314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228234881062724e-05,
+      "loss": 0.9057,
+      "step": 5315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225587543977198e-05,
+      "loss": 0.9164,
+      "step": 5316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222940042774971e-05,
+      "loss": 0.8767,
+      "step": 5317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220292377651271e-05,
+      "loss": 0.962,
+      "step": 5318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217644548801335e-05,
+      "loss": 0.9227,
+      "step": 5319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214996556420424e-05,
+      "loss": 0.9533,
+      "step": 5320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212348400703797e-05,
+      "loss": 0.8742,
+      "step": 5321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209700081846731e-05,
+      "loss": 0.9366,
+      "step": 5322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220705160004452e-05,
+      "loss": 0.9097,
+      "step": 5323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220440295549246e-05,
+      "loss": 0.9082,
+      "step": 5324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220175414838587e-05,
+      "loss": 0.9808,
+      "step": 5325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219910517892007e-05,
+      "loss": 0.8179,
+      "step": 5326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2196456047290402e-05,
+      "loss": 1.0225,
+      "step": 5327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193806753692208e-05,
+      "loss": 0.831,
+      "step": 5328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191157298320855e-05,
+      "loss": 0.9112,
+      "step": 5329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188507681371717e-05,
+      "loss": 0.9347,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185857903040174e-05,
+      "loss": 0.9044,
+      "step": 5331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183207963521629e-05,
+      "loss": 0.9193,
+      "step": 5332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180557863011484e-05,
+      "loss": 0.8741,
+      "step": 5333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177907601705163e-05,
+      "loss": 0.9147,
+      "step": 5334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175257179798104e-05,
+      "loss": 0.9809,
+      "step": 5335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217260659748574e-05,
+      "loss": 0.903,
+      "step": 5336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169955854963536e-05,
+      "loss": 0.8985,
+      "step": 5337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167304952426955e-05,
+      "loss": 0.8738,
+      "step": 5338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164653890071481e-05,
+      "loss": 0.8753,
+      "step": 5339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21620026680926e-05,
+      "loss": 0.9138,
+      "step": 5340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159351286685826e-05,
+      "loss": 1.0469,
+      "step": 5341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156699746046661e-05,
+      "loss": 0.8953,
+      "step": 5342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154048046370636e-05,
+      "loss": 1.0113,
+      "step": 5343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151396187853294e-05,
+      "loss": 0.8459,
+      "step": 5344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148744170690185e-05,
+      "loss": 0.902,
+      "step": 5345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146091995076867e-05,
+      "loss": 0.9202,
+      "step": 5346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143439661208916e-05,
+      "loss": 0.9642,
+      "step": 5347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140787169281913e-05,
+      "loss": 0.8985,
+      "step": 5348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213813451949146e-05,
+      "loss": 0.9074,
+      "step": 5349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135481712033165e-05,
+      "loss": 0.9453,
+      "step": 5350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132828747102651e-05,
+      "loss": 0.9276,
+      "step": 5351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213017562489554e-05,
+      "loss": 0.8952,
+      "step": 5352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127522345607485e-05,
+      "loss": 0.7733,
+      "step": 5353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124868909434138e-05,
+      "loss": 0.8996,
+      "step": 5354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122215316571163e-05,
+      "loss": 0.9087,
+      "step": 5355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119561567214246e-05,
+      "loss": 0.9523,
+      "step": 5356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116907661559068e-05,
+      "loss": 0.9345,
+      "step": 5357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114253599801331e-05,
+      "loss": 0.9859,
+      "step": 5358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111599382136754e-05,
+      "loss": 0.7947,
+      "step": 5359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108945008761057e-05,
+      "loss": 0.8539,
+      "step": 5360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106290479869976e-05,
+      "loss": 0.9418,
+      "step": 5361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103635795659258e-05,
+      "loss": 0.9289,
+      "step": 5362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100980956324663e-05,
+      "loss": 0.9058,
+      "step": 5363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098325962061957e-05,
+      "loss": 0.2698,
+      "step": 5364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095670813066929e-05,
+      "loss": 0.8546,
+      "step": 5365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093015509535368e-05,
+      "loss": 0.8064,
+      "step": 5366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090360051663075e-05,
+      "loss": 0.8522,
+      "step": 5367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208770443964587e-05,
+      "loss": 0.9193,
+      "step": 5368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208504867367958e-05,
+      "loss": 0.2183,
+      "step": 5369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082392753960042e-05,
+      "loss": 0.2476,
+      "step": 5370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079736680683102e-05,
+      "loss": 0.2634,
+      "step": 5371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077080454044632e-05,
+      "loss": 0.9404,
+      "step": 5372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074424074240491e-05,
+      "loss": 0.9405,
+      "step": 5373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071767541466572e-05,
+      "loss": 0.9081,
+      "step": 5374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069110855918767e-05,
+      "loss": 0.9148,
+      "step": 5375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066454017792984e-05,
+      "loss": 0.9329,
+      "step": 5376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063797027285135e-05,
+      "loss": 0.8546,
+      "step": 5377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061139884591158e-05,
+      "loss": 0.2728,
+      "step": 5378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058482589906984e-05,
+      "loss": 0.9051,
+      "step": 5379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055825143428566e-05,
+      "loss": 0.9143,
+      "step": 5380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205316754535187e-05,
+      "loss": 0.8814,
+      "step": 5381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050509795872866e-05,
+      "loss": 0.8469,
+      "step": 5382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047851895187544e-05,
+      "loss": 0.8856,
+      "step": 5383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045193843491894e-05,
+      "loss": 0.976,
+      "step": 5384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042535640981925e-05,
+      "loss": 0.9128,
+      "step": 5385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039877287853653e-05,
+      "loss": 0.836,
+      "step": 5386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037218784303113e-05,
+      "loss": 0.8475,
+      "step": 5387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034560130526341e-05,
+      "loss": 1.019,
+      "step": 5388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031901326719387e-05,
+      "loss": 0.9271,
+      "step": 5389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029242373078316e-05,
+      "loss": 0.9342,
+      "step": 5390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20265832697992e-05,
+      "loss": 0.2769,
+      "step": 5391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023924017078126e-05,
+      "loss": 0.8795,
+      "step": 5392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202126461511119e-05,
+      "loss": 0.9052,
+      "step": 5393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018605064094493e-05,
+      "loss": 0.9175,
+      "step": 5394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015945364224155e-05,
+      "loss": 0.8506,
+      "step": 5395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201328551569631e-05,
+      "loss": 0.9059,
+      "step": 5396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010625518707092e-05,
+      "loss": 0.8824,
+      "step": 5397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200796537345265e-05,
+      "loss": 0.2652,
+      "step": 5398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005305080129148e-05,
+      "loss": 0.863,
+      "step": 5399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200264463893276e-05,
+      "loss": 0.3031,
+      "step": 5400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999984050059666e-05,
+      "loss": 0.9953,
+      "step": 5401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997323313706063e-05,
+      "loss": 0.7899,
+      "step": 5402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994662430068153e-05,
+      "loss": 0.9434,
+      "step": 5403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992001399342152e-05,
+      "loss": 0.9651,
+      "step": 5404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989340221724288e-05,
+      "loss": 1.0208,
+      "step": 5405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986678897410799e-05,
+      "loss": 0.9385,
+      "step": 5406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984017426597931e-05,
+      "loss": 0.9207,
+      "step": 5407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198135580948195e-05,
+      "loss": 0.8676,
+      "step": 5408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978694046259117e-05,
+      "loss": 0.9354,
+      "step": 5409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976032137125713e-05,
+      "loss": 0.8728,
+      "step": 5410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973370082278035e-05,
+      "loss": 0.9942,
+      "step": 5411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970707881912385e-05,
+      "loss": 0.9117,
+      "step": 5412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968045536225072e-05,
+      "loss": 0.9351,
+      "step": 5413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196538304541242e-05,
+      "loss": 0.8807,
+      "step": 5414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962720409670768e-05,
+      "loss": 0.8771,
+      "step": 5415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960057629196453e-05,
+      "loss": 0.8549,
+      "step": 5416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957394704185837e-05,
+      "loss": 0.9398,
+      "step": 5417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954731634835287e-05,
+      "loss": 0.2827,
+      "step": 5418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952068421341174e-05,
+      "loss": 0.907,
+      "step": 5419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949405063899893e-05,
+      "loss": 0.9641,
+      "step": 5420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946741562707835e-05,
+      "loss": 0.9134,
+      "step": 5421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944077917961415e-05,
+      "loss": 0.2845,
+      "step": 5422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941414129857048e-05,
+      "loss": 0.924,
+      "step": 5423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938750198591167e-05,
+      "loss": 0.8746,
+      "step": 5424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193608612436021e-05,
+      "loss": 0.8547,
+      "step": 5425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933421907360627e-05,
+      "loss": 0.8429,
+      "step": 5426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1930757547788885e-05,
+      "loss": 0.8741,
+      "step": 5427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928093045841453e-05,
+      "loss": 0.809,
+      "step": 5428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925428401714814e-05,
+      "loss": 0.8593,
+      "step": 5429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192276361560546e-05,
+      "loss": 0.9767,
+      "step": 5430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920098687709893e-05,
+      "loss": 0.9527,
+      "step": 5431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917433618224629e-05,
+      "loss": 0.8742,
+      "step": 5432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19147684073462e-05,
+      "loss": 0.9347,
+      "step": 5433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912103055271125e-05,
+      "loss": 0.9301,
+      "step": 5434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909437562195963e-05,
+      "loss": 0.7955,
+      "step": 5435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906771928317265e-05,
+      "loss": 0.8051,
+      "step": 5436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904106153831593e-05,
+      "loss": 0.8799,
+      "step": 5437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901440238935528e-05,
+      "loss": 0.9457,
+      "step": 5438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898774183825661e-05,
+      "loss": 0.8264,
+      "step": 5439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896107988698581e-05,
+      "loss": 0.8236,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893441653750897e-05,
+      "loss": 0.8294,
+      "step": 5441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189077517917923e-05,
+      "loss": 0.8472,
+      "step": 5442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888108565180207e-05,
+      "loss": 0.955,
+      "step": 5443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885441811950467e-05,
+      "loss": 0.8595,
+      "step": 5444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882774919686654e-05,
+      "loss": 0.8601,
+      "step": 5445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880107888585435e-05,
+      "loss": 0.9508,
+      "step": 5446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877440718843469e-05,
+      "loss": 0.8044,
+      "step": 5447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874773410657446e-05,
+      "loss": 0.8896,
+      "step": 5448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187210596422405e-05,
+      "loss": 0.8836,
+      "step": 5449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869438379739977e-05,
+      "loss": 0.9568,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866770657401944e-05,
+      "loss": 0.8635,
+      "step": 5451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864102797406668e-05,
+      "loss": 0.9336,
+      "step": 5452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186143479995088e-05,
+      "loss": 0.9011,
+      "step": 5453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858766665231323e-05,
+      "loss": 1.0244,
+      "step": 5454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185609839344474e-05,
+      "loss": 0.9927,
+      "step": 5455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853429984787894e-05,
+      "loss": 0.7954,
+      "step": 5456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850761439457563e-05,
+      "loss": 0.8911,
+      "step": 5457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184809275765052e-05,
+      "loss": 0.956,
+      "step": 5458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845423939563558e-05,
+      "loss": 0.8419,
+      "step": 5459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184275498539348e-05,
+      "loss": 0.9106,
+      "step": 5460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840085895337094e-05,
+      "loss": 0.8378,
+      "step": 5461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183741666959122e-05,
+      "loss": 0.8131,
+      "step": 5462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834747308352694e-05,
+      "loss": 0.8601,
+      "step": 5463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832077811818357e-05,
+      "loss": 0.8459,
+      "step": 5464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829408180185049e-05,
+      "loss": 0.2761,
+      "step": 5465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826738413649643e-05,
+      "loss": 0.8983,
+      "step": 5466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824068512409005e-05,
+      "loss": 0.7663,
+      "step": 5467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821398476660018e-05,
+      "loss": 0.8365,
+      "step": 5468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181872830659957e-05,
+      "loss": 0.9324,
+      "step": 5469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816058002424566e-05,
+      "loss": 0.906,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813387564331907e-05,
+      "loss": 0.9457,
+      "step": 5471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810716992518522e-05,
+      "loss": 0.8722,
+      "step": 5472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180804628718134e-05,
+      "loss": 0.8482,
+      "step": 5473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805375448517302e-05,
+      "loss": 0.8434,
+      "step": 5474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802704476723352e-05,
+      "loss": 0.975,
+      "step": 5475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800033371996456e-05,
+      "loss": 0.8858,
+      "step": 5476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797362134533577e-05,
+      "loss": 0.8479,
+      "step": 5477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794690764531703e-05,
+      "loss": 0.9276,
+      "step": 5478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792019262187816e-05,
+      "loss": 0.8974,
+      "step": 5479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789347627698917e-05,
+      "loss": 0.8914,
+      "step": 5480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786675861262013e-05,
+      "loss": 0.9212,
+      "step": 5481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178400396307413e-05,
+      "loss": 0.813,
+      "step": 5482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781331933332286e-05,
+      "loss": 0.8431,
+      "step": 5483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177865977223352e-05,
+      "loss": 0.2718,
+      "step": 5484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775987479974889e-05,
+      "loss": 0.8731,
+      "step": 5485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773315056753436e-05,
+      "loss": 0.2498,
+      "step": 5486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177064250276624e-05,
+      "loss": 0.8491,
+      "step": 5487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767969818210373e-05,
+      "loss": 0.9627,
+      "step": 5488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765297003282918e-05,
+      "loss": 0.9479,
+      "step": 5489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762624058180974e-05,
+      "loss": 0.9637,
+      "step": 5490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759950983101644e-05,
+      "loss": 0.9003,
+      "step": 5491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757277778242046e-05,
+      "loss": 0.9167,
+      "step": 5492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17546044437993e-05,
+      "loss": 0.9427,
+      "step": 5493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751930979970546e-05,
+      "loss": 0.8407,
+      "step": 5494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749257386952921e-05,
+      "loss": 0.9507,
+      "step": 5495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746583664943583e-05,
+      "loss": 0.7549,
+      "step": 5496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174390981413969e-05,
+      "loss": 0.8652,
+      "step": 5497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741235834738421e-05,
+      "loss": 0.9259,
+      "step": 5498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738561726936948e-05,
+      "loss": 0.8414,
+      "step": 5499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173588749093247e-05,
+      "loss": 0.902,
+      "step": 5500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733213126922185e-05,
+      "loss": 0.9158,
+      "step": 5501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17305386351033e-05,
+      "loss": 0.8923,
+      "step": 5502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727864015673033e-05,
+      "loss": 0.8979,
+      "step": 5503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725189268828622e-05,
+      "loss": 0.9142,
+      "step": 5504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722514394767298e-05,
+      "loss": 0.9567,
+      "step": 5505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171983939368631e-05,
+      "loss": 0.8889,
+      "step": 5506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717164265782915e-05,
+      "loss": 0.9564,
+      "step": 5507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714489011254374e-05,
+      "loss": 0.9471,
+      "step": 5508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711813630297973e-05,
+      "loss": 0.985,
+      "step": 5509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709138123110987e-05,
+      "loss": 0.9172,
+      "step": 5510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706462489890717e-05,
+      "loss": 0.9641,
+      "step": 5511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703786730834463e-05,
+      "loss": 0.9744,
+      "step": 5512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701110846139539e-05,
+      "loss": 0.8934,
+      "step": 5513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698434836003263e-05,
+      "loss": 0.8691,
+      "step": 5514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695758700622975e-05,
+      "loss": 0.8961,
+      "step": 5515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693082440196008e-05,
+      "loss": 0.9163,
+      "step": 5516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690406054919711e-05,
+      "loss": 0.9151,
+      "step": 5517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168772954499145e-05,
+      "loss": 0.9507,
+      "step": 5518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685052910608587e-05,
+      "loss": 0.923,
+      "step": 5519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16823761519685e-05,
+      "loss": 0.9143,
+      "step": 5520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167969926926858e-05,
+      "loss": 0.8632,
+      "step": 5521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677022262706219e-05,
+      "loss": 0.8958,
+      "step": 5522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674345132478817e-05,
+      "loss": 0.9065,
+      "step": 5523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671667878783796e-05,
+      "loss": 0.8584,
+      "step": 5524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668990501818578e-05,
+      "loss": 0.8572,
+      "step": 5525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166631300178059e-05,
+      "loss": 0.9542,
+      "step": 5526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663635378867275e-05,
+      "loss": 0.8642,
+      "step": 5527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660957633276087e-05,
+      "loss": 0.8677,
+      "step": 5528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658279765204478e-05,
+      "loss": 0.862,
+      "step": 5529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655601774849923e-05,
+      "loss": 0.9171,
+      "step": 5530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652923662409898e-05,
+      "loss": 0.9409,
+      "step": 5531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650245428081885e-05,
+      "loss": 0.8573,
+      "step": 5532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647567072063386e-05,
+      "loss": 0.9048,
+      "step": 5533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16448885945519e-05,
+      "loss": 0.9203,
+      "step": 5534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642209995744943e-05,
+      "loss": 0.9448,
+      "step": 5535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163953127584003e-05,
+      "loss": 0.9117,
+      "step": 5536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636852435034704e-05,
+      "loss": 0.2485,
+      "step": 5537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634173473526496e-05,
+      "loss": 0.8976,
+      "step": 5538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631494391512959e-05,
+      "loss": 0.9409,
+      "step": 5539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628815189191652e-05,
+      "loss": 0.9157,
+      "step": 5540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626135866760136e-05,
+      "loss": 0.9005,
+      "step": 5541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623456424415988e-05,
+      "loss": 0.9271,
+      "step": 5542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620776862356798e-05,
+      "loss": 0.9671,
+      "step": 5543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618097180780153e-05,
+      "loss": 0.8822,
+      "step": 5544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615417379883656e-05,
+      "loss": 1.0927,
+      "step": 5545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161273745986492e-05,
+      "loss": 0.9484,
+      "step": 5546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610057420921566e-05,
+      "loss": 0.2639,
+      "step": 5547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1607377263251217e-05,
+      "loss": 0.9242,
+      "step": 5548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604696987051513e-05,
+      "loss": 0.8792,
+      "step": 5549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602016592520103e-05,
+      "loss": 0.8987,
+      "step": 5550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599336079854634e-05,
+      "loss": 0.9517,
+      "step": 5551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596655449252776e-05,
+      "loss": 0.8645,
+      "step": 5552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593974700912198e-05,
+      "loss": 0.8678,
+      "step": 5553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159129383503058e-05,
+      "loss": 0.9255,
+      "step": 5554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588612851805614e-05,
+      "loss": 0.895,
+      "step": 5555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1585931751434996e-05,
+      "loss": 0.9391,
+      "step": 5556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583250534116433e-05,
+      "loss": 0.7971,
+      "step": 5557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158056920004764e-05,
+      "loss": 0.9742,
+      "step": 5558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577887749426341e-05,
+      "loss": 0.99,
+      "step": 5559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575206182450267e-05,
+      "loss": 0.2507,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572524499317164e-05,
+      "loss": 0.8058,
+      "step": 5561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569842700224779e-05,
+      "loss": 0.9819,
+      "step": 5562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567160785370865e-05,
+      "loss": 0.8477,
+      "step": 5563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564478754953194e-05,
+      "loss": 0.7942,
+      "step": 5564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561796609169542e-05,
+      "loss": 0.2598,
+      "step": 5565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559114348217688e-05,
+      "loss": 0.8702,
+      "step": 5566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155643197229543e-05,
+      "loss": 0.8955,
+      "step": 5567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553749481600565e-05,
+      "loss": 0.9863,
+      "step": 5568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551066876330901e-05,
+      "loss": 1.0057,
+      "step": 5569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548384156684259e-05,
+      "loss": 0.9422,
+      "step": 5570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545701322858463e-05,
+      "loss": 0.9018,
+      "step": 5571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543018375051344e-05,
+      "loss": 0.8783,
+      "step": 5572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540335313460755e-05,
+      "loss": 0.9221,
+      "step": 5573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537652138284539e-05,
+      "loss": 0.869,
+      "step": 5574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534968849720554e-05,
+      "loss": 0.894,
+      "step": 5575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532285447966676e-05,
+      "loss": 0.8716,
+      "step": 5576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529601933220774e-05,
+      "loss": 0.8733,
+      "step": 5577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526918305680737e-05,
+      "loss": 0.9193,
+      "step": 5578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524234565544455e-05,
+      "loss": 0.2431,
+      "step": 5579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521550713009831e-05,
+      "loss": 0.2696,
+      "step": 5580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151886674827477e-05,
+      "loss": 0.9196,
+      "step": 5581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15161826715372e-05,
+      "loss": 1.0383,
+      "step": 5582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151349848299504e-05,
+      "loss": 0.8481,
+      "step": 5583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510814182846224e-05,
+      "loss": 0.7961,
+      "step": 5584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508129771288696e-05,
+      "loss": 0.8185,
+      "step": 5585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505445248520407e-05,
+      "loss": 0.8433,
+      "step": 5586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502760614739316e-05,
+      "loss": 0.934,
+      "step": 5587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150007587014339e-05,
+      "loss": 0.8257,
+      "step": 5588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497391014930604e-05,
+      "loss": 0.8959,
+      "step": 5589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494706049298939e-05,
+      "loss": 0.8307,
+      "step": 5590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492020973446391e-05,
+      "loss": 0.9091,
+      "step": 5591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148933578757096e-05,
+      "loss": 0.9765,
+      "step": 5592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148665049187065e-05,
+      "loss": 0.8496,
+      "step": 5593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483965086543479e-05,
+      "loss": 0.831,
+      "step": 5594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148127957178747e-05,
+      "loss": 0.9723,
+      "step": 5595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478593947800659e-05,
+      "loss": 0.9557,
+      "step": 5596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147590821478108e-05,
+      "loss": 0.8841,
+      "step": 5597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473222372926784e-05,
+      "loss": 0.9023,
+      "step": 5598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470536422435829e-05,
+      "loss": 0.9021,
+      "step": 5599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146785036350628e-05,
+      "loss": 0.8861,
+      "step": 5600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465164196336208e-05,
+      "loss": 0.9007,
+      "step": 5601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146247792112369e-05,
+      "loss": 0.9383,
+      "step": 5602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459791538066813e-05,
+      "loss": 0.8747,
+      "step": 5603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457105047363683e-05,
+      "loss": 0.9328,
+      "step": 5604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454418449212396e-05,
+      "loss": 0.8631,
+      "step": 5605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451731743811063e-05,
+      "loss": 0.9416,
+      "step": 5606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449044931357808e-05,
+      "loss": 0.9337,
+      "step": 5607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144635801205076e-05,
+      "loss": 0.8439,
+      "step": 5608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144367098608805e-05,
+      "loss": 0.8603,
+      "step": 5609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440983853667827e-05,
+      "loss": 0.7864,
+      "step": 5610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438296614988235e-05,
+      "loss": 0.9514,
+      "step": 5611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435609270247438e-05,
+      "loss": 0.8674,
+      "step": 5612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432921819643606e-05,
+      "loss": 0.8426,
+      "step": 5613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430234263374911e-05,
+      "loss": 0.8963,
+      "step": 5614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142754660163953e-05,
+      "loss": 0.9253,
+      "step": 5615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424858834635661e-05,
+      "loss": 0.7275,
+      "step": 5616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14221709625615e-05,
+      "loss": 0.9478,
+      "step": 5617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419482985615251e-05,
+      "loss": 0.9566,
+      "step": 5618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416794903995134e-05,
+      "loss": 0.8784,
+      "step": 5619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1414106717899365e-05,
+      "loss": 0.9456,
+      "step": 5620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411418427526168e-05,
+      "loss": 0.8986,
+      "step": 5621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408730033073792e-05,
+      "loss": 0.8402,
+      "step": 5622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406041534740473e-05,
+      "loss": 0.8951,
+      "step": 5623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403352932724465e-05,
+      "loss": 0.8273,
+      "step": 5624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400664227224029e-05,
+      "loss": 0.9257,
+      "step": 5625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139797541843743e-05,
+      "loss": 0.9429,
+      "step": 5626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395286506562946e-05,
+      "loss": 0.821,
+      "step": 5627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392597491798857e-05,
+      "loss": 0.8368,
+      "step": 5628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389908374343454e-05,
+      "loss": 0.8959,
+      "step": 5629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387219154395037e-05,
+      "loss": 0.9894,
+      "step": 5630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384529832151912e-05,
+      "loss": 0.8659,
+      "step": 5631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381840407812386e-05,
+      "loss": 0.9066,
+      "step": 5632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137915088157478e-05,
+      "loss": 0.7525,
+      "step": 5633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137646125363743e-05,
+      "loss": 0.9138,
+      "step": 5634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373771524198664e-05,
+      "loss": 0.9284,
+      "step": 5635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371081693456828e-05,
+      "loss": 0.9134,
+      "step": 5636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368391761610269e-05,
+      "loss": 0.9336,
+      "step": 5637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136570172885735e-05,
+      "loss": 0.9403,
+      "step": 5638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363011595396431e-05,
+      "loss": 0.8654,
+      "step": 5639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360321361425892e-05,
+      "loss": 0.8151,
+      "step": 5640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357631027144107e-05,
+      "loss": 0.8921,
+      "step": 5641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354940592749462e-05,
+      "loss": 0.8595,
+      "step": 5642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352250058440357e-05,
+      "loss": 0.8978,
+      "step": 5643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349559424415193e-05,
+      "loss": 0.9138,
+      "step": 5644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346868690872376e-05,
+      "loss": 0.8248,
+      "step": 5645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344177858010328e-05,
+      "loss": 0.9493,
+      "step": 5646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341486926027472e-05,
+      "loss": 0.2481,
+      "step": 5647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338795895122235e-05,
+      "loss": 0.8392,
+      "step": 5648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133610476549306e-05,
+      "loss": 0.9727,
+      "step": 5649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333413537338397e-05,
+      "loss": 0.8879,
+      "step": 5650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330722210856692e-05,
+      "loss": 0.9503,
+      "step": 5651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328030786246411e-05,
+      "loss": 0.9413,
+      "step": 5652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325339263706017e-05,
+      "loss": 0.995,
+      "step": 5653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132264764343399e-05,
+      "loss": 0.9444,
+      "step": 5654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131995592562881e-05,
+      "loss": 0.8713,
+      "step": 5655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317264110488966e-05,
+      "loss": 0.8412,
+      "step": 5656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314572198212954e-05,
+      "loss": 0.8762,
+      "step": 5657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311880188999281e-05,
+      "loss": 0.869,
+      "step": 5658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309188083046457e-05,
+      "loss": 0.8937,
+      "step": 5659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306495880553e-05,
+      "loss": 0.9056,
+      "step": 5660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303803581717432e-05,
+      "loss": 0.9391,
+      "step": 5661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301111186738289e-05,
+      "loss": 0.7875,
+      "step": 5662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298418695814113e-05,
+      "loss": 0.9361,
+      "step": 5663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295726109143443e-05,
+      "loss": 0.2588,
+      "step": 5664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293033426924838e-05,
+      "loss": 0.9019,
+      "step": 5665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290340649356859e-05,
+      "loss": 0.9251,
+      "step": 5666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128764777663807e-05,
+      "loss": 0.8175,
+      "step": 5667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284954808967053e-05,
+      "loss": 0.8465,
+      "step": 5668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128226174654238e-05,
+      "loss": 1.0245,
+      "step": 5669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279568589562643e-05,
+      "loss": 0.8478,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276875338226445e-05,
+      "loss": 0.9358,
+      "step": 5671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274181992732378e-05,
+      "loss": 0.8203,
+      "step": 5672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271488553279056e-05,
+      "loss": 0.8319,
+      "step": 5673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268795020065097e-05,
+      "loss": 0.8525,
+      "step": 5674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266101393289124e-05,
+      "loss": 0.9326,
+      "step": 5675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263407673149764e-05,
+      "loss": 0.9308,
+      "step": 5676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126071385984566e-05,
+      "loss": 0.9055,
+      "step": 5677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125801995357545e-05,
+      "loss": 0.9094,
+      "step": 5678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255325954537788e-05,
+      "loss": 0.8529,
+      "step": 5679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252631862931332e-05,
+      "loss": 0.866,
+      "step": 5680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249937678954748e-05,
+      "loss": 0.9605,
+      "step": 5681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12472434028067e-05,
+      "loss": 0.8425,
+      "step": 5682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244549034685875e-05,
+      "loss": 0.8867,
+      "step": 5683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241854574790953e-05,
+      "loss": 0.9756,
+      "step": 5684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239160023320628e-05,
+      "loss": 0.9431,
+      "step": 5685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12364653804736e-05,
+      "loss": 0.9089,
+      "step": 5686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123377064644857e-05,
+      "loss": 0.8701,
+      "step": 5687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231075821444251e-05,
+      "loss": 0.885,
+      "step": 5688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228380905659365e-05,
+      "loss": 0.7905,
+      "step": 5689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225685899292631e-05,
+      "loss": 0.8324,
+      "step": 5690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222990802542787e-05,
+      "loss": 0.9,
+      "step": 5691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220295615608572e-05,
+      "loss": 0.904,
+      "step": 5692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217600338688729e-05,
+      "loss": 0.9218,
+      "step": 5693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214904971982005e-05,
+      "loss": 0.8538,
+      "step": 5694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212209515687167e-05,
+      "loss": 0.9078,
+      "step": 5695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209513970002979e-05,
+      "loss": 0.915,
+      "step": 5696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120681833512821e-05,
+      "loss": 0.8454,
+      "step": 5697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204122611261637e-05,
+      "loss": 0.8858,
+      "step": 5698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201426798602046e-05,
+      "loss": 0.8369,
+      "step": 5699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198730897348231e-05,
+      "loss": 0.9767,
+      "step": 5700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196034907698989e-05,
+      "loss": 0.8844,
+      "step": 5701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193338829853126e-05,
+      "loss": 0.9119,
+      "step": 5702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1190642664009448e-05,
+      "loss": 0.8753,
+      "step": 5703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118794641036678e-05,
+      "loss": 0.8707,
+      "step": 5704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185250069123939e-05,
+      "loss": 0.9755,
+      "step": 5705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182553640479757e-05,
+      "loss": 0.9378,
+      "step": 5706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179857124633073e-05,
+      "loss": 0.8517,
+      "step": 5707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117716052178273e-05,
+      "loss": 0.9155,
+      "step": 5708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174463832127576e-05,
+      "loss": 0.9438,
+      "step": 5709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117176705586647e-05,
+      "loss": 0.8732,
+      "step": 5710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169070193198272e-05,
+      "loss": 0.8695,
+      "step": 5711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1166373244321852e-05,
+      "loss": 0.878,
+      "step": 5712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163676209436088e-05,
+      "loss": 0.2561,
+      "step": 5713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160979088739857e-05,
+      "loss": 0.8372,
+      "step": 5714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115828188243205e-05,
+      "loss": 0.9583,
+      "step": 5715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155584590711555e-05,
+      "loss": 0.8287,
+      "step": 5716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152887213777285e-05,
+      "loss": 0.8145,
+      "step": 5717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150189751828133e-05,
+      "loss": 0.7701,
+      "step": 5718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147492205063022e-05,
+      "loss": 0.2379,
+      "step": 5719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144794573680869e-05,
+      "loss": 0.9503,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142096857880595e-05,
+      "loss": 0.8847,
+      "step": 5721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139399057861138e-05,
+      "loss": 0.9638,
+      "step": 5722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136701173821434e-05,
+      "loss": 0.9248,
+      "step": 5723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134003205960424e-05,
+      "loss": 0.9158,
+      "step": 5724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1131305154477061e-05,
+      "loss": 0.8894,
+      "step": 5725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128607019570306e-05,
+      "loss": 0.894,
+      "step": 5726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125908801439113e-05,
+      "loss": 0.8437,
+      "step": 5727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123210500282458e-05,
+      "loss": 0.864,
+      "step": 5728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120512116299313e-05,
+      "loss": 0.7751,
+      "step": 5729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111781364968866e-05,
+      "loss": 0.7725,
+      "step": 5730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115115100649484e-05,
+      "loss": 0.9108,
+      "step": 5731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111241646938078e-05,
+      "loss": 0.9403,
+      "step": 5732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110971775608155e-05,
+      "loss": 0.874,
+      "step": 5733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107018960950793e-05,
+      "loss": 0.3047,
+      "step": 5734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104320084187527e-05,
+      "loss": 0.2742,
+      "step": 5735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101621125990767e-05,
+      "loss": 1.0291,
+      "step": 5736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098922086559532e-05,
+      "loss": 0.9069,
+      "step": 5737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109622296609286e-05,
+      "loss": 0.794,
+      "step": 5738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109352376478978e-05,
+      "loss": 0.9011,
+      "step": 5739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090824482849334e-05,
+      "loss": 0.7863,
+      "step": 5740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088125120470572e-05,
+      "loss": 0.8342,
+      "step": 5741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1085425677852547e-05,
+      "loss": 0.9733,
+      "step": 5742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082726155194316e-05,
+      "loss": 0.8677,
+      "step": 5743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080026552694944e-05,
+      "loss": 0.7331,
+      "step": 5744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077326870553506e-05,
+      "loss": 0.88,
+      "step": 5745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074627108969072e-05,
+      "loss": 0.9187,
+      "step": 5746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107192726814073e-05,
+      "loss": 0.2516,
+      "step": 5747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106922734826757e-05,
+      "loss": 0.9723,
+      "step": 5748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066527349548682e-05,
+      "loss": 0.8854,
+      "step": 5749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063827272183165e-05,
+      "loss": 0.9001,
+      "step": 5750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061127116370132e-05,
+      "loss": 0.8598,
+      "step": 5751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058426882308687e-05,
+      "loss": 0.9017,
+      "step": 5752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055726570197953e-05,
+      "loss": 0.2612,
+      "step": 5753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105302618023705e-05,
+      "loss": 0.8233,
+      "step": 5754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050325712625107e-05,
+      "loss": 0.9334,
+      "step": 5755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047625167561263e-05,
+      "loss": 0.262,
+      "step": 5756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044924545244655e-05,
+      "loss": 0.9365,
+      "step": 5757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042223845874428e-05,
+      "loss": 0.8969,
+      "step": 5758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039523069649738e-05,
+      "loss": 0.917,
+      "step": 5759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103682221676974e-05,
+      "loss": 0.8211,
+      "step": 5760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034121287433594e-05,
+      "loss": 0.2756,
+      "step": 5761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031420281840475e-05,
+      "loss": 1.0439,
+      "step": 5762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028719200189553e-05,
+      "loss": 0.8544,
+      "step": 5763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026018042680012e-05,
+      "loss": 0.9057,
+      "step": 5764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023316809511033e-05,
+      "loss": 0.9186,
+      "step": 5765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102061550088181e-05,
+      "loss": 0.921,
+      "step": 5766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101791411699154e-05,
+      "loss": 0.9281,
+      "step": 5767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015212658039421e-05,
+      "loss": 0.9196,
+      "step": 5768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101251112422467e-05,
+      "loss": 0.8642,
+      "step": 5769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009809515746488e-05,
+      "loss": 0.8471,
+      "step": 5770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007107832804104e-05,
+      "loss": 0.8825,
+      "step": 5771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100440607559674e-05,
+      "loss": 0.8272,
+      "step": 5772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001704244323626e-05,
+      "loss": 0.8681,
+      "step": 5773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999002339183993e-05,
+      "loss": 0.9596,
+      "step": 5774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996300360377088e-05,
+      "loss": 0.8676,
+      "step": 5775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993598308102152e-05,
+      "loss": 0.9083,
+      "step": 5776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990896182558437e-05,
+      "loss": 0.9765,
+      "step": 5777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988193983945203e-05,
+      "loss": 0.8816,
+      "step": 5778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985491712461713e-05,
+      "loss": 0.9152,
+      "step": 5779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982789368307231e-05,
+      "loss": 0.8609,
+      "step": 5780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980086951681036e-05,
+      "loss": 0.7992,
+      "step": 5781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977384462782399e-05,
+      "loss": 0.8945,
+      "step": 5782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974681901810607e-05,
+      "loss": 0.8736,
+      "step": 5783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971979268964952e-05,
+      "loss": 0.8323,
+      "step": 5784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969276564444724e-05,
+      "loss": 0.3141,
+      "step": 5785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966573788449223e-05,
+      "loss": 0.783,
+      "step": 5786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963870941177759e-05,
+      "loss": 0.9139,
+      "step": 5787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961168022829635e-05,
+      "loss": 0.8457,
+      "step": 5788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958465033604173e-05,
+      "loss": 0.8769,
+      "step": 5789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955761973700693e-05,
+      "loss": 0.8201,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953058843318517e-05,
+      "loss": 0.8425,
+      "step": 5791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950355642656975e-05,
+      "loss": 0.9568,
+      "step": 5792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094765237191541e-05,
+      "loss": 0.9865,
+      "step": 5793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094494903129316e-05,
+      "loss": 0.9019,
+      "step": 5794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094224562098957e-05,
+      "loss": 0.8891,
+      "step": 5795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939542141203993e-05,
+      "loss": 0.8753,
+      "step": 5796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0936838592135787e-05,
+      "loss": 0.8643,
+      "step": 5797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934134973984311e-05,
+      "loss": 0.919,
+      "step": 5798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931431286948936e-05,
+      "loss": 0.9388,
+      "step": 5799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928727531229037e-05,
+      "loss": 0.8998,
+      "step": 5800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926023707023981e-05,
+      "loss": 0.9016,
+      "step": 5801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923319814533155e-05,
+      "loss": 0.9349,
+      "step": 5802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920615853955949e-05,
+      "loss": 0.8166,
+      "step": 5803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917911825491752e-05,
+      "loss": 0.8884,
+      "step": 5804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915207729339967e-05,
+      "loss": 0.8931,
+      "step": 5805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091250356569999e-05,
+      "loss": 0.9062,
+      "step": 5806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909799334771227e-05,
+      "loss": 0.9001,
+      "step": 5807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907095036753097e-05,
+      "loss": 0.8732,
+      "step": 5808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0904390671845013e-05,
+      "loss": 0.8452,
+      "step": 5809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901686240246399e-05,
+      "loss": 0.2635,
+      "step": 5810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089898174215668e-05,
+      "loss": 0.8161,
+      "step": 5811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089627717777529e-05,
+      "loss": 0.8641,
+      "step": 5812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893572547301662e-05,
+      "loss": 0.8978,
+      "step": 5813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890867850935246e-05,
+      "loss": 0.8931,
+      "step": 5814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888163088875482e-05,
+      "loss": 0.9124,
+      "step": 5815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088545826132182e-05,
+      "loss": 0.8983,
+      "step": 5816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088275336847372e-05,
+      "loss": 0.9297,
+      "step": 5817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880048410530644e-05,
+      "loss": 0.873,
+      "step": 5818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877343387692055e-05,
+      "loss": 0.9545,
+      "step": 5819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874638300157422e-05,
+      "loss": 0.92,
+      "step": 5820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871933148126225e-05,
+      "loss": 0.9025,
+      "step": 5821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869227931797941e-05,
+      "loss": 0.8664,
+      "step": 5822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866522651372059e-05,
+      "loss": 0.9462,
+      "step": 5823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863817307048063e-05,
+      "loss": 0.8799,
+      "step": 5824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861111899025454e-05,
+      "loss": 0.8301,
+      "step": 5825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858406427503724e-05,
+      "loss": 0.9326,
+      "step": 5826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855700892682382e-05,
+      "loss": 0.8816,
+      "step": 5827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852995294760932e-05,
+      "loss": 0.8772,
+      "step": 5828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085028963393889e-05,
+      "loss": 0.9264,
+      "step": 5829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847583910415775e-05,
+      "loss": 0.8973,
+      "step": 5830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844878124391106e-05,
+      "loss": 0.8441,
+      "step": 5831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842172276064413e-05,
+      "loss": 0.9629,
+      "step": 5832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839466365635225e-05,
+      "loss": 0.8572,
+      "step": 5833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836760393303081e-05,
+      "loss": 0.9263,
+      "step": 5834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834054359267516e-05,
+      "loss": 0.9659,
+      "step": 5835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831348263728083e-05,
+      "loss": 0.9013,
+      "step": 5836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828642106884328e-05,
+      "loss": 1.0118,
+      "step": 5837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825935888935801e-05,
+      "loss": 0.8939,
+      "step": 5838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082322961008207e-05,
+      "loss": 0.8945,
+      "step": 5839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082052327052269e-05,
+      "loss": 0.7921,
+      "step": 5840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817816870457232e-05,
+      "loss": 0.9152,
+      "step": 5841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081511041008527e-05,
+      "loss": 0.8817,
+      "step": 5842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812403889606378e-05,
+      "loss": 0.8266,
+      "step": 5843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809697309220136e-05,
+      "loss": 0.8768,
+      "step": 5844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806990669126132e-05,
+      "loss": 0.8956,
+      "step": 5845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804283969523958e-05,
+      "loss": 0.2738,
+      "step": 5846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801577210613203e-05,
+      "loss": 0.986,
+      "step": 5847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798870392593473e-05,
+      "loss": 0.9012,
+      "step": 5848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796163515664362e-05,
+      "loss": 0.8951,
+      "step": 5849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793456580025481e-05,
+      "loss": 0.9032,
+      "step": 5850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790749585876448e-05,
+      "loss": 0.9722,
+      "step": 5851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078804253341687e-05,
+      "loss": 0.8147,
+      "step": 5852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078533542284637e-05,
+      "loss": 0.9718,
+      "step": 5853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782628254364576e-05,
+      "loss": 0.9491,
+      "step": 5854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779921028171112e-05,
+      "loss": 1.0156,
+      "step": 5855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777213744465614e-05,
+      "loss": 0.2947,
+      "step": 5856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774506403447722e-05,
+      "loss": 0.9563,
+      "step": 5857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771799005317071e-05,
+      "loss": 1.0027,
+      "step": 5858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769091550273311e-05,
+      "loss": 0.8876,
+      "step": 5859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766384038516093e-05,
+      "loss": 0.8545,
+      "step": 5860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763676470245068e-05,
+      "loss": 0.9493,
+      "step": 5861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760968845659898e-05,
+      "loss": 0.8456,
+      "step": 5862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758261164960242e-05,
+      "loss": 0.9223,
+      "step": 5863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755553428345768e-05,
+      "loss": 0.8901,
+      "step": 5864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752845636016147e-05,
+      "loss": 0.9646,
+      "step": 5865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750137788171054e-05,
+      "loss": 0.89,
+      "step": 5866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074742988501017e-05,
+      "loss": 0.8924,
+      "step": 5867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744721926733172e-05,
+      "loss": 0.8855,
+      "step": 5868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742013913539755e-05,
+      "loss": 0.8867,
+      "step": 5869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739305845629604e-05,
+      "loss": 0.9271,
+      "step": 5870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736597723202417e-05,
+      "loss": 0.9372,
+      "step": 5871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733889546457892e-05,
+      "loss": 0.933,
+      "step": 5872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731181315595733e-05,
+      "loss": 0.8943,
+      "step": 5873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728473030815647e-05,
+      "loss": 0.8322,
+      "step": 5874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725764692317344e-05,
+      "loss": 0.8545,
+      "step": 5875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723056300300543e-05,
+      "loss": 0.961,
+      "step": 5876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720347854964958e-05,
+      "loss": 0.9308,
+      "step": 5877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717639356510318e-05,
+      "loss": 0.8643,
+      "step": 5878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714930805136345e-05,
+      "loss": 0.9697,
+      "step": 5879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071222220104277e-05,
+      "loss": 0.79,
+      "step": 5880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709513544429331e-05,
+      "loss": 0.9317,
+      "step": 5881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706804835495765e-05,
+      "loss": 0.7621,
+      "step": 5882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704096074441813e-05,
+      "loss": 0.9489,
+      "step": 5883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701387261467223e-05,
+      "loss": 0.9004,
+      "step": 5884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698678396771745e-05,
+      "loss": 0.831,
+      "step": 5885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695969480555133e-05,
+      "loss": 0.9316,
+      "step": 5886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693260513017143e-05,
+      "loss": 1.0429,
+      "step": 5887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690551494357542e-05,
+      "loss": 0.9265,
+      "step": 5888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687842424776087e-05,
+      "loss": 0.9067,
+      "step": 5889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685133304472553e-05,
+      "loss": 0.8113,
+      "step": 5890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682424133646712e-05,
+      "loss": 0.8348,
+      "step": 5891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679714912498339e-05,
+      "loss": 0.8741,
+      "step": 5892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677005641227214e-05,
+      "loss": 1.0313,
+      "step": 5893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674296320033125e-05,
+      "loss": 0.8389,
+      "step": 5894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671586949115855e-05,
+      "loss": 0.8067,
+      "step": 5895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668877528675195e-05,
+      "loss": 0.8845,
+      "step": 5896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666168058910944e-05,
+      "loss": 0.7963,
+      "step": 5897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663458540022897e-05,
+      "loss": 0.8402,
+      "step": 5898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066074897221086e-05,
+      "loss": 0.272,
+      "step": 5899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658039355674637e-05,
+      "loss": 0.8958,
+      "step": 5900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655329690614033e-05,
+      "loss": 0.9076,
+      "step": 5901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652619977228866e-05,
+      "loss": 0.9385,
+      "step": 5902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649910215718958e-05,
+      "loss": 0.8291,
+      "step": 5903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0647200406284115e-05,
+      "loss": 0.8771,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644490549124169e-05,
+      "loss": 0.9152,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641780644438949e-05,
+      "loss": 0.946,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639070692428281e-05,
+      "loss": 0.9892,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636360693292003e-05,
+      "loss": 0.2661,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063365064722995e-05,
+      "loss": 0.9316,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630940554441965e-05,
+      "loss": 0.8259,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628230415127888e-05,
+      "loss": 0.8631,
+      "step": 5911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625520229487571e-05,
+      "loss": 0.8139,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622809997720868e-05,
+      "loss": 0.8875,
+      "step": 5913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620099720027625e-05,
+      "loss": 0.8545,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617389396607712e-05,
+      "loss": 0.9111,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614679027660978e-05,
+      "loss": 0.9155,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611968613387298e-05,
+      "loss": 0.9305,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609258153986535e-05,
+      "loss": 0.8744,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606547649658564e-05,
+      "loss": 0.8496,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0603837100603256e-05,
+      "loss": 0.9372,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601126507020492e-05,
+      "loss": 0.9069,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059841586911015e-05,
+      "loss": 0.8432,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595705187072117e-05,
+      "loss": 0.8671,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592994461106285e-05,
+      "loss": 1.0216,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059028369141254e-05,
+      "loss": 0.9219,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0587572878190775e-05,
+      "loss": 0.8602,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058486202164089e-05,
+      "loss": 0.8973,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058215112196279e-05,
+      "loss": 0.8427,
+      "step": 5928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579440179356376e-05,
+      "loss": 0.8737,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057672919402155e-05,
+      "loss": 0.2524,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574018166158228e-05,
+      "loss": 0.8643,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571307095966323e-05,
+      "loss": 0.942,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056859598364575e-05,
+      "loss": 0.972,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565884829396433e-05,
+      "loss": 0.9327,
+      "step": 5934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563173633418289e-05,
+      "loss": 0.866,
+      "step": 5935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560462395911246e-05,
+      "loss": 0.9845,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557751117075234e-05,
+      "loss": 0.9662,
+      "step": 5937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555039797110185e-05,
+      "loss": 0.9145,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055232843621603e-05,
+      "loss": 0.8756,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549617034592713e-05,
+      "loss": 0.9156,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054690559244017e-05,
+      "loss": 0.8893,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054419410995835e-05,
+      "loss": 0.8853,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541482587347196e-05,
+      "loss": 0.8971,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053877102480666e-05,
+      "loss": 0.8939,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536059422536694e-05,
+      "loss": 0.8386,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533347780737254e-05,
+      "loss": 0.8456,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530636099608299e-05,
+      "loss": 1.0215,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527924379349787e-05,
+      "loss": 0.9743,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525212620161693e-05,
+      "loss": 0.2889,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052250082224397e-05,
+      "loss": 0.9253,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519788985796601e-05,
+      "loss": 0.8859,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517077111019551e-05,
+      "loss": 0.9388,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514365198112802e-05,
+      "loss": 0.8632,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511653247276325e-05,
+      "loss": 0.8313,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050894125871011e-05,
+      "loss": 0.9021,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506229232614135e-05,
+      "loss": 0.9293,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503517169188389e-05,
+      "loss": 0.8888,
+      "step": 5957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500805068632864e-05,
+      "loss": 1.0074,
+      "step": 5958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498092931147549e-05,
+      "loss": 0.9217,
+      "step": 5959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495380756932443e-05,
+      "loss": 0.9592,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492668546187544e-05,
+      "loss": 0.9097,
+      "step": 5961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489956299112847e-05,
+      "loss": 0.8963,
+      "step": 5962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048724401590836e-05,
+      "loss": 0.8767,
+      "step": 5963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484531696774087e-05,
+      "loss": 0.8879,
+      "step": 5964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481819341910044e-05,
+      "loss": 0.9,
+      "step": 5965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479106951516229e-05,
+      "loss": 0.945,
+      "step": 5966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476394525792668e-05,
+      "loss": 0.27,
+      "step": 5967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473682064939372e-05,
+      "loss": 0.8188,
+      "step": 5968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470969569156357e-05,
+      "loss": 0.9216,
+      "step": 5969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468257038643657e-05,
+      "loss": 0.8498,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465544473601281e-05,
+      "loss": 0.8115,
+      "step": 5971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462831874229265e-05,
+      "loss": 0.8536,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460119240727635e-05,
+      "loss": 0.8892,
+      "step": 5973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457406573296428e-05,
+      "loss": 0.8353,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045469387213567e-05,
+      "loss": 0.9436,
+      "step": 5975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451981137445406e-05,
+      "loss": 0.7917,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449268369425669e-05,
+      "loss": 0.8422,
+      "step": 5977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446555568276504e-05,
+      "loss": 0.9857,
+      "step": 5978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443842734197955e-05,
+      "loss": 0.278,
+      "step": 5979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0441129867390071e-05,
+      "loss": 0.9095,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438416968052895e-05,
+      "loss": 0.8841,
+      "step": 5981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435704036386486e-05,
+      "loss": 0.8215,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043299107259089e-05,
+      "loss": 0.8975,
+      "step": 5983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430278076866165e-05,
+      "loss": 0.8947,
+      "step": 5984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427565049412376e-05,
+      "loss": 0.8914,
+      "step": 5985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042485199042958e-05,
+      "loss": 0.8526,
+      "step": 5986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422138900117837e-05,
+      "loss": 0.9055,
+      "step": 5987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041942577867722e-05,
+      "loss": 0.9963,
+      "step": 5988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416712626307788e-05,
+      "loss": 0.888,
+      "step": 5989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413999443209616e-05,
+      "loss": 0.8775,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411286229582778e-05,
+      "loss": 0.8044,
+      "step": 5991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408572985627345e-05,
+      "loss": 0.2987,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405859711543393e-05,
+      "loss": 0.8934,
+      "step": 5993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403146407531008e-05,
+      "loss": 0.833,
+      "step": 5994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400433073790266e-05,
+      "loss": 0.8979,
+      "step": 5995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039771971052125e-05,
+      "loss": 0.8557,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395006317924048e-05,
+      "loss": 0.8846,
+      "step": 5997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392292896198746e-05,
+      "loss": 0.9899,
+      "step": 5998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389579445545436e-05,
+      "loss": 0.9067,
+      "step": 5999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386865966164208e-05,
+      "loss": 0.8848,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384152458255162e-05,
+      "loss": 1.0334,
+      "step": 6001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381438922018381e-05,
+      "loss": 0.9201,
+      "step": 6002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378725357653978e-05,
+      "loss": 0.9555,
+      "step": 6003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376011765362046e-05,
+      "loss": 0.8163,
+      "step": 6004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373298145342692e-05,
+      "loss": 0.8781,
+      "step": 6005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370584497796015e-05,
+      "loss": 0.9215,
+      "step": 6006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367870822922126e-05,
+      "loss": 0.8578,
+      "step": 6007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365157120921132e-05,
+      "loss": 0.992,
+      "step": 6008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362443391993142e-05,
+      "loss": 0.9584,
+      "step": 6009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359729636338272e-05,
+      "loss": 0.8527,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357015854156635e-05,
+      "loss": 0.8354,
+      "step": 6011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035430204564835e-05,
+      "loss": 0.9515,
+      "step": 6012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0351588211013532e-05,
+      "loss": 0.9008,
+      "step": 6013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348874350452303e-05,
+      "loss": 0.9649,
+      "step": 6014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346160464164783e-05,
+      "loss": 0.852,
+      "step": 6015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343446552351099e-05,
+      "loss": 0.93,
+      "step": 6016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034073261521138e-05,
+      "loss": 0.8624,
+      "step": 6017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338018652945746e-05,
+      "loss": 0.9198,
+      "step": 6018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335304665754334e-05,
+      "loss": 0.9803,
+      "step": 6019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332590653837274e-05,
+      "loss": 0.8325,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329876617394694e-05,
+      "loss": 0.8877,
+      "step": 6021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032716255662674e-05,
+      "loss": 0.889,
+      "step": 6022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324448471733541e-05,
+      "loss": 0.852,
+      "step": 6023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321734362915235e-05,
+      "loss": 0.8052,
+      "step": 6024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319020230371967e-05,
+      "loss": 0.8217,
+      "step": 6025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316306074303881e-05,
+      "loss": 0.9009,
+      "step": 6026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313591894911116e-05,
+      "loss": 0.8632,
+      "step": 6027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310877692393824e-05,
+      "loss": 0.9321,
+      "step": 6028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308163466952143e-05,
+      "loss": 0.9279,
+      "step": 6029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030544921878623e-05,
+      "loss": 0.8884,
+      "step": 6030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302734948096236e-05,
+      "loss": 0.9075,
+      "step": 6031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300020655082313e-05,
+      "loss": 0.8652,
+      "step": 6032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297306339944608e-05,
+      "loss": 0.8834,
+      "step": 6033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294592002883286e-05,
+      "loss": 0.8759,
+      "step": 6034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291877644098503e-05,
+      "loss": 0.8947,
+      "step": 6035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289163263790412e-05,
+      "loss": 0.2966,
+      "step": 6036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286448862159182e-05,
+      "loss": 0.3037,
+      "step": 6037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283734439404971e-05,
+      "loss": 0.8024,
+      "step": 6038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281019995727939e-05,
+      "loss": 0.8724,
+      "step": 6039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278305531328258e-05,
+      "loss": 0.9752,
+      "step": 6040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275591046406095e-05,
+      "loss": 0.926,
+      "step": 6041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272876541161614e-05,
+      "loss": 0.9345,
+      "step": 6042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270162015794984e-05,
+      "loss": 0.2698,
+      "step": 6043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267447470506381e-05,
+      "loss": 0.9393,
+      "step": 6044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264732905495975e-05,
+      "loss": 0.9187,
+      "step": 6045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262018320963945e-05,
+      "loss": 0.9064,
+      "step": 6046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025930371711046e-05,
+      "loss": 0.9002,
+      "step": 6047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02565890941357e-05,
+      "loss": 0.9081,
+      "step": 6048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253874452239848e-05,
+      "loss": 1.0175,
+      "step": 6049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251159791623075e-05,
+      "loss": 0.8363,
+      "step": 6050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024844511248557e-05,
+      "loss": 0.8357,
+      "step": 6051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245730415027509e-05,
+      "loss": 0.9364,
+      "step": 6052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243015699449084e-05,
+      "loss": 0.8538,
+      "step": 6053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240300965950473e-05,
+      "loss": 0.9001,
+      "step": 6054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237586214731868e-05,
+      "loss": 0.8445,
+      "step": 6055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234871445993455e-05,
+      "loss": 0.8556,
+      "step": 6056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232156659935424e-05,
+      "loss": 0.9147,
+      "step": 6057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229441856757964e-05,
+      "loss": 0.2605,
+      "step": 6058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022672703666127e-05,
+      "loss": 0.8746,
+      "step": 6059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224012199845532e-05,
+      "loss": 0.8528,
+      "step": 6060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221297346510943e-05,
+      "loss": 0.9005,
+      "step": 6061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218582476857704e-05,
+      "loss": 0.8425,
+      "step": 6062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215867591086006e-05,
+      "loss": 0.8244,
+      "step": 6063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0213152689396053e-05,
+      "loss": 0.2529,
+      "step": 6064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021043777198804e-05,
+      "loss": 0.9394,
+      "step": 6065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207722839062168e-05,
+      "loss": 0.824,
+      "step": 6066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205007890818637e-05,
+      "loss": 0.2786,
+      "step": 6067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202292927457655e-05,
+      "loss": 0.8038,
+      "step": 6068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019957794917942e-05,
+      "loss": 0.9009,
+      "step": 6069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196862956184135e-05,
+      "loss": 0.8798,
+      "step": 6070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194147948672013e-05,
+      "loss": 0.8539,
+      "step": 6071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019143292684326e-05,
+      "loss": 0.82,
+      "step": 6072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188717890898076e-05,
+      "loss": 0.9005,
+      "step": 6073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018600284103668e-05,
+      "loss": 0.9629,
+      "step": 6074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183287777459276e-05,
+      "loss": 0.7778,
+      "step": 6075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180572700366076e-05,
+      "loss": 0.8934,
+      "step": 6076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177857609957294e-05,
+      "loss": 0.9142,
+      "step": 6077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175142506433142e-05,
+      "loss": 0.9203,
+      "step": 6078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017242738999383e-05,
+      "loss": 0.8846,
+      "step": 6079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169712260839581e-05,
+      "loss": 0.8284,
+      "step": 6080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166997119170608e-05,
+      "loss": 0.9642,
+      "step": 6081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164281965187121e-05,
+      "loss": 0.8424,
+      "step": 6082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161566799089349e-05,
+      "loss": 0.8698,
+      "step": 6083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0158851621077503e-05,
+      "loss": 0.8949,
+      "step": 6084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156136431351802e-05,
+      "loss": 0.9532,
+      "step": 6085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153421230112473e-05,
+      "loss": 0.8862,
+      "step": 6086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015070601755973e-05,
+      "loss": 0.87,
+      "step": 6087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0147990793893798e-05,
+      "loss": 0.7869,
+      "step": 6088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145275559314904e-05,
+      "loss": 0.8906,
+      "step": 6089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142560314023265e-05,
+      "loss": 0.904,
+      "step": 6090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139845058219106e-05,
+      "loss": 0.8877,
+      "step": 6091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137129792102658e-05,
+      "loss": 0.8284,
+      "step": 6092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134414515874143e-05,
+      "loss": 0.9171,
+      "step": 6093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131699229733788e-05,
+      "loss": 0.8964,
+      "step": 6094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128983933881818e-05,
+      "loss": 0.86,
+      "step": 6095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126268628518468e-05,
+      "loss": 0.8863,
+      "step": 6096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123553313843956e-05,
+      "loss": 0.8769,
+      "step": 6097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120837990058525e-05,
+      "loss": 0.8506,
+      "step": 6098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01181226573624e-05,
+      "loss": 1.0092,
+      "step": 6099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115407315955806e-05,
+      "loss": 0.9446,
+      "step": 6100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112691966038982e-05,
+      "loss": 0.8722,
+      "step": 6101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109976607812155e-05,
+      "loss": 0.8779,
+      "step": 6102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107261241475562e-05,
+      "loss": 0.937,
+      "step": 6103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104545867229434e-05,
+      "loss": 0.8677,
+      "step": 6104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101830485274004e-05,
+      "loss": 0.8818,
+      "step": 6105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099115095809508e-05,
+      "loss": 0.8546,
+      "step": 6106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096399699036184e-05,
+      "loss": 0.8798,
+      "step": 6107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093684295154264e-05,
+      "loss": 0.8862,
+      "step": 6108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090968884363986e-05,
+      "loss": 0.9005,
+      "step": 6109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088253466865585e-05,
+      "loss": 0.874,
+      "step": 6110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085538042859297e-05,
+      "loss": 0.9141,
+      "step": 6111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082822612545364e-05,
+      "loss": 0.9523,
+      "step": 6112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008010717612402e-05,
+      "loss": 0.9431,
+      "step": 6113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077391733795504e-05,
+      "loss": 0.8971,
+      "step": 6114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074676285760061e-05,
+      "loss": 0.8661,
+      "step": 6115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071960832217923e-05,
+      "loss": 0.8831,
+      "step": 6116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069245373369331e-05,
+      "loss": 0.9217,
+      "step": 6117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006652990941453e-05,
+      "loss": 0.8082,
+      "step": 6118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063814440553754e-05,
+      "loss": 0.9362,
+      "step": 6119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006109896698725e-05,
+      "loss": 0.921,
+      "step": 6120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058383488915258e-05,
+      "loss": 0.8958,
+      "step": 6121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055668006538014e-05,
+      "loss": 0.8953,
+      "step": 6122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052952520055765e-05,
+      "loss": 0.9099,
+      "step": 6123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050237029668754e-05,
+      "loss": 0.935,
+      "step": 6124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047521535577222e-05,
+      "loss": 0.9052,
+      "step": 6125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044806037981414e-05,
+      "loss": 0.9205,
+      "step": 6126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042090537081568e-05,
+      "loss": 0.9282,
+      "step": 6127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003937503307793e-05,
+      "loss": 0.8506,
+      "step": 6128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036659526170744e-05,
+      "loss": 0.8825,
+      "step": 6129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033944016560257e-05,
+      "loss": 0.9203,
+      "step": 6130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031228504446704e-05,
+      "loss": 0.8653,
+      "step": 6131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028512990030339e-05,
+      "loss": 0.9047,
+      "step": 6132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00257974735114e-05,
+      "loss": 0.7822,
+      "step": 6133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023081955090133e-05,
+      "loss": 0.9271,
+      "step": 6134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020366434966789e-05,
+      "loss": 0.8937,
+      "step": 6135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00176509133416e-05,
+      "loss": 0.8722,
+      "step": 6136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001493539041482e-05,
+      "loss": 0.8854,
+      "step": 6137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012219866386692e-05,
+      "loss": 0.8826,
+      "step": 6138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000950434145746e-05,
+      "loss": 0.8241,
+      "step": 6139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000678881582737e-05,
+      "loss": 0.7737,
+      "step": 6140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000407328969667e-05,
+      "loss": 0.9101,
+      "step": 6141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001357763265598e-05,
+      "loss": 0.9375,
+      "step": 6142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998642236734405e-06,
+      "loss": 0.8617,
+      "step": 6143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995926710303337e-06,
+      "loss": 0.9271,
+      "step": 6144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993211184172635e-06,
+      "loss": 0.2511,
+      "step": 6145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990495658542543e-06,
+      "loss": 0.7816,
+      "step": 6146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987780133613311e-06,
+      "loss": 0.9278,
+      "step": 6147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985064609585184e-06,
+      "loss": 0.8947,
+      "step": 6148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982349086658401e-06,
+      "loss": 0.7322,
+      "step": 6149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979633565033216e-06,
+      "loss": 0.8658,
+      "step": 6150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976918044909867e-06,
+      "loss": 0.9471,
+      "step": 6151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.974202526488603e-06,
+      "loss": 0.9145,
+      "step": 6152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971487009969666e-06,
+      "loss": 0.9241,
+      "step": 6153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9687714955533e-06,
+      "loss": 0.7651,
+      "step": 6154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96605598343975e-06,
+      "loss": 0.8194,
+      "step": 6155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96334047382926e-06,
+      "loss": 0.8619,
+      "step": 6156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960624966922075e-06,
+      "loss": 0.8107,
+      "step": 6157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957909462918435e-06,
+      "loss": 0.8985,
+      "step": 6158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95519396201859e-06,
+      "loss": 0.9454,
+      "step": 6159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95247846442278e-06,
+      "loss": 0.9189,
+      "step": 6160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949762970331248e-06,
+      "loss": 0.9689,
+      "step": 6161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947047479944235e-06,
+      "loss": 0.8708,
+      "step": 6162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944331993461988e-06,
+      "loss": 0.903,
+      "step": 6163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941616511084747e-06,
+      "loss": 0.975,
+      "step": 6164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938901033012753e-06,
+      "loss": 0.828,
+      "step": 6165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936185559446248e-06,
+      "loss": 0.8165,
+      "step": 6166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933470090585473e-06,
+      "loss": 0.8587,
+      "step": 6167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93075462663067e-06,
+      "loss": 0.9671,
+      "step": 6168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928039167782079e-06,
+      "loss": 0.9879,
+      "step": 6169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925323714239942e-06,
+      "loss": 0.8452,
+      "step": 6170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922608266204494e-06,
+      "loss": 0.9119,
+      "step": 6171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91989282387598e-06,
+      "loss": 0.8605,
+      "step": 6172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917177387454638e-06,
+      "loss": 0.9016,
+      "step": 6173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914461957140706e-06,
+      "loss": 0.2262,
+      "step": 6174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91174653313442e-06,
+      "loss": 0.8452,
+      "step": 6175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909031115636017e-06,
+      "loss": 0.9122,
+      "step": 6176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906315704845738e-06,
+      "loss": 0.8331,
+      "step": 6177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903600300963818e-06,
+      "loss": 0.8799,
+      "step": 6178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900884904190493e-06,
+      "loss": 0.8741,
+      "step": 6179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898169514725998e-06,
+      "loss": 0.8989,
+      "step": 6180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89545413277057e-06,
+      "loss": 0.8041,
+      "step": 6181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892738758524442e-06,
+      "loss": 0.799,
+      "step": 6182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890023392187845e-06,
+      "loss": 0.9401,
+      "step": 6183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887308033961025e-06,
+      "loss": 0.7284,
+      "step": 6184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884592684044198e-06,
+      "loss": 0.9161,
+      "step": 6185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881877342637605e-06,
+      "loss": 0.8201,
+      "step": 6186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879162009941478e-06,
+      "loss": 0.2748,
+      "step": 6187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876446686156046e-06,
+      "loss": 0.8705,
+      "step": 6188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873731371481537e-06,
+      "loss": 0.9201,
+      "step": 6189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871016066118184e-06,
+      "loss": 0.89,
+      "step": 6190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868300770266217e-06,
+      "loss": 0.8145,
+      "step": 6191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865585484125858e-06,
+      "loss": 0.9373,
+      "step": 6192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862870207897343e-06,
+      "loss": 0.8256,
+      "step": 6193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860154941780896e-06,
+      "loss": 0.9331,
+      "step": 6194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85743968597674e-06,
+      "loss": 0.8461,
+      "step": 6195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854724440685101e-06,
+      "loss": 0.9305,
+      "step": 6196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.852009206106206e-06,
+      "loss": 0.8812,
+      "step": 6197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849293982440273e-06,
+      "loss": 0.9385,
+      "step": 6198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84657876988753e-06,
+      "loss": 0.9996,
+      "step": 6199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8438635686482e-06,
+      "loss": 0.9449,
+      "step": 6200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841148378922499e-06,
+      "loss": 0.8585,
+      "step": 6201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838433200910653e-06,
+      "loss": 0.9333,
+      "step": 6202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835718034812879e-06,
+      "loss": 0.8379,
+      "step": 6203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833002880829395e-06,
+      "loss": 0.2404,
+      "step": 6204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830287739160424e-06,
+      "loss": 0.805,
+      "step": 6205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827572610006173e-06,
+      "loss": 0.7942,
+      "step": 6206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824857493566863e-06,
+      "loss": 0.9459,
+      "step": 6207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822142390042711e-06,
+      "loss": 0.9354,
+      "step": 6208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819427299633927e-06,
+      "loss": 0.8141,
+      "step": 6209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816712222540726e-06,
+      "loss": 0.9474,
+      "step": 6210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813997158963323e-06,
+      "loss": 0.9451,
+      "step": 6211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811282109101926e-06,
+      "loss": 0.9195,
+      "step": 6212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808567073156746e-06,
+      "loss": 0.8278,
+      "step": 6213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805852051327988e-06,
+      "loss": 0.9371,
+      "step": 6214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80313704381587e-06,
+      "loss": 0.9723,
+      "step": 6215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800422050820586e-06,
+      "loss": 0.9284,
+      "step": 6216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797707072542351e-06,
+      "loss": 0.9021,
+      "step": 6217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794992109181368e-06,
+      "loss": 0.8328,
+      "step": 6218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792277160937835e-06,
+      "loss": 0.9291,
+      "step": 6219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789562228011964e-06,
+      "loss": 0.9044,
+      "step": 6220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786847310603949e-06,
+      "loss": 0.8577,
+      "step": 6221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784132408913996e-06,
+      "loss": 0.2995,
+      "step": 6222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781417523142298e-06,
+      "loss": 0.8885,
+      "step": 6223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778702653489058e-06,
+      "loss": 1.039,
+      "step": 6224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775987800154472e-06,
+      "loss": 0.9026,
+      "step": 6225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773272963338736e-06,
+      "loss": 0.9905,
+      "step": 6226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770558143242041e-06,
+      "loss": 0.8705,
+      "step": 6227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76784334006458e-06,
+      "loss": 0.9074,
+      "step": 6228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765128554006548e-06,
+      "loss": 0.8622,
+      "step": 6229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762413785268134e-06,
+      "loss": 0.9205,
+      "step": 6230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75969903404953e-06,
+      "loss": 0.8861,
+      "step": 6231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75698430055092e-06,
+      "loss": 0.9002,
+      "step": 6232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754269584972493e-06,
+      "loss": 0.8641,
+      "step": 6233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751554887514435e-06,
+      "loss": 0.9871,
+      "step": 6234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748840208376925e-06,
+      "loss": 0.8154,
+      "step": 6235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746125547760159e-06,
+      "loss": 0.9112,
+      "step": 6236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743410905864301e-06,
+      "loss": 0.9815,
+      "step": 6237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740696282889542e-06,
+      "loss": 0.8959,
+      "step": 6238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737981679036058e-06,
+      "loss": 0.8975,
+      "step": 6239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735267094504026e-06,
+      "loss": 0.9558,
+      "step": 6240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73255252949362e-06,
+      "loss": 0.9806,
+      "step": 6241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729837984205018e-06,
+      "loss": 0.9258,
+      "step": 6242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72712345883839e-06,
+      "loss": 0.905,
+      "step": 6243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724408953593907e-06,
+      "loss": 0.9017,
+      "step": 6244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721694468671741e-06,
+      "loss": 0.8797,
+      "step": 6245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718980004272063e-06,
+      "loss": 0.9281,
+      "step": 6246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716265560595036e-06,
+      "loss": 0.282,
+      "step": 6247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713551137840823e-06,
+      "loss": 0.8275,
+      "step": 6248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710836736209593e-06,
+      "loss": 0.8381,
+      "step": 6249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7081223559015e-06,
+      "loss": 0.2312,
+      "step": 6250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705407997116717e-06,
+      "loss": 0.8703,
+      "step": 6251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702693660055395e-06,
+      "loss": 0.8277,
+      "step": 6252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69997934491769e-06,
+      "loss": 1.0229,
+      "step": 6253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697265051903765e-06,
+      "loss": 0.8948,
+      "step": 6254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694550781213771e-06,
+      "loss": 0.8675,
+      "step": 6255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691836533047859e-06,
+      "loss": 0.9286,
+      "step": 6256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689122307606183e-06,
+      "loss": 0.8336,
+      "step": 6257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686408105088889e-06,
+      "loss": 0.9067,
+      "step": 6258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683693925696122e-06,
+      "loss": 0.8752,
+      "step": 6259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680979769628034e-06,
+      "loss": 0.7739,
+      "step": 6260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678265637084767e-06,
+      "loss": 0.7999,
+      "step": 6261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675551528266462e-06,
+      "loss": 0.9281,
+      "step": 6262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672837443373263e-06,
+      "loss": 0.8272,
+      "step": 6263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670123382605306e-06,
+      "loss": 0.8871,
+      "step": 6264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.667409346162731e-06,
+      "loss": 0.9117,
+      "step": 6265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664695334245666e-06,
+      "loss": 0.8852,
+      "step": 6266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661981347054259e-06,
+      "loss": 0.8832,
+      "step": 6267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659267384788625e-06,
+      "loss": 0.9583,
+      "step": 6268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656553447648905e-06,
+      "loss": 0.8644,
+      "step": 6269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653839535835222e-06,
+      "loss": 0.8234,
+      "step": 6270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6511256495477e-06,
+      "loss": 0.8548,
+      "step": 6271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648411788986471e-06,
+      "loss": 0.9511,
+      "step": 6272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645697954351652e-06,
+      "loss": 1.0075,
+      "step": 6273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642984145843366e-06,
+      "loss": 0.9263,
+      "step": 6274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640270363661728e-06,
+      "loss": 0.8755,
+      "step": 6275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637556608006858e-06,
+      "loss": 0.8992,
+      "step": 6276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634842879078873e-06,
+      "loss": 0.8934,
+      "step": 6277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632129177077879e-06,
+      "loss": 0.8646,
+      "step": 6278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629415502203989e-06,
+      "loss": 0.9158,
+      "step": 6279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626701854657311e-06,
+      "loss": 0.8067,
+      "step": 6280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623988234637956e-06,
+      "loss": 0.8391,
+      "step": 6281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621274642346024e-06,
+      "loss": 0.9479,
+      "step": 6282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61856107798162e-06,
+      "loss": 0.8871,
+      "step": 6283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615847541744843e-06,
+      "loss": 0.7616,
+      "step": 6284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613134033835794e-06,
+      "loss": 0.8646,
+      "step": 6285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610420554454565e-06,
+      "loss": 0.8064,
+      "step": 6286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607707103801259e-06,
+      "loss": 1.0479,
+      "step": 6287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604993682075957e-06,
+      "loss": 0.8789,
+      "step": 6288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602280289478753e-06,
+      "loss": 0.2831,
+      "step": 6289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599566926209737e-06,
+      "loss": 0.8564,
+      "step": 6290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596853592468995e-06,
+      "loss": 0.9641,
+      "step": 6291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594140288456609e-06,
+      "loss": 0.8699,
+      "step": 6292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591427014372657e-06,
+      "loss": 0.7987,
+      "step": 6293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588713770417226e-06,
+      "loss": 0.757,
+      "step": 6294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586000556790385e-06,
+      "loss": 0.8537,
+      "step": 6295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583287373692213e-06,
+      "loss": 0.9242,
+      "step": 6296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580574221322782e-06,
+      "loss": 0.9396,
+      "step": 6297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577861099882166e-06,
+      "loss": 0.8695,
+      "step": 6298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575148009570424e-06,
+      "loss": 0.8865,
+      "step": 6299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572434950587627e-06,
+      "loss": 0.8463,
+      "step": 6300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569721923133836e-06,
+      "loss": 0.9601,
+      "step": 6301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567008927409113e-06,
+      "loss": 0.8671,
+      "step": 6302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56429596361352e-06,
+      "loss": 0.8683,
+      "step": 6303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561583031947108e-06,
+      "loss": 0.8772,
+      "step": 6304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55887013260993e-06,
+      "loss": 0.9197,
+      "step": 6305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556157265802045e-06,
+      "loss": 0.916,
+      "step": 6306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553444431723497e-06,
+      "loss": 0.8608,
+      "step": 6307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550731630574335e-06,
+      "loss": 0.8591,
+      "step": 6308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548018862554599e-06,
+      "loss": 0.8965,
+      "step": 6309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545306127864333e-06,
+      "loss": 0.8396,
+      "step": 6310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542593426703576e-06,
+      "loss": 0.9032,
+      "step": 6311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539880759272367e-06,
+      "loss": 0.8398,
+      "step": 6312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537168125770738e-06,
+      "loss": 0.8587,
+      "step": 6313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53445552639872e-06,
+      "loss": 0.896,
+      "step": 6314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531742961356348e-06,
+      "loss": 0.9933,
+      "step": 6315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529030430843642e-06,
+      "loss": 0.893,
+      "step": 6316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526317935060632e-06,
+      "loss": 0.2847,
+      "step": 6317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523605474207337e-06,
+      "loss": 0.2496,
+      "step": 6318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520893048483774e-06,
+      "loss": 1.0048,
+      "step": 6319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518180658089961e-06,
+      "loss": 0.9431,
+      "step": 6320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515468303225914e-06,
+      "loss": 0.9377,
+      "step": 6321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512755984091643e-06,
+      "loss": 0.9294,
+      "step": 6322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510043700887156e-06,
+      "loss": 0.8819,
+      "step": 6323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507331453812461e-06,
+      "loss": 0.9138,
+      "step": 6324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504619243067559e-06,
+      "loss": 0.9329,
+      "step": 6325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501907068852453e-06,
+      "loss": 0.2436,
+      "step": 6326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499194931367138e-06,
+      "loss": 0.8945,
+      "step": 6327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496482830811611e-06,
+      "loss": 0.8441,
+      "step": 6328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493770767385868e-06,
+      "loss": 0.7979,
+      "step": 6329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491058741289894e-06,
+      "loss": 0.937,
+      "step": 6330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488346752723678e-06,
+      "loss": 0.9352,
+      "step": 6331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485634801887202e-06,
+      "loss": 0.9442,
+      "step": 6332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48292288898045e-06,
+      "loss": 0.9883,
+      "step": 6333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480211014203402e-06,
+      "loss": 0.852,
+      "step": 6334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477499177756031e-06,
+      "loss": 0.8959,
+      "step": 6335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47478737983831e-06,
+      "loss": 0.8615,
+      "step": 6336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472075620650213e-06,
+      "loss": 0.9277,
+      "step": 6337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469363900391703e-06,
+      "loss": 0.8936,
+      "step": 6338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466652219262751e-06,
+      "loss": 0.9703,
+      "step": 6339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463940577463311e-06,
+      "loss": 0.8695,
+      "step": 6340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461228975193344e-06,
+      "loss": 0.7983,
+      "step": 6341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458517412652806e-06,
+      "loss": 0.8446,
+      "step": 6342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455805890041654e-06,
+      "loss": 0.8822,
+      "step": 6343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453094407559833e-06,
+      "loss": 0.8696,
+      "step": 6344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45038296540729e-06,
+      "loss": 0.9521,
+      "step": 6345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447671563783973e-06,
+      "loss": 0.8704,
+      "step": 6346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44496020288982e-06,
+      "loss": 0.9032,
+      "step": 6347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442248882924768e-06,
+      "loss": 0.9706,
+      "step": 6348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43953760408876e-06,
+      "loss": 0.8998,
+      "step": 6349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436826366581714e-06,
+      "loss": 0.9252,
+      "step": 6350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43411517060357e-06,
+      "loss": 0.8723,
+      "step": 6351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431404016354251e-06,
+      "loss": 0.953,
+      "step": 6352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42869290403368e-06,
+      "loss": 0.8801,
+      "step": 6353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425981833841773e-06,
+      "loss": 0.8484,
+      "step": 6354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423270805978453e-06,
+      "loss": 0.9064,
+      "step": 6355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42055982064363e-06,
+      "loss": 0.8412,
+      "step": 6356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41784887803721e-06,
+      "loss": 0.8213,
+      "step": 6357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415137978359109e-06,
+      "loss": 0.909,
+      "step": 6358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.412427121809227e-06,
+      "loss": 0.9016,
+      "step": 6359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409716308587465e-06,
+      "loss": 0.877,
+      "step": 6360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40700553889372e-06,
+      "loss": 0.8545,
+      "step": 6361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404294812927885e-06,
+      "loss": 0.9179,
+      "step": 6362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401584130889853e-06,
+      "loss": 0.9074,
+      "step": 6363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398873492979513e-06,
+      "loss": 0.8652,
+      "step": 6364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396162899396749e-06,
+      "loss": 0.2493,
+      "step": 6365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393452350341437e-06,
+      "loss": 0.8024,
+      "step": 6366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390741846013465e-06,
+      "loss": 0.9587,
+      "step": 6367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388031386612704e-06,
+      "loss": 0.9623,
+      "step": 6368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385320972339023e-06,
+      "loss": 0.8784,
+      "step": 6369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382610603392295e-06,
+      "loss": 0.8845,
+      "step": 6370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379900279972378e-06,
+      "loss": 0.8567,
+      "step": 6371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377190002279137e-06,
+      "loss": 0.8345,
+      "step": 6372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374479770512432e-06,
+      "loss": 0.9427,
+      "step": 6373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371769584872117e-06,
+      "loss": 1.0001,
+      "step": 6374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36905944555804e-06,
+      "loss": 0.813,
+      "step": 6375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366349352770054e-06,
+      "loss": 0.9285,
+      "step": 6376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363639306708e-06,
+      "loss": 0.9234,
+      "step": 6377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360929307571722e-06,
+      "loss": 0.9012,
+      "step": 6378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358219355561053e-06,
+      "loss": 0.859,
+      "step": 6379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355509450875836e-06,
+      "loss": 0.8917,
+      "step": 6380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35279959371589e-06,
+      "loss": 1.0455,
+      "step": 6381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350089784281049e-06,
+      "loss": 0.9644,
+      "step": 6382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347380022771136e-06,
+      "loss": 0.9291,
+      "step": 6383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344670309385969e-06,
+      "loss": 0.9512,
+      "step": 6384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341960644325366e-06,
+      "loss": 0.8858,
+      "step": 6385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339251027789142e-06,
+      "loss": 0.9348,
+      "step": 6386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336541459977105e-06,
+      "loss": 0.9366,
+      "step": 6387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333831941089056e-06,
+      "loss": 0.9093,
+      "step": 6388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331122471324807e-06,
+      "loss": 0.2552,
+      "step": 6389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328413050884148e-06,
+      "loss": 0.8163,
+      "step": 6390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32570367996688e-06,
+      "loss": 0.87,
+      "step": 6391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32299435877279e-06,
+      "loss": 0.9316,
+      "step": 6392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320285087501665e-06,
+      "loss": 0.9291,
+      "step": 6393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317575866353293e-06,
+      "loss": 0.9146,
+      "step": 6394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31486669552745e-06,
+      "loss": 0.8596,
+      "step": 6395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312157575223916e-06,
+      "loss": 0.8881,
+      "step": 6396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309448505642463e-06,
+      "loss": 1.0474,
+      "step": 6397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306739486982859e-06,
+      "loss": 0.7678,
+      "step": 6398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30403051944487e-06,
+      "loss": 0.837,
+      "step": 6399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301321603228255e-06,
+      "loss": 0.8317,
+      "step": 6400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298612738532783e-06,
+      "loss": 0.9397,
+      "step": 6401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29590392555819e-06,
+      "loss": 0.8599,
+      "step": 6402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29319516450424e-06,
+      "loss": 0.925,
+      "step": 6403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290486455570672e-06,
+      "loss": 0.8784,
+      "step": 6404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287777798957234e-06,
+      "loss": 0.9457,
+      "step": 6405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285069194863657e-06,
+      "loss": 1.0133,
+      "step": 6406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282360643489685e-06,
+      "loss": 0.9069,
+      "step": 6407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279652145035043e-06,
+      "loss": 0.9982,
+      "step": 6408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276943699699457e-06,
+      "loss": 0.822,
+      "step": 6409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274235307682656e-06,
+      "loss": 0.8991,
+      "step": 6410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271526969184357e-06,
+      "loss": 0.9782,
+      "step": 6411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268818684404272e-06,
+      "loss": 0.817,
+      "step": 6412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266110453542113e-06,
+      "loss": 0.9944,
+      "step": 6413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263402276797586e-06,
+      "loss": 0.8707,
+      "step": 6414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260694154370399e-06,
+      "loss": 0.8305,
+      "step": 6415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25798608646025e-06,
+      "loss": 0.8403,
+      "step": 6416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25527807326683e-06,
+      "loss": 0.8964,
+      "step": 6417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252570114989832e-06,
+      "loss": 0.9389,
+      "step": 6418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249862211828946e-06,
+      "loss": 0.8602,
+      "step": 6419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247154363983853e-06,
+      "loss": 0.8741,
+      "step": 6420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24444657165423e-06,
+      "loss": 0.9096,
+      "step": 6421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241738835039763e-06,
+      "loss": 0.8688,
+      "step": 6422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239031154340105e-06,
+      "loss": 0.9634,
+      "step": 6423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236323529754933e-06,
+      "loss": 0.8836,
+      "step": 6424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23361596148391e-06,
+      "loss": 0.9256,
+      "step": 6425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230908449726692e-06,
+      "loss": 0.9164,
+      "step": 6426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22820099468293e-06,
+      "loss": 0.921,
+      "step": 6427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225493596552282e-06,
+      "loss": 1.012,
+      "step": 6428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222786255534388e-06,
+      "loss": 0.8757,
+      "step": 6429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220078971828888e-06,
+      "loss": 0.8595,
+      "step": 6430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217371745635426e-06,
+      "loss": 0.8828,
+      "step": 6431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214664577153634e-06,
+      "loss": 0.7843,
+      "step": 6432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211957466583135e-06,
+      "loss": 0.8429,
+      "step": 6433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209250414123558e-06,
+      "loss": 0.8722,
+      "step": 6434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206543419974522e-06,
+      "loss": 0.9083,
+      "step": 6435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203836484335642e-06,
+      "loss": 0.2815,
+      "step": 6436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201129607406532e-06,
+      "loss": 0.2482,
+      "step": 6437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198422789386799e-06,
+      "loss": 0.8318,
+      "step": 6438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195716030476043e-06,
+      "loss": 0.8335,
+      "step": 6439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193009330873867e-06,
+      "loss": 0.8655,
+      "step": 6440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190302690779866e-06,
+      "loss": 0.2984,
+      "step": 6441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187596110393628e-06,
+      "loss": 0.884,
+      "step": 6442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184889589914736e-06,
+      "loss": 0.8493,
+      "step": 6443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182183129542773e-06,
+      "loss": 0.8983,
+      "step": 6444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179476729477314e-06,
+      "loss": 0.2285,
+      "step": 6445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176770389917936e-06,
+      "loss": 0.9548,
+      "step": 6446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1740641110642e-06,
+      "loss": 0.8941,
+      "step": 6447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171357893115675e-06,
+      "loss": 0.9177,
+      "step": 6448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168651736271919e-06,
+      "loss": 0.9553,
+      "step": 6449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165945640732484e-06,
+      "loss": 0.9179,
+      "step": 6450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163239606696922e-06,
+      "loss": 1.0273,
+      "step": 6451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160533634364775e-06,
+      "loss": 0.8988,
+      "step": 6452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157827723935592e-06,
+      "loss": 0.9141,
+      "step": 6453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155121875608897e-06,
+      "loss": 0.9777,
+      "step": 6454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152416089584228e-06,
+      "loss": 0.9298,
+      "step": 6455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149710366061113e-06,
+      "loss": 0.2361,
+      "step": 6456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14700470523907e-06,
+      "loss": 0.8494,
+      "step": 6457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144299107317621e-06,
+      "loss": 0.8244,
+      "step": 6458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141593572496279e-06,
+      "loss": 0.8935,
+      "step": 6459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13888810097455e-06,
+      "loss": 0.9177,
+      "step": 6460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136182692951936e-06,
+      "loss": 0.9034,
+      "step": 6461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133477348627943e-06,
+      "loss": 0.8787,
+      "step": 6462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130772068202062e-06,
+      "loss": 0.8343,
+      "step": 6463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12806685187378e-06,
+      "loss": 0.8886,
+      "step": 6464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125361699842583e-06,
+      "loss": 0.9407,
+      "step": 6465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12265661230795e-06,
+      "loss": 0.9152,
+      "step": 6466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11995158946936e-06,
+      "loss": 0.8945,
+      "step": 6467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117246631526284e-06,
+      "loss": 0.9446,
+      "step": 6468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114541738678183e-06,
+      "loss": 0.8878,
+      "step": 6469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11183691112452e-06,
+      "loss": 0.8533,
+      "step": 6470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109132149064756e-06,
+      "loss": 0.8977,
+      "step": 6471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106427452698338e-06,
+      "loss": 0.9196,
+      "step": 6472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103722822224715e-06,
+      "loss": 0.8741,
+      "step": 6473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101018257843325e-06,
+      "loss": 0.8732,
+      "step": 6474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098313759753605e-06,
+      "loss": 0.9487,
+      "step": 6475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09560932815499e-06,
+      "loss": 0.8361,
+      "step": 6476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092904963246908e-06,
+      "loss": 0.8805,
+      "step": 6477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090200665228777e-06,
+      "loss": 0.7645,
+      "step": 6478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087496434300014e-06,
+      "loss": 0.938,
+      "step": 6479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084792270660037e-06,
+      "loss": 0.9551,
+      "step": 6480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08208817450825e-06,
+      "loss": 0.9685,
+      "step": 6481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079384146044051e-06,
+      "loss": 0.9437,
+      "step": 6482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076680185466845e-06,
+      "loss": 0.8692,
+      "step": 6483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073976292976024e-06,
+      "loss": 0.9249,
+      "step": 6484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071272468770968e-06,
+      "loss": 0.8281,
+      "step": 6485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068568713051065e-06,
+      "loss": 0.8563,
+      "step": 6486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06586502601569e-06,
+      "loss": 0.8194,
+      "step": 6487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063161407864214e-06,
+      "loss": 0.9511,
+      "step": 6488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060457858796009e-06,
+      "loss": 0.8812,
+      "step": 6489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057754379010433e-06,
+      "loss": 0.8665,
+      "step": 6490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05505096870684e-06,
+      "loss": 0.884,
+      "step": 6491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05234762808459e-06,
+      "loss": 0.8911,
+      "step": 6492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049644357343025e-06,
+      "loss": 0.9504,
+      "step": 6493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046941156681488e-06,
+      "loss": 0.921,
+      "step": 6494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044238026299312e-06,
+      "loss": 0.9086,
+      "step": 6495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04153496639583e-06,
+      "loss": 0.9467,
+      "step": 6496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038831977170366e-06,
+      "loss": 0.9429,
+      "step": 6497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036129058822244e-06,
+      "loss": 0.8273,
+      "step": 6498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03342621155078e-06,
+      "loss": 0.7829,
+      "step": 6499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03072343555528e-06,
+      "loss": 0.9127,
+      "step": 6500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028020731035052e-06,
+      "loss": 0.8062,
+      "step": 6501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025318098189395e-06,
+      "loss": 0.8688,
+      "step": 6502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022615537217604e-06,
+      "loss": 0.8758,
+      "step": 6503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01991304831897e-06,
+      "loss": 0.864,
+      "step": 6504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017210631692772e-06,
+      "loss": 0.8923,
+      "step": 6505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01450828753829e-06,
+      "loss": 0.8407,
+      "step": 6506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0118060160548e-06,
+      "loss": 0.9904,
+      "step": 6507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009103817441566e-06,
+      "loss": 0.8533,
+      "step": 6508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006401691897852e-06,
+      "loss": 1.0273,
+      "step": 6509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003699639622916e-06,
+      "loss": 0.8817,
+      "step": 6510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000997660816009e-06,
+      "loss": 0.9534,
+      "step": 6511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998295755676378e-06,
+      "loss": 0.8709,
+      "step": 6512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995593924403259e-06,
+      "loss": 0.8728,
+      "step": 6513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992892167195896e-06,
+      "loss": 1.021,
+      "step": 6514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990190484253513e-06,
+      "loss": 0.8657,
+      "step": 6515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987488875775337e-06,
+      "loss": 0.9312,
+      "step": 6516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984787341960582e-06,
+      "loss": 0.8214,
+      "step": 6517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982085883008463e-06,
+      "loss": 0.9131,
+      "step": 6518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979384499118193e-06,
+      "loss": 0.8439,
+      "step": 6519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97668319048897e-06,
+      "loss": 0.881,
+      "step": 6520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973981957319992e-06,
+      "loss": 0.9368,
+      "step": 6521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971280799810447e-06,
+      "loss": 0.9341,
+      "step": 6522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968579718159527e-06,
+      "loss": 0.9071,
+      "step": 6523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965878712566407e-06,
+      "loss": 0.802,
+      "step": 6524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963177783230266e-06,
+      "loss": 0.8387,
+      "step": 6525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960476930350267e-06,
+      "loss": 0.8826,
+      "step": 6526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957776154125573e-06,
+      "loss": 0.9525,
+      "step": 6527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955075454755349e-06,
+      "loss": 0.8698,
+      "step": 6528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95237483243874e-06,
+      "loss": 0.9561,
+      "step": 6529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949674287374895e-06,
+      "loss": 0.8694,
+      "step": 6530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946973819762952e-06,
+      "loss": 0.8505,
+      "step": 6531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94427342980205e-06,
+      "loss": 0.7938,
+      "step": 6532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941573117691315e-06,
+      "loss": 0.7326,
+      "step": 6533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93887288362987e-06,
+      "loss": 0.9851,
+      "step": 6534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93617272781684e-06,
+      "loss": 0.8597,
+      "step": 6535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933472650451322e-06,
+      "loss": 0.9368,
+      "step": 6536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930772651732435e-06,
+      "loss": 0.8728,
+      "step": 6537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928072731859272e-06,
+      "loss": 0.784,
+      "step": 6538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925372891030931e-06,
+      "loss": 0.8621,
+      "step": 6539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922673129446497e-06,
+      "loss": 0.9536,
+      "step": 6540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919973447305057e-06,
+      "loss": 0.8187,
+      "step": 6541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917273844805687e-06,
+      "loss": 0.8447,
+      "step": 6542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.914574322147455e-06,
+      "loss": 0.9254,
+      "step": 6543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911874879529428e-06,
+      "loss": 0.8738,
+      "step": 6544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90917551715067e-06,
+      "loss": 0.8699,
+      "step": 6545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906476235210224e-06,
+      "loss": 0.9302,
+      "step": 6546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903777033907145e-06,
+      "loss": 0.9476,
+      "step": 6547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901077913440472e-06,
+      "loss": 0.8474,
+      "step": 6548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898378874009236e-06,
+      "loss": 0.8602,
+      "step": 6549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895679915812476e-06,
+      "loss": 0.8795,
+      "step": 6550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892981039049209e-06,
+      "loss": 0.9229,
+      "step": 6551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890282243918453e-06,
+      "loss": 0.8613,
+      "step": 6552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88758353061922e-06,
+      "loss": 0.8995,
+      "step": 6553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884884899350518e-06,
+      "loss": 0.7972,
+      "step": 6554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882186350311342e-06,
+      "loss": 0.9615,
+      "step": 6555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879487883700692e-06,
+      "loss": 0.905,
+      "step": 6556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876789499717546e-06,
+      "loss": 0.8058,
+      "step": 6557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87409119856089e-06,
+      "loss": 0.8802,
+      "step": 6558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871392980429699e-06,
+      "loss": 0.8591,
+      "step": 6559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86869484552294e-06,
+      "loss": 0.8996,
+      "step": 6560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865996794039578e-06,
+      "loss": 0.9339,
+      "step": 6561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86329882617857e-06,
+      "loss": 0.8492,
+      "step": 6562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860600942138865e-06,
+      "loss": 0.9454,
+      "step": 6563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857903142119408e-06,
+      "loss": 0.8626,
+      "step": 6564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855205426319133e-06,
+      "loss": 0.8308,
+      "step": 6565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852507794936983e-06,
+      "loss": 0.9543,
+      "step": 6566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84981024817187e-06,
+      "loss": 0.9535,
+      "step": 6567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84711278622272e-06,
+      "loss": 0.8823,
+      "step": 6568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844415409288447e-06,
+      "loss": 0.2734,
+      "step": 6569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841718117567954e-06,
+      "loss": 0.9172,
+      "step": 6570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839020911260145e-06,
+      "loss": 0.9608,
+      "step": 6571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836323790563914e-06,
+      "loss": 0.8494,
+      "step": 6572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83362675567815e-06,
+      "loss": 0.8473,
+      "step": 6573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830929806801728e-06,
+      "loss": 0.9634,
+      "step": 6574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82823294413353e-06,
+      "loss": 0.8277,
+      "step": 6575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825536167872425e-06,
+      "loss": 0.9268,
+      "step": 6576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822839478217275e-06,
+      "loss": 0.9554,
+      "step": 6577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820142875366932e-06,
+      "loss": 0.8963,
+      "step": 6578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817446359520245e-06,
+      "loss": 0.8337,
+      "step": 6579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814749930876066e-06,
+      "loss": 0.9316,
+      "step": 6580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812053589633226e-06,
+      "loss": 0.943,
+      "step": 6581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.809357335990553e-06,
+      "loss": 0.8028,
+      "step": 6582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806661170146876e-06,
+      "loss": 0.8776,
+      "step": 6583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803965092301011e-06,
+      "loss": 0.8761,
+      "step": 6584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801269102651769e-06,
+      "loss": 0.9008,
+      "step": 6585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798573201397953e-06,
+      "loss": 0.8417,
+      "step": 6586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795877388738368e-06,
+      "loss": 0.8782,
+      "step": 6587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793181664871795e-06,
+      "loss": 0.9482,
+      "step": 6588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790486029997025e-06,
+      "loss": 0.8922,
+      "step": 6589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787790484312834e-06,
+      "loss": 0.8851,
+      "step": 6590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785095028017998e-06,
+      "loss": 0.8813,
+      "step": 6591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782399661311275e-06,
+      "loss": 0.9909,
+      "step": 6592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779704384391432e-06,
+      "loss": 0.9971,
+      "step": 6593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777009197457214e-06,
+      "loss": 0.8613,
+      "step": 6594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774314100707369e-06,
+      "loss": 0.7376,
+      "step": 6595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771619094340637e-06,
+      "loss": 0.9175,
+      "step": 6596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768924178555754e-06,
+      "loss": 0.8393,
+      "step": 6597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766229353551435e-06,
+      "loss": 0.9431,
+      "step": 6598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763534619526404e-06,
+      "loss": 0.8799,
+      "step": 6599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760839976679376e-06,
+      "loss": 0.8919,
+      "step": 6600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758145425209049e-06,
+      "loss": 0.7598,
+      "step": 6601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755450965314128e-06,
+      "loss": 0.2889,
+      "step": 6602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752756597193303e-06,
+      "loss": 0.9492,
+      "step": 6603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750062321045256e-06,
+      "loss": 0.8762,
+      "step": 6604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74736813706867e-06,
+      "loss": 0.8245,
+      "step": 6605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744674045462213e-06,
+      "loss": 0.949,
+      "step": 6606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741980046424553e-06,
+      "loss": 0.9431,
+      "step": 6607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739286140154345e-06,
+      "loss": 0.9039,
+      "step": 6608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73659232685024e-06,
+      "loss": 0.9385,
+      "step": 6609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733898606710878e-06,
+      "loss": 0.956,
+      "step": 6610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731204979934905e-06,
+      "loss": 0.9406,
+      "step": 6611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728511446720946e-06,
+      "loss": 0.8048,
+      "step": 6612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725818007267624e-06,
+      "loss": 0.9094,
+      "step": 6613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723124661773559e-06,
+      "loss": 0.8546,
+      "step": 6614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720431410437357e-06,
+      "loss": 0.8716,
+      "step": 6615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717738253457623e-06,
+      "loss": 0.8366,
+      "step": 6616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715045191032948e-06,
+      "loss": 0.9211,
+      "step": 6617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712352223361933e-06,
+      "loss": 0.92,
+      "step": 6618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709659350643144e-06,
+      "loss": 0.8014,
+      "step": 6619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706966573075164e-06,
+      "loss": 0.8257,
+      "step": 6620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70427389085656e-06,
+      "loss": 0.9026,
+      "step": 6621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70158130418589e-06,
+      "loss": 0.8608,
+      "step": 6622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698888813261713e-06,
+      "loss": 0.8431,
+      "step": 6623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696196418282571e-06,
+      "loss": 0.7747,
+      "step": 6624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693504119447006e-06,
+      "loss": 0.8587,
+      "step": 6625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690811916953545e-06,
+      "loss": 0.9207,
+      "step": 6626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68811981100072e-06,
+      "loss": 0.8526,
+      "step": 6627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68542780178705e-06,
+      "loss": 0.9413,
+      "step": 6628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68273588951104e-06,
+      "loss": 0.7584,
+      "step": 6629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680044074371196e-06,
+      "loss": 0.9012,
+      "step": 6630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677352356566013e-06,
+      "loss": 0.8189,
+      "step": 6631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674660736293985e-06,
+      "loss": 0.8398,
+      "step": 6632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671969213753594e-06,
+      "loss": 0.8378,
+      "step": 6633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66927778914331e-06,
+      "loss": 0.9914,
+      "step": 6634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666586462661604e-06,
+      "loss": 0.8383,
+      "step": 6635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663895234506938e-06,
+      "loss": 0.9117,
+      "step": 6636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661204104877765e-06,
+      "loss": 0.8951,
+      "step": 6637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658513073972533e-06,
+      "loss": 0.2243,
+      "step": 6638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.655822141989676e-06,
+      "loss": 1.0009,
+      "step": 6639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653131309127625e-06,
+      "loss": 0.9051,
+      "step": 6640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65044057558481e-06,
+      "loss": 0.8301,
+      "step": 6641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647749941559647e-06,
+      "loss": 0.8709,
+      "step": 6642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645059407250542e-06,
+      "loss": 0.7863,
+      "step": 6643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.642368972855896e-06,
+      "loss": 0.894,
+      "step": 6644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63967863857411e-06,
+      "loss": 0.9464,
+      "step": 6645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636988404603569e-06,
+      "loss": 0.8779,
+      "step": 6646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634298271142649e-06,
+      "loss": 0.7926,
+      "step": 6647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631608238389731e-06,
+      "loss": 0.9269,
+      "step": 6648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628918306543176e-06,
+      "loss": 0.8842,
+      "step": 6649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62622847580134e-06,
+      "loss": 0.966,
+      "step": 6650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623538746362574e-06,
+      "loss": 0.8813,
+      "step": 6651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620849118425223e-06,
+      "loss": 0.8773,
+      "step": 6652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618159592187618e-06,
+      "loss": 0.9178,
+      "step": 6653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615470167848092e-06,
+      "loss": 0.9189,
+      "step": 6654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612780845604965e-06,
+      "loss": 0.8956,
+      "step": 6655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610091625656544e-06,
+      "loss": 0.8582,
+      "step": 6656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607402508201144e-06,
+      "loss": 0.923,
+      "step": 6657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604713493437056e-06,
+      "loss": 0.8215,
+      "step": 6658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602024581562573e-06,
+      "loss": 0.838,
+      "step": 6659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599335772775976e-06,
+      "loss": 0.9203,
+      "step": 6660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59664706727554e-06,
+      "loss": 0.9221,
+      "step": 6661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593958465259529e-06,
+      "loss": 0.7772,
+      "step": 6662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591269966926211e-06,
+      "loss": 0.8978,
+      "step": 6663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588581572473834e-06,
+      "loss": 0.8923,
+      "step": 6664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58589328210064e-06,
+      "loss": 0.8667,
+      "step": 6665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583205096004867e-06,
+      "loss": 0.9087,
+      "step": 6666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580517014384749e-06,
+      "loss": 0.8307,
+      "step": 6667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577829037438503e-06,
+      "loss": 0.8184,
+      "step": 6668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575141165364344e-06,
+      "loss": 0.8789,
+      "step": 6669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572453398360475e-06,
+      "loss": 0.936,
+      "step": 6670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569765736625096e-06,
+      "loss": 0.9013,
+      "step": 6671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567078180356397e-06,
+      "loss": 0.8927,
+      "step": 6672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564390729752564e-06,
+      "loss": 0.8623,
+      "step": 6673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561703385011767e-06,
+      "loss": 0.8443,
+      "step": 6674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559016146332178e-06,
+      "loss": 0.9425,
+      "step": 6675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556329013911951e-06,
+      "loss": 0.9103,
+      "step": 6676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553641987949242e-06,
+      "loss": 0.8664,
+      "step": 6677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550955068642192e-06,
+      "loss": 0.8639,
+      "step": 6678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548268256188937e-06,
+      "loss": 0.8817,
+      "step": 6679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545581550787609e-06,
+      "loss": 0.8123,
+      "step": 6680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542894952636322e-06,
+      "loss": 0.9184,
+      "step": 6681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540208461933189e-06,
+      "loss": 0.8637,
+      "step": 6682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537522078876314e-06,
+      "loss": 0.9412,
+      "step": 6683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534835803663797e-06,
+      "loss": 0.9102,
+      "step": 6684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532149636493723e-06,
+      "loss": 0.8152,
+      "step": 6685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529463577564173e-06,
+      "loss": 0.8495,
+      "step": 6686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526777627073216e-06,
+      "loss": 1.0449,
+      "step": 6687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524091785218921e-06,
+      "loss": 0.9294,
+      "step": 6688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521406052199345e-06,
+      "loss": 0.8494,
+      "step": 6689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518720428212533e-06,
+      "loss": 0.2539,
+      "step": 6690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516034913456526e-06,
+      "loss": 0.9327,
+      "step": 6691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513349508129354e-06,
+      "loss": 0.8317,
+      "step": 6692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510664212429044e-06,
+      "loss": 0.871,
+      "step": 6693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507979026553612e-06,
+      "loss": 0.8386,
+      "step": 6694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505293950701065e-06,
+      "loss": 0.8909,
+      "step": 6695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5026089850694e-06,
+      "loss": 0.9164,
+      "step": 6696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499924129856614e-06,
+      "loss": 0.8836,
+      "step": 6697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497239385260687e-06,
+      "loss": 0.9505,
+      "step": 6698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494554751479593e-06,
+      "loss": 0.9665,
+      "step": 6699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49187022871131e-06,
+      "loss": 1.0304,
+      "step": 6700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48918581715378e-06,
+      "loss": 0.831,
+      "step": 6701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486501517004964e-06,
+      "loss": 0.9046,
+      "step": 6702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483817328462804e-06,
+      "loss": 0.83,
+      "step": 6703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481133251725232e-06,
+      "loss": 0.8639,
+      "step": 6704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478449286990172e-06,
+      "loss": 0.8498,
+      "step": 6705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475765434455549e-06,
+      "loss": 0.931,
+      "step": 6706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473081694319268e-06,
+      "loss": 1.0171,
+      "step": 6707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470398066779226e-06,
+      "loss": 0.9103,
+      "step": 6708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467714552033325e-06,
+      "loss": 0.9052,
+      "step": 6709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465031150279446e-06,
+      "loss": 0.8737,
+      "step": 6710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462347861715466e-06,
+      "loss": 0.9043,
+      "step": 6711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459664686539248e-06,
+      "loss": 0.8356,
+      "step": 6712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456981624948657e-06,
+      "loss": 0.9002,
+      "step": 6713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45429867714154e-06,
+      "loss": 0.8792,
+      "step": 6714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451615843315743e-06,
+      "loss": 0.9141,
+      "step": 6715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448933123669102e-06,
+      "loss": 0.8453,
+      "step": 6716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446250518399436e-06,
+      "loss": 0.9529,
+      "step": 6717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443568027704571e-06,
+      "loss": 0.8832,
+      "step": 6718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440885651782312e-06,
+      "loss": 0.9408,
+      "step": 6719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438203390830461e-06,
+      "loss": 0.8601,
+      "step": 6720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43552124504681e-06,
+      "loss": 0.8985,
+      "step": 6721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43283921462914e-06,
+      "loss": 0.7501,
+      "step": 6722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430157299775226e-06,
+      "loss": 0.8299,
+      "step": 6723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427475500682839e-06,
+      "loss": 0.8776,
+      "step": 6724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424793817549735e-06,
+      "loss": 0.9311,
+      "step": 6725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42211225057366e-06,
+      "loss": 0.935,
+      "step": 6726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419430799952362e-06,
+      "loss": 0.9277,
+      "step": 6727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41674946588357e-06,
+      "loss": 0.9351,
+      "step": 6728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414068248565007e-06,
+      "loss": 0.8623,
+      "step": 6729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411387148194387e-06,
+      "loss": 0.9467,
+      "step": 6730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408706164969425e-06,
+      "loss": 0.9531,
+      "step": 6731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406025299087807e-06,
+      "loss": 0.927,
+      "step": 6732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40334455074723e-06,
+      "loss": 0.834,
+      "step": 6733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40066392014537e-06,
+      "loss": 0.9573,
+      "step": 6734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397983407479902e-06,
+      "loss": 0.8305,
+      "step": 6735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39530301294849e-06,
+      "loss": 0.8657,
+      "step": 6736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.392622736748785e-06,
+      "loss": 0.8111,
+      "step": 6737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389942579078439e-06,
+      "loss": 0.8814,
+      "step": 6738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.38726254013508e-06,
+      "loss": 0.8429,
+      "step": 6739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384582620116344e-06,
+      "loss": 0.824,
+      "step": 6740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381902819219849e-06,
+      "loss": 0.9203,
+      "step": 6741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379223137643207e-06,
+      "loss": 0.8275,
+      "step": 6742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376543575584015e-06,
+      "loss": 0.8589,
+      "step": 6743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373864133239868e-06,
+      "loss": 0.8558,
+      "step": 6744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371184810808353e-06,
+      "loss": 0.8752,
+      "step": 6745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368505608487043e-06,
+      "loss": 0.9248,
+      "step": 6746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365826526473506e-06,
+      "loss": 0.8579,
+      "step": 6747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363147564965297e-06,
+      "loss": 0.9272,
+      "step": 6748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360468724159971e-06,
+      "loss": 0.9118,
+      "step": 6749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357790004255062e-06,
+      "loss": 0.8954,
+      "step": 6750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355111405448101e-06,
+      "loss": 0.9624,
+      "step": 6751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352432927936619e-06,
+      "loss": 0.2404,
+      "step": 6752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349754571918117e-06,
+      "loss": 0.8607,
+      "step": 6753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347076337590105e-06,
+      "loss": 0.831,
+      "step": 6754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344398225150079e-06,
+      "loss": 0.9406,
+      "step": 6755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341720234795525e-06,
+      "loss": 0.8567,
+      "step": 6756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339042366723917e-06,
+      "loss": 0.9304,
+      "step": 6757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336364621132726e-06,
+      "loss": 0.9087,
+      "step": 6758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333686998219414e-06,
+      "loss": 0.8142,
+      "step": 6759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331009498181423e-06,
+      "loss": 0.8879,
+      "step": 6760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328332121216204e-06,
+      "loss": 0.7042,
+      "step": 6761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325654867521185e-06,
+      "loss": 0.9053,
+      "step": 6762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322977737293786e-06,
+      "loss": 0.8139,
+      "step": 6763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320300730731424e-06,
+      "loss": 0.9034,
+      "step": 6764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317623848031504e-06,
+      "loss": 0.8969,
+      "step": 6765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314947089391415e-06,
+      "loss": 0.8907,
+      "step": 6766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312270455008553e-06,
+      "loss": 0.8096,
+      "step": 6767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30959394508029e-06,
+      "loss": 0.7754,
+      "step": 6768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306917559803993e-06,
+      "loss": 0.8862,
+      "step": 6769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304241299377027e-06,
+      "loss": 0.925,
+      "step": 6770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301565163996737e-06,
+      "loss": 0.9089,
+      "step": 6771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298889153860465e-06,
+      "loss": 0.8889,
+      "step": 6772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296213269165542e-06,
+      "loss": 0.8951,
+      "step": 6773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293537510109288e-06,
+      "loss": 0.98,
+      "step": 6774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290861876889016e-06,
+      "loss": 0.9594,
+      "step": 6775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28818636970203e-06,
+      "loss": 0.8652,
+      "step": 6776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285510988745628e-06,
+      "loss": 0.857,
+      "step": 6777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28283573421709e-06,
+      "loss": 0.9182,
+      "step": 6778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280160606313693e-06,
+      "loss": 0.8157,
+      "step": 6779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277485605232704e-06,
+      "loss": 0.854,
+      "step": 6780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27481073117138e-06,
+      "loss": 0.8732,
+      "step": 6781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272135984326965e-06,
+      "loss": 0.9116,
+      "step": 6782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269461364896707e-06,
+      "loss": 0.9347,
+      "step": 6783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266786873077821e-06,
+      "loss": 0.8867,
+      "step": 6784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264112509067534e-06,
+      "loss": 0.8382,
+      "step": 6785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261438273063055e-06,
+      "loss": 0.8752,
+      "step": 6786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258764165261582e-06,
+      "loss": 0.8839,
+      "step": 6787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256090185860311e-06,
+      "loss": 0.8616,
+      "step": 6788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25341633505642e-06,
+      "loss": 0.9553,
+      "step": 6789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25074261304708e-06,
+      "loss": 0.9122,
+      "step": 6790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248069020029456e-06,
+      "loss": 0.8254,
+      "step": 6791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2453955562007e-06,
+      "loss": 0.9253,
+      "step": 6792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242722221757959e-06,
+      "loss": 0.8607,
+      "step": 6793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24004901689836e-06,
+      "loss": 0.8766,
+      "step": 6794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237375941819031e-06,
+      "loss": 0.9154,
+      "step": 6795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234702996717086e-06,
+      "loss": 0.9842,
+      "step": 6796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232030181789632e-06,
+      "loss": 0.9081,
+      "step": 6797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229357497233763e-06,
+      "loss": 0.911,
+      "step": 6798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226684943246566e-06,
+      "loss": 0.2337,
+      "step": 6799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224012520025115e-06,
+      "loss": 0.9239,
+      "step": 6800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221340227766482e-06,
+      "loss": 0.8578,
+      "step": 6801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218668066667718e-06,
+      "loss": 0.8902,
+      "step": 6802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215996036925878e-06,
+      "loss": 0.2347,
+      "step": 6803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21332413873799e-06,
+      "loss": 0.836,
+      "step": 6804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210652372301087e-06,
+      "loss": 0.9445,
+      "step": 6805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207980737812188e-06,
+      "loss": 0.9604,
+      "step": 6806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205309235468302e-06,
+      "loss": 0.9621,
+      "step": 6807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202637865466427e-06,
+      "loss": 0.893,
+      "step": 6808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199966628003547e-06,
+      "loss": 0.9015,
+      "step": 6809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19729552327665e-06,
+      "loss": 0.8809,
+      "step": 6810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194624551482701e-06,
+      "loss": 0.8472,
+      "step": 6811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19195371281866e-06,
+      "loss": 0.8788,
+      "step": 6812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189283007481476e-06,
+      "loss": 0.8533,
+      "step": 6813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186612435668095e-06,
+      "loss": 0.817,
+      "step": 6814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183941997575439e-06,
+      "loss": 0.9343,
+      "step": 6815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181271693400434e-06,
+      "loss": 0.9407,
+      "step": 6816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178601523339987e-06,
+      "loss": 0.9022,
+      "step": 6817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175931487590996e-06,
+      "loss": 0.8718,
+      "step": 6818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173261586350359e-06,
+      "loss": 0.8191,
+      "step": 6819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170591819814953e-06,
+      "loss": 1.0076,
+      "step": 6820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167922188181648e-06,
+      "loss": 0.8943,
+      "step": 6821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165252691647308e-06,
+      "loss": 0.8516,
+      "step": 6822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16258333040878e-06,
+      "loss": 0.8905,
+      "step": 6823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15991410466291e-06,
+      "loss": 0.853,
+      "step": 6824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157245014606526e-06,
+      "loss": 0.7205,
+      "step": 6825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154576060436447e-06,
+      "loss": 0.9644,
+      "step": 6826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151907242349482e-06,
+      "loss": 0.8141,
+      "step": 6827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14923856054244e-06,
+      "loss": 0.9163,
+      "step": 6828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146570015212107e-06,
+      "loss": 0.8938,
+      "step": 6829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143901606555262e-06,
+      "loss": 0.9117,
+      "step": 6830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14123333476868e-06,
+      "loss": 0.8758,
+      "step": 6831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13856520004912e-06,
+      "loss": 0.8379,
+      "step": 6832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135897202593333e-06,
+      "loss": 0.9182,
+      "step": 6833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133229342598059e-06,
+      "loss": 0.9083,
+      "step": 6834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130561620260026e-06,
+      "loss": 0.8778,
+      "step": 6835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127894035775955e-06,
+      "loss": 0.9101,
+      "step": 6836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125226589342557e-06,
+      "loss": 0.881,
+      "step": 6837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122559281156534e-06,
+      "loss": 0.9454,
+      "step": 6838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119892111414568e-06,
+      "loss": 0.8613,
+      "step": 6839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117225080313347e-06,
+      "loss": 0.7842,
+      "step": 6840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114558188049538e-06,
+      "loss": 0.8708,
+      "step": 6841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111891434819796e-06,
+      "loss": 0.9264,
+      "step": 6842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10922482082077e-06,
+      "loss": 0.9193,
+      "step": 6843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106558346249103e-06,
+      "loss": 0.2565,
+      "step": 6844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103892011301424e-06,
+      "loss": 0.854,
+      "step": 6845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101225816174344e-06,
+      "loss": 0.7473,
+      "step": 6846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098559761064473e-06,
+      "loss": 0.259,
+      "step": 6847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095893846168409e-06,
+      "loss": 0.8929,
+      "step": 6848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093228071682739e-06,
+      "loss": 0.9851,
+      "step": 6849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090562437804039e-06,
+      "loss": 0.8403,
+      "step": 6850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087896944728877e-06,
+      "loss": 0.9259,
+      "step": 6851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085231592653803e-06,
+      "loss": 0.8618,
+      "step": 6852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082566381775369e-06,
+      "loss": 0.917,
+      "step": 6853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079901312290109e-06,
+      "loss": 0.8301,
+      "step": 6854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077236384394547e-06,
+      "loss": 0.9203,
+      "step": 6855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074571598285193e-06,
+      "loss": 0.8856,
+      "step": 6856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07190695415855e-06,
+      "loss": 0.9516,
+      "step": 6857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069242452211118e-06,
+      "loss": 0.8743,
+      "step": 6858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066578092639375e-06,
+      "loss": 0.9229,
+      "step": 6859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063913875639796e-06,
+      "loss": 0.8521,
+      "step": 6860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061249801408836e-06,
+      "loss": 0.8828,
+      "step": 6861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058585870142954e-06,
+      "loss": 0.2843,
+      "step": 6862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055922082038587e-06,
+      "loss": 0.8445,
+      "step": 6863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053258437292165e-06,
+      "loss": 0.9601,
+      "step": 6864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050594936100114e-06,
+      "loss": 0.9228,
+      "step": 6865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047931578658828e-06,
+      "loss": 0.8644,
+      "step": 6866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045268365164718e-06,
+      "loss": 0.8477,
+      "step": 6867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042605295814165e-06,
+      "loss": 0.942,
+      "step": 6868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03994237080355e-06,
+      "loss": 0.813,
+      "step": 6869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037279590329237e-06,
+      "loss": 0.882,
+      "step": 6870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034616954587581e-06,
+      "loss": 0.8255,
+      "step": 6871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031954463774931e-06,
+      "loss": 0.8819,
+      "step": 6872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029292118087615e-06,
+      "loss": 0.9042,
+      "step": 6873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026629917721965e-06,
+      "loss": 0.8479,
+      "step": 6874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023967862874287e-06,
+      "loss": 0.8548,
+      "step": 6875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021305953740888e-06,
+      "loss": 0.8057,
+      "step": 6876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018644190518056e-06,
+      "loss": 0.871,
+      "step": 6877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01598257340207e-06,
+      "loss": 0.8695,
+      "step": 6878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013321102589203e-06,
+      "loss": 0.9052,
+      "step": 6879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010659778275714e-06,
+      "loss": 0.2921,
+      "step": 6880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007998600657851e-06,
+      "loss": 0.7435,
+      "step": 6881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005337569931849e-06,
+      "loss": 0.8565,
+      "step": 6882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002676686293939e-06,
+      "loss": 0.8612,
+      "step": 6883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000015949940336e-06,
+      "loss": 0.2962,
+      "step": 6884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997355361067243e-06,
+      "loss": 0.8928,
+      "step": 6885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994694919870856e-06,
+      "loss": 0.9971,
+      "step": 6886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992034626547356e-06,
+      "loss": 0.2752,
+      "step": 6887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989374481292913e-06,
+      "loss": 0.9196,
+      "step": 6888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986714484303694e-06,
+      "loss": 0.9326,
+      "step": 6889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984054635775847e-06,
+      "loss": 0.9327,
+      "step": 6890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981394935905509e-06,
+      "loss": 0.89,
+      "step": 6891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978735384888815e-06,
+      "loss": 0.9566,
+      "step": 6892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976075982921875e-06,
+      "loss": 0.865,
+      "step": 6893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973416730200803e-06,
+      "loss": 0.9173,
+      "step": 6894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970757626921686e-06,
+      "loss": 0.8906,
+      "step": 6895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96809867328062e-06,
+      "loss": 0.8538,
+      "step": 6896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965439869473664e-06,
+      "loss": 0.9492,
+      "step": 6897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962781215696892e-06,
+      "loss": 0.8369,
+      "step": 6898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960122712146349e-06,
+      "loss": 0.9147,
+      "step": 6899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957464359018077e-06,
+      "loss": 0.8723,
+      "step": 6900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95480615650811e-06,
+      "loss": 0.8598,
+      "step": 6901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95214810481246e-06,
+      "loss": 0.2583,
+      "step": 6902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949490204127136e-06,
+      "loss": 0.8973,
+      "step": 6903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946832454648132e-06,
+      "loss": 0.8165,
+      "step": 6904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944174856571434e-06,
+      "loss": 0.9056,
+      "step": 6905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941517410093019e-06,
+      "loss": 0.8546,
+      "step": 6906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938860115408849e-06,
+      "loss": 0.9597,
+      "step": 6907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936202972714868e-06,
+      "loss": 0.9003,
+      "step": 6908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93354598220702e-06,
+      "loss": 0.9176,
+      "step": 6909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930889144081235e-06,
+      "loss": 0.8315,
+      "step": 6910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928232458533431e-06,
+      "loss": 0.8697,
+      "step": 6911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925575925759512e-06,
+      "loss": 0.8108,
+      "step": 6912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922919545955373e-06,
+      "loss": 0.916,
+      "step": 6913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9202633193169e-06,
+      "loss": 0.9449,
+      "step": 6914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917607246039962e-06,
+      "loss": 0.8451,
+      "step": 6915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914951326320421e-06,
+      "loss": 0.8129,
+      "step": 6916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912295560354135e-06,
+      "loss": 0.8546,
+      "step": 6917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909639948336929e-06,
+      "loss": 0.2691,
+      "step": 6918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906984490464637e-06,
+      "loss": 0.8521,
+      "step": 6919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904329186933074e-06,
+      "loss": 0.9036,
+      "step": 6920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901674037938045e-06,
+      "loss": 0.8533,
+      "step": 6921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89901904367534e-06,
+      "loss": 0.7784,
+      "step": 6922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896364204340745e-06,
+      "loss": 0.8831,
+      "step": 6923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893709520130027e-06,
+      "loss": 0.8611,
+      "step": 6924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891054991238945e-06,
+      "loss": 0.9122,
+      "step": 6925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888400617863247e-06,
+      "loss": 0.8221,
+      "step": 6926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885746400198672e-06,
+      "loss": 0.9141,
+      "step": 6927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883092338440937e-06,
+      "loss": 0.9055,
+      "step": 6928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880438432785759e-06,
+      "loss": 0.9081,
+      "step": 6929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877784683428839e-06,
+      "loss": 0.2722,
+      "step": 6930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875131090565864e-06,
+      "loss": 0.8848,
+      "step": 6931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872477654392517e-06,
+      "loss": 0.8942,
+      "step": 6932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869824375104462e-06,
+      "loss": 0.8472,
+      "step": 6933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867171252897352e-06,
+      "loss": 0.2374,
+      "step": 6934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864518287966835e-06,
+      "loss": 0.9437,
+      "step": 6935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861865480508541e-06,
+      "loss": 0.8738,
+      "step": 6936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859212830718085e-06,
+      "loss": 0.8768,
+      "step": 6937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856560338791091e-06,
+      "loss": 0.8426,
+      "step": 6938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853908004923136e-06,
+      "loss": 0.9165,
+      "step": 6939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851255829309817e-06,
+      "loss": 0.2955,
+      "step": 6940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848603812146707e-06,
+      "loss": 0.93,
+      "step": 6941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845951953629365e-06,
+      "loss": 1.0025,
+      "step": 6942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84330025395334e-06,
+      "loss": 0.8951,
+      "step": 6943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840648713314179e-06,
+      "loss": 0.2171,
+      "step": 6944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8379973319074e-06,
+      "loss": 0.8175,
+      "step": 6945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835346109928519e-06,
+      "loss": 0.845,
+      "step": 6946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832695047573045e-06,
+      "loss": 0.9151,
+      "step": 6947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830044145036467e-06,
+      "loss": 0.8699,
+      "step": 6948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827393402514263e-06,
+      "loss": 0.8118,
+      "step": 6949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824742820201901e-06,
+      "loss": 0.8836,
+      "step": 6950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82209239829484e-06,
+      "loss": 0.837,
+      "step": 6951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819442136988518e-06,
+      "loss": 0.8656,
+      "step": 6952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816792036478374e-06,
+      "loss": 0.9359,
+      "step": 6953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814142096959828e-06,
+      "loss": 0.8786,
+      "step": 6954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811492318628285e-06,
+      "loss": 0.2545,
+      "step": 6955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808842701679146e-06,
+      "loss": 0.9117,
+      "step": 6956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806193246307793e-06,
+      "loss": 0.8584,
+      "step": 6957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.803543952709604e-06,
+      "loss": 0.8237,
+      "step": 6958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800894821079934e-06,
+      "loss": 0.9189,
+      "step": 6959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798245851614136e-06,
+      "loss": 0.8228,
+      "step": 6960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795597044507542e-06,
+      "loss": 0.8269,
+      "step": 6961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792948399955484e-06,
+      "loss": 0.9017,
+      "step": 6962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790299918153272e-06,
+      "loss": 0.8942,
+      "step": 6963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787651599296206e-06,
+      "loss": 0.9124,
+      "step": 6964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785003443579578e-06,
+      "loss": 0.8804,
+      "step": 6965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782355451198666e-06,
+      "loss": 0.8906,
+      "step": 6966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779707622348732e-06,
+      "loss": 0.9683,
+      "step": 6967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777059957225029e-06,
+      "loss": 0.8746,
+      "step": 6968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774412456022805e-06,
+      "loss": 0.967,
+      "step": 6969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77176511893728e-06,
+      "loss": 0.9099,
+      "step": 6970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769117946163674e-06,
+      "loss": 0.8906,
+      "step": 6971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766470937897194e-06,
+      "loss": 0.9434,
+      "step": 6972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763824094333028e-06,
+      "loss": 0.8569,
+      "step": 6973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761177415666362e-06,
+      "loss": 0.947,
+      "step": 6974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75853090209236e-06,
+      "loss": 0.8646,
+      "step": 6975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75588455380618e-06,
+      "loss": 0.9375,
+      "step": 6976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753238371002965e-06,
+      "loss": 0.8384,
+      "step": 6977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750592353877848e-06,
+      "loss": 0.8383,
+      "step": 6978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747946502625951e-06,
+      "loss": 0.8782,
+      "step": 6979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745300817442377e-06,
+      "loss": 0.8618,
+      "step": 6980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74265529852222e-06,
+      "loss": 0.2122,
+      "step": 6981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740009946060565e-06,
+      "loss": 0.9123,
+      "step": 6982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737364760252483e-06,
+      "loss": 0.8172,
+      "step": 6983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734719741293035e-06,
+      "loss": 0.8451,
+      "step": 6984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732074889377262e-06,
+      "loss": 0.8199,
+      "step": 6985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729430204700197e-06,
+      "loss": 0.8683,
+      "step": 6986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726785687456866e-06,
+      "loss": 0.8296,
+      "step": 6987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724141337842276e-06,
+      "loss": 0.8888,
+      "step": 6988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721497156051425e-06,
+      "loss": 0.8743,
+      "step": 6989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718853142279295e-06,
+      "loss": 0.89,
+      "step": 6990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716209296720853e-06,
+      "loss": 0.9988,
+      "step": 6991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713565619571068e-06,
+      "loss": 0.9117,
+      "step": 6992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710922111024884e-06,
+      "loss": 0.9142,
+      "step": 6993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708278771277231e-06,
+      "loss": 0.8166,
+      "step": 6994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705635600523034e-06,
+      "loss": 0.8287,
+      "step": 6995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702992598957204e-06,
+      "loss": 0.8762,
+      "step": 6996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700349766774638e-06,
+      "loss": 0.8252,
+      "step": 6997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697707104170216e-06,
+      "loss": 0.8891,
+      "step": 6998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695064611338818e-06,
+      "loss": 0.8039,
+      "step": 6999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692422288475301e-06,
+      "loss": 0.8135,
+      "step": 7000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68978013577451e-06,
+      "loss": 0.8452,
+      "step": 7001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687138153431279e-06,
+      "loss": 0.2715,
+      "step": 7002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68449634164043e-06,
+      "loss": 0.908,
+      "step": 7003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681854700596772e-06,
+      "loss": 0.8829,
+      "step": 7004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679213230495107e-06,
+      "loss": 0.8393,
+      "step": 7005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676571931530216e-06,
+      "loss": 0.244,
+      "step": 7006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673930803896868e-06,
+      "loss": 0.8083,
+      "step": 7007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671289847789824e-06,
+      "loss": 0.8643,
+      "step": 7008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668649063403833e-06,
+      "loss": 0.9442,
+      "step": 7009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666008450933628e-06,
+      "loss": 0.8883,
+      "step": 7010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663368010573927e-06,
+      "loss": 0.8546,
+      "step": 7011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660727742519439e-06,
+      "loss": 0.8835,
+      "step": 7012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658087646964858e-06,
+      "loss": 0.8115,
+      "step": 7013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655447724104871e-06,
+      "loss": 0.8556,
+      "step": 7014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652807974134147e-06,
+      "loss": 0.9485,
+      "step": 7015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65016839724734e-06,
+      "loss": 0.8836,
+      "step": 7016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6475289936391e-06,
+      "loss": 0.9374,
+      "step": 7017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644889763504055e-06,
+      "loss": 0.8929,
+      "step": 7018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642250707036825e-06,
+      "loss": 0.9441,
+      "step": 7019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63961182443202e-06,
+      "loss": 0.7968,
+      "step": 7020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636973115884227e-06,
+      "loss": 0.8083,
+      "step": 7021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634334581588028e-06,
+      "loss": 0.9144,
+      "step": 7022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631696221737993e-06,
+      "loss": 0.9343,
+      "step": 7023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629058036528677e-06,
+      "loss": 0.7807,
+      "step": 7024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62642002615462e-06,
+      "loss": 0.8357,
+      "step": 7025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623782190810352e-06,
+      "loss": 0.8563,
+      "step": 7026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6211445306903904e-06,
+      "loss": 0.8439,
+      "step": 7027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618507045989239e-06,
+      "loss": 0.8678,
+      "step": 7028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6158697369013825e-06,
+      "loss": 0.8248,
+      "step": 7029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613232603621305e-06,
+      "loss": 0.8624,
+      "step": 7030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61059564634347e-06,
+      "loss": 0.9315,
+      "step": 7031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607958865262326e-06,
+      "loss": 0.8504,
+      "step": 7032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605322260572312e-06,
+      "loss": 0.8355,
+      "step": 7033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602685832467853e-06,
+      "loss": 0.7994,
+      "step": 7034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600049581143364e-06,
+      "loss": 0.8749,
+      "step": 7035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597413506793243e-06,
+      "loss": 0.9561,
+      "step": 7036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594777609611877e-06,
+      "loss": 0.8857,
+      "step": 7037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592141889793636e-06,
+      "loss": 0.8607,
+      "step": 7038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5895063475328856e-06,
+      "loss": 0.8116,
+      "step": 7039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586870983023971e-06,
+      "loss": 0.8807,
+      "step": 7040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5842357964612255e-06,
+      "loss": 0.9041,
+      "step": 7041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581600788038969e-06,
+      "loss": 0.9054,
+      "step": 7042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5789659579515075e-06,
+      "loss": 0.8561,
+      "step": 7043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576331306393141e-06,
+      "loss": 0.7958,
+      "step": 7044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573696833558147e-06,
+      "loss": 0.8214,
+      "step": 7045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571062539640796e-06,
+      "loss": 0.8656,
+      "step": 7046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568428424835339e-06,
+      "loss": 0.8,
+      "step": 7047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5657944893360225e-06,
+      "loss": 0.8999,
+      "step": 7048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563160733337075e-06,
+      "loss": 0.8905,
+      "step": 7049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5605271570327065e-06,
+      "loss": 0.9953,
+      "step": 7050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55789376061713e-06,
+      "loss": 0.9153,
+      "step": 7051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555260544284522e-06,
+      "loss": 0.8508,
+      "step": 7052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5526275082290646e-06,
+      "loss": 0.8136,
+      "step": 7053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549994652644918e-06,
+      "loss": 0.82,
+      "step": 7054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547361977726233e-06,
+      "loss": 0.7881,
+      "step": 7055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544729483667143e-06,
+      "loss": 0.8612,
+      "step": 7056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5420971706617715e-06,
+      "loss": 0.9212,
+      "step": 7057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53946503890423e-06,
+      "loss": 0.7981,
+      "step": 7058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5368330885886084e-06,
+      "loss": 0.2517,
+      "step": 7059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534201319908995e-06,
+      "loss": 0.2525,
+      "step": 7060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531569733059456e-06,
+      "loss": 0.8674,
+      "step": 7061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528938328234048e-06,
+      "loss": 0.8886,
+      "step": 7062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526307105626811e-06,
+      "loss": 0.9646,
+      "step": 7063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523676065431773e-06,
+      "loss": 0.8881,
+      "step": 7064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52104520784295e-06,
+      "loss": 0.8908,
+      "step": 7065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518414533054346e-06,
+      "loss": 0.8531,
+      "step": 7066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515784041259946e-06,
+      "loss": 0.9133,
+      "step": 7067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513153732653724e-06,
+      "loss": 0.9533,
+      "step": 7068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510523607429647e-06,
+      "loss": 0.8835,
+      "step": 7069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507893665781658e-06,
+      "loss": 0.9202,
+      "step": 7070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505263907903691e-06,
+      "loss": 0.9175,
+      "step": 7071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5026343339896715e-06,
+      "loss": 0.8193,
+      "step": 7072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5000049442334985e-06,
+      "loss": 0.8814,
+      "step": 7073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497375738829069e-06,
+      "loss": 0.7885,
+      "step": 7074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494746717970265e-06,
+      "loss": 0.8808,
+      "step": 7075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49211788185095e-06,
+      "loss": 0.2676,
+      "step": 7076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489489230664976e-06,
+      "loss": 0.9201,
+      "step": 7077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486860764606185e-06,
+      "loss": 0.9532,
+      "step": 7078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484232483868402e-06,
+      "loss": 0.8408,
+      "step": 7079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481604388645437e-06,
+      "loss": 0.8616,
+      "step": 7080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478976479131086e-06,
+      "loss": 0.9814,
+      "step": 7081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476348755519143e-06,
+      "loss": 0.9426,
+      "step": 7082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473721218003365e-06,
+      "loss": 0.9387,
+      "step": 7083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471093866777516e-06,
+      "loss": 0.8663,
+      "step": 7084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4684667020353385e-06,
+      "loss": 0.9626,
+      "step": 7085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465839723970559e-06,
+      "loss": 0.8671,
+      "step": 7086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4632129327768985e-06,
+      "loss": 0.9385,
+      "step": 7087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460586328648056e-06,
+      "loss": 0.9686,
+      "step": 7088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457959911777719e-06,
+      "loss": 0.9086,
+      "step": 7089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45533368235956e-06,
+      "loss": 0.8321,
+      "step": 7090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452707640587244e-06,
+      "loss": 0.9102,
+      "step": 7091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4500817866544175e-06,
+      "loss": 0.7875,
+      "step": 7092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447456120754708e-06,
+      "loss": 0.8594,
+      "step": 7093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444830643081737e-06,
+      "loss": 0.9568,
+      "step": 7094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442205353829106e-06,
+      "loss": 0.9227,
+      "step": 7095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439580253190414e-06,
+      "loss": 0.86,
+      "step": 7096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436955341359232e-06,
+      "loss": 0.9459,
+      "step": 7097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434330618529125e-06,
+      "loss": 0.8577,
+      "step": 7098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431706084893641e-06,
+      "loss": 0.8986,
+      "step": 7099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4290817406463175e-06,
+      "loss": 0.7981,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426457585980675e-06,
+      "loss": 0.8082,
+      "step": 7101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42383362109022e-06,
+      "loss": 0.8266,
+      "step": 7102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421209846168451e-06,
+      "loss": 0.8277,
+      "step": 7103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418586261408838e-06,
+      "loss": 0.8752,
+      "step": 7104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415962867004851e-06,
+      "loss": 0.881,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4133396631499445e-06,
+      "loss": 0.8638,
+      "step": 7106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410716650037551e-06,
+      "loss": 0.8829,
+      "step": 7107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408093827861094e-06,
+      "loss": 0.8272,
+      "step": 7108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405471196813987e-06,
+      "loss": 0.8589,
+      "step": 7109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402848757089621e-06,
+      "loss": 0.9673,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400226508881377e-06,
+      "loss": 0.75,
+      "step": 7111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397604452382625e-06,
+      "loss": 0.2722,
+      "step": 7112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3949825877867165e-06,
+      "loss": 0.2961,
+      "step": 7113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392360915286989e-06,
+      "loss": 0.9121,
+      "step": 7114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389739435076766e-06,
+      "loss": 0.87,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387118147349357e-06,
+      "loss": 0.8611,
+      "step": 7116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38449705229806e-06,
+      "loss": 0.9353,
+      "step": 7117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381876150116156e-06,
+      "loss": 0.8942,
+      "step": 7118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379255440996915e-06,
+      "loss": 0.8458,
+      "step": 7119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376634925133586e-06,
+      "loss": 0.2657,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374014602719411e-06,
+      "loss": 0.9514,
+      "step": 7121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371394473947614e-06,
+      "loss": 0.9048,
+      "step": 7122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3687745390114076e-06,
+      "loss": 0.9284,
+      "step": 7123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366154798103985e-06,
+      "loss": 0.9512,
+      "step": 7124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363535251418527e-06,
+      "loss": 0.9707,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360915899148204e-06,
+      "loss": 0.8429,
+      "step": 7126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358296741486168e-06,
+      "loss": 0.791,
+      "step": 7127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35567777862556e-06,
+      "loss": 0.9295,
+      "step": 7128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3530590107595e-06,
+      "loss": 0.9115,
+      "step": 7129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350440438081104e-06,
+      "loss": 0.8757,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3478220607834624e-06,
+      "loss": 0.8371,
+      "step": 7131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345203879059661e-06,
+      "loss": 0.8083,
+      "step": 7132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342585893102763e-06,
+      "loss": 0.8985,
+      "step": 7133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3399681031058275e-06,
+      "loss": 0.7751,
+      "step": 7134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337350509261882e-06,
+      "loss": 0.7935,
+      "step": 7135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334733111763957e-06,
+      "loss": 0.8839,
+      "step": 7136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33211591080506e-06,
+      "loss": 0.8683,
+      "step": 7137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3294989065781855e-06,
+      "loss": 0.8305,
+      "step": 7138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3268820992763145e-06,
+      "loss": 0.9051,
+      "step": 7139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324265489092412e-06,
+      "loss": 0.954,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321649076219431e-06,
+      "loss": 0.8309,
+      "step": 7141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3190328608503016e-06,
+      "loss": 0.9065,
+      "step": 7142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316416843177953e-06,
+      "loss": 0.8553,
+      "step": 7143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313801023395293e-06,
+      "loss": 0.8745,
+      "step": 7144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3111854016952085e-06,
+      "loss": 0.9288,
+      "step": 7145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3085699782705795e-06,
+      "loss": 0.9064,
+      "step": 7146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305954753314268e-06,
+      "loss": 0.8531,
+      "step": 7147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30333972701913e-06,
+      "loss": 0.8569,
+      "step": 7148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3007248995779925e-06,
+      "loss": 0.7682,
+      "step": 7149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298110271183678e-06,
+      "loss": 0.9612,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295495842028988e-06,
+      "loss": 0.2208,
+      "step": 7151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292881612306719e-06,
+      "loss": 0.8362,
+      "step": 7152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290267582209642e-06,
+      "loss": 0.8701,
+      "step": 7153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287653751930523e-06,
+      "loss": 0.9146,
+      "step": 7154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285040121662099e-06,
+      "loss": 0.8597,
+      "step": 7155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282426691597107e-06,
+      "loss": 0.8742,
+      "step": 7156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279813461928263e-06,
+      "loss": 0.9245,
+      "step": 7157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.277200432848269e-06,
+      "loss": 0.7932,
+      "step": 7158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274587604549812e-06,
+      "loss": 0.8964,
+      "step": 7159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2719749772255625e-06,
+      "loss": 0.9122,
+      "step": 7160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26936255106818e-06,
+      "loss": 0.8599,
+      "step": 7161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266750326270304e-06,
+      "loss": 0.8788,
+      "step": 7162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264138303024565e-06,
+      "loss": 0.7923,
+      "step": 7163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261526481523576e-06,
+      "loss": 0.8654,
+      "step": 7164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258914861959936e-06,
+      "loss": 0.994,
+      "step": 7165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256303444526225e-06,
+      "loss": 0.9468,
+      "step": 7166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25369222941501e-06,
+      "loss": 0.9483,
+      "step": 7167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251081216818848e-06,
+      "loss": 1.0024,
+      "step": 7168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248470406930274e-06,
+      "loss": 0.9615,
+      "step": 7169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245859799941813e-06,
+      "loss": 0.9189,
+      "step": 7170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243249396045975e-06,
+      "loss": 0.2469,
+      "step": 7171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2406391954352485e-06,
+      "loss": 0.8333,
+      "step": 7172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2380291983021166e-06,
+      "loss": 0.8386,
+      "step": 7173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235419404839041e-06,
+      "loss": 0.8734,
+      "step": 7174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232809815238474e-06,
+      "loss": 0.8725,
+      "step": 7175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230200429692842e-06,
+      "loss": 0.8749,
+      "step": 7176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227591248394568e-06,
+      "loss": 0.8115,
+      "step": 7177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22498227153605e-06,
+      "loss": 0.9155,
+      "step": 7178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2223734993096805e-06,
+      "loss": 0.8322,
+      "step": 7179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219764931907834e-06,
+      "loss": 0.8365,
+      "step": 7180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217156569522863e-06,
+      "loss": 0.2514,
+      "step": 7181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214548412347117e-06,
+      "loss": 0.8879,
+      "step": 7182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21194046057292e-06,
+      "loss": 0.8777,
+      "step": 7183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209332714392583e-06,
+      "loss": 0.8426,
+      "step": 7184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20672517399841e-06,
+      "loss": 0.8025,
+      "step": 7185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204117839582675e-06,
+      "loss": 0.9374,
+      "step": 7186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201510711337647e-06,
+      "loss": 0.9103,
+      "step": 7187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198903789455582e-06,
+      "loss": 0.8994,
+      "step": 7188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196297074128713e-06,
+      "loss": 0.8294,
+      "step": 7189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193690565549261e-06,
+      "loss": 0.9899,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191084263909435e-06,
+      "loss": 0.754,
+      "step": 7191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188478169401425e-06,
+      "loss": 0.8395,
+      "step": 7192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185872282217406e-06,
+      "loss": 0.8203,
+      "step": 7193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1832666025495364e-06,
+      "loss": 0.8564,
+      "step": 7194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180661130589965e-06,
+      "loss": 0.7592,
+      "step": 7195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178055866530823e-06,
+      "loss": 0.8789,
+      "step": 7196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.175450810564218e-06,
+      "loss": 0.8171,
+      "step": 7197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172845962882253e-06,
+      "loss": 0.7918,
+      "step": 7198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170241323677009e-06,
+      "loss": 0.2698,
+      "step": 7199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16763689314056e-06,
+      "loss": 0.8679,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165032671464953e-06,
+      "loss": 1.0192,
+      "step": 7201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162428658842228e-06,
+      "loss": 0.8157,
+      "step": 7202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159824855464407e-06,
+      "loss": 0.8885,
+      "step": 7203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157221261523497e-06,
+      "loss": 0.8496,
+      "step": 7204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154617877211489e-06,
+      "loss": 0.8747,
+      "step": 7205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152014702720361e-06,
+      "loss": 0.8235,
+      "step": 7206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149411738242069e-06,
+      "loss": 0.8823,
+      "step": 7207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146808983968558e-06,
+      "loss": 0.9673,
+      "step": 7208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144206440091759e-06,
+      "loss": 0.9239,
+      "step": 7209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141604106803588e-06,
+      "loss": 0.8936,
+      "step": 7210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139001984295939e-06,
+      "loss": 0.9343,
+      "step": 7211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136400072760696e-06,
+      "loss": 0.8833,
+      "step": 7212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133798372389729e-06,
+      "loss": 0.8575,
+      "step": 7213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131196883374887e-06,
+      "loss": 0.9246,
+      "step": 7214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128595605908004e-06,
+      "loss": 0.9195,
+      "step": 7215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1259945401809094e-06,
+      "loss": 0.9298,
+      "step": 7216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1233936863853945e-06,
+      "loss": 0.8299,
+      "step": 7217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120793044713258e-06,
+      "loss": 0.8528,
+      "step": 7218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11819261535627e-06,
+      "loss": 0.8595,
+      "step": 7219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11559239850619e-06,
+      "loss": 0.8679,
+      "step": 7220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112992394354757e-06,
+      "loss": 0.9133,
+      "step": 7221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110392603093703e-06,
+      "loss": 0.933,
+      "step": 7222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107793024914734e-06,
+      "loss": 0.9267,
+      "step": 7223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105193660009545e-06,
+      "loss": 0.8337,
+      "step": 7224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10259450856982e-06,
+      "loss": 0.9077,
+      "step": 7225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0999955707872185e-06,
+      "loss": 0.7916,
+      "step": 7226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097396846853391e-06,
+      "loss": 0.9525,
+      "step": 7227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094798336959967e-06,
+      "loss": 0.8967,
+      "step": 7228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092200041298565e-06,
+      "loss": 0.9124,
+      "step": 7229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0896019600607816e-06,
+      "loss": 0.8329,
+      "step": 7230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087004093438205e-06,
+      "loss": 0.9162,
+      "step": 7231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084406441622404e-06,
+      "loss": 0.8683,
+      "step": 7232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08180900480493e-06,
+      "loss": 0.8736,
+      "step": 7233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079211783177323e-06,
+      "loss": 0.9961,
+      "step": 7234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076614776931103e-06,
+      "loss": 0.2366,
+      "step": 7235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074017986257774e-06,
+      "loss": 0.872,
+      "step": 7236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071421411348829e-06,
+      "loss": 0.9026,
+      "step": 7237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068825052395737e-06,
+      "loss": 0.2679,
+      "step": 7238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066228909589956e-06,
+      "loss": 0.8063,
+      "step": 7239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063632983122931e-06,
+      "loss": 0.7933,
+      "step": 7240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061037273186086e-06,
+      "loss": 0.8895,
+      "step": 7241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05844177997083e-06,
+      "loss": 0.8954,
+      "step": 7242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055846503668562e-06,
+      "loss": 0.8505,
+      "step": 7243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053251444470653e-06,
+      "loss": 0.8033,
+      "step": 7244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050656602568468e-06,
+      "loss": 0.8954,
+      "step": 7245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048061978153353e-06,
+      "loss": 0.8973,
+      "step": 7246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045467571416642e-06,
+      "loss": 0.9689,
+      "step": 7247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042873382549637e-06,
+      "loss": 0.799,
+      "step": 7248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040279411743647e-06,
+      "loss": 0.8608,
+      "step": 7249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037685659189949e-06,
+      "loss": 0.8608,
+      "step": 7250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0350921250798065e-06,
+      "loss": 0.8715,
+      "step": 7251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0324988096044734e-06,
+      "loss": 0.8641,
+      "step": 7252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029905712955181e-06,
+      "loss": 0.8609,
+      "step": 7253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027312835323145e-06,
+      "loss": 0.9024,
+      "step": 7254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024720176899568e-06,
+      "loss": 0.8966,
+      "step": 7255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022127737875635e-06,
+      "loss": 0.736,
+      "step": 7256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019535518442515e-06,
+      "loss": 0.9657,
+      "step": 7257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016943518791361e-06,
+      "loss": 0.8473,
+      "step": 7258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014351739113307e-06,
+      "loss": 0.9607,
+      "step": 7259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011760179599471e-06,
+      "loss": 0.9148,
+      "step": 7260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0091688404409605e-06,
+      "loss": 0.9027,
+      "step": 7261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006577721828864e-06,
+      "loss": 0.2627,
+      "step": 7262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003986823954252e-06,
+      "loss": 0.8505,
+      "step": 7263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001396147008175e-06,
+      "loss": 0.8712,
+      "step": 7264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998805691181676e-06,
+      "loss": 0.8793,
+      "step": 7265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996215456665778e-06,
+      "loss": 0.9255,
+      "step": 7266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993625443651484e-06,
+      "loss": 0.8983,
+      "step": 7267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99103565232979e-06,
+      "loss": 0.9395,
+      "step": 7268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98844608289166e-06,
+      "loss": 0.8618,
+      "step": 7269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985856735528058e-06,
+      "loss": 0.9106,
+      "step": 7270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983267610429922e-06,
+      "loss": 0.8487,
+      "step": 7271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980678707788177e-06,
+      "loss": 0.9765,
+      "step": 7272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978090027793728e-06,
+      "loss": 0.8348,
+      "step": 7273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975501570637471e-06,
+      "loss": 0.8133,
+      "step": 7274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9729133365102805e-06,
+      "loss": 0.8443,
+      "step": 7275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97032532560301e-06,
+      "loss": 0.8923,
+      "step": 7276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967737538106508e-06,
+      "loss": 0.925,
+      "step": 7277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9651499742116005e-06,
+      "loss": 0.8438,
+      "step": 7278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962562634109089e-06,
+      "loss": 0.8523,
+      "step": 7279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959975517989772e-06,
+      "loss": 0.7784,
+      "step": 7280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957388626044426e-06,
+      "loss": 0.8446,
+      "step": 7281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954801958463805e-06,
+      "loss": 0.8792,
+      "step": 7282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952215515438659e-06,
+      "loss": 0.283,
+      "step": 7283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94962929715971e-06,
+      "loss": 0.8515,
+      "step": 7284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947043303817668e-06,
+      "loss": 0.8405,
+      "step": 7285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94445753560323e-06,
+      "loss": 0.8218,
+      "step": 7286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94187199270707e-06,
+      "loss": 0.8751,
+      "step": 7287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939286675319849e-06,
+      "loss": 0.8445,
+      "step": 7288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936701583632211e-06,
+      "loss": 0.8468,
+      "step": 7289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934116717834781e-06,
+      "loss": 0.9023,
+      "step": 7290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931532078118168e-06,
+      "loss": 0.8167,
+      "step": 7291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928947664672969e-06,
+      "loss": 0.9257,
+      "step": 7292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926363477689759e-06,
+      "loss": 0.9204,
+      "step": 7293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923779517359097e-06,
+      "loss": 0.2662,
+      "step": 7294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92119578387153e-06,
+      "loss": 0.9786,
+      "step": 7295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9186122774175835e-06,
+      "loss": 0.8479,
+      "step": 7296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916028998187765e-06,
+      "loss": 0.8829,
+      "step": 7297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913445946372568e-06,
+      "loss": 0.8515,
+      "step": 7298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910863122162477e-06,
+      "loss": 0.8912,
+      "step": 7299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908280525747937e-06,
+      "loss": 0.8767,
+      "step": 7300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905698157319402e-06,
+      "loss": 0.2151,
+      "step": 7301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903116017067294e-06,
+      "loss": 0.848,
+      "step": 7302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900534105182022e-06,
+      "loss": 0.7917,
+      "step": 7303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897952421853979e-06,
+      "loss": 0.8782,
+      "step": 7304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895370967273543e-06,
+      "loss": 0.2737,
+      "step": 7305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8927897416310695e-06,
+      "loss": 0.2381,
+      "step": 7306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8902087451168986e-06,
+      "loss": 0.8882,
+      "step": 7307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887627977921361e-06,
+      "loss": 0.9124,
+      "step": 7308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885047440234762e-06,
+      "loss": 0.848,
+      "step": 7309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88246713224739e-06,
+      "loss": 0.9211,
+      "step": 7310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8798870541495215e-06,
+      "loss": 0.2394,
+      "step": 7311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877307206131411e-06,
+      "loss": 0.8759,
+      "step": 7312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874727588383303e-06,
+      "loss": 0.7147,
+      "step": 7313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872148201095419e-06,
+      "loss": 0.9294,
+      "step": 7314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869569044457965e-06,
+      "loss": 0.934,
+      "step": 7315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866990118661126e-06,
+      "loss": 0.9125,
+      "step": 7316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864411423895081e-06,
+      "loss": 1.011,
+      "step": 7317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861832960349982e-06,
+      "loss": 0.9303,
+      "step": 7318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859254728215966e-06,
+      "loss": 0.8625,
+      "step": 7319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856676727683162e-06,
+      "loss": 0.8598,
+      "step": 7320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854098958941659e-06,
+      "loss": 0.9229,
+      "step": 7321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851521422181555e-06,
+      "loss": 0.9153,
+      "step": 7322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848944117592918e-06,
+      "loss": 0.8607,
+      "step": 7323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846367045365798e-06,
+      "loss": 0.9094,
+      "step": 7324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843790205690231e-06,
+      "loss": 0.8324,
+      "step": 7325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841213598756238e-06,
+      "loss": 0.7925,
+      "step": 7326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8386372247538165e-06,
+      "loss": 0.9543,
+      "step": 7327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836061083872954e-06,
+      "loss": 0.8344,
+      "step": 7328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833485176303615e-06,
+      "loss": 0.9097,
+      "step": 7329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830909502235752e-06,
+      "loss": 0.7791,
+      "step": 7330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828334061859294e-06,
+      "loss": 0.9243,
+      "step": 7331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8257588553641575e-06,
+      "loss": 0.2462,
+      "step": 7332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823183882940239e-06,
+      "loss": 0.8008,
+      "step": 7333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820609144777419e-06,
+      "loss": 0.8843,
+      "step": 7334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818034641065564e-06,
+      "loss": 0.8768,
+      "step": 7335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815460371994518e-06,
+      "loss": 0.8963,
+      "step": 7336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8128863377541095e-06,
+      "loss": 0.823,
+      "step": 7337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810312538534152e-06,
+      "loss": 0.9637,
+      "step": 7338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807738974524437e-06,
+      "loss": 0.9523,
+      "step": 7339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805165645914745e-06,
+      "loss": 0.8858,
+      "step": 7340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802592552894831e-06,
+      "loss": 0.8789,
+      "step": 7341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800019695654438e-06,
+      "loss": 0.7764,
+      "step": 7342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797447074383292e-06,
+      "loss": 0.8581,
+      "step": 7343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7948746892711004e-06,
+      "loss": 0.8214,
+      "step": 7344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792302540507552e-06,
+      "loss": 0.8948,
+      "step": 7345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789730628282319e-06,
+      "loss": 0.9202,
+      "step": 7346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7871589527850566e-06,
+      "loss": 0.9052,
+      "step": 7347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784587514205404e-06,
+      "loss": 0.8247,
+      "step": 7348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782016312732981e-06,
+      "loss": 0.8307,
+      "step": 7349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77944534855739e-06,
+      "loss": 0.9466,
+      "step": 7350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776874621868214e-06,
+      "loss": 0.9039,
+      "step": 7351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7743041328550195e-06,
+      "loss": 0.8051,
+      "step": 7352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771733881707361e-06,
+      "loss": 0.9106,
+      "step": 7353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769163868614769e-06,
+      "loss": 0.9565,
+      "step": 7354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7665940937667565e-06,
+      "loss": 0.847,
+      "step": 7355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764024557352824e-06,
+      "loss": 0.8648,
+      "step": 7356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761455259562449e-06,
+      "loss": 0.8038,
+      "step": 7357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758886200585095e-06,
+      "loss": 0.899,
+      "step": 7358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756317380610206e-06,
+      "loss": 0.853,
+      "step": 7359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75374879982721e-06,
+      "loss": 0.9859,
+      "step": 7360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751180458425517e-06,
+      "loss": 0.8355,
+      "step": 7361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748612356594516e-06,
+      "loss": 0.9059,
+      "step": 7362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746044494523582e-06,
+      "loss": 0.8805,
+      "step": 7363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743476872402069e-06,
+      "loss": 0.8321,
+      "step": 7364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74090949041932e-06,
+      "loss": 0.8343,
+      "step": 7365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738342348764654e-06,
+      "loss": 0.8408,
+      "step": 7366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735775447627374e-06,
+      "loss": 0.9414,
+      "step": 7367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733208787196763e-06,
+      "loss": 0.8514,
+      "step": 7368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730642367662095e-06,
+      "loss": 0.8381,
+      "step": 7369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7280761892126145e-06,
+      "loss": 0.8697,
+      "step": 7370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725510252037559e-06,
+      "loss": 0.8171,
+      "step": 7371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722944556326136e-06,
+      "loss": 0.8437,
+      "step": 7372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720379102267544e-06,
+      "loss": 0.8847,
+      "step": 7373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717813890050964e-06,
+      "loss": 0.8167,
+      "step": 7374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715248919865558e-06,
+      "loss": 0.9172,
+      "step": 7375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712684191900465e-06,
+      "loss": 0.8661,
+      "step": 7376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710119706344811e-06,
+      "loss": 0.8199,
+      "step": 7377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707555463387707e-06,
+      "loss": 0.8674,
+      "step": 7378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70499146321824e-06,
+      "loss": 0.98,
+      "step": 7379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7024277060254795e-06,
+      "loss": 0.8513,
+      "step": 7380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699864191998486e-06,
+      "loss": 0.8368,
+      "step": 7381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697300921326286e-06,
+      "loss": 0.7906,
+      "step": 7382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694737894197903e-06,
+      "loss": 0.7986,
+      "step": 7383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692175110802336e-06,
+      "loss": 0.8958,
+      "step": 7384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689612571328565e-06,
+      "loss": 0.8772,
+      "step": 7385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687050275965554e-06,
+      "loss": 0.8947,
+      "step": 7386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6844882249022525e-06,
+      "loss": 0.8323,
+      "step": 7387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681926418327584e-06,
+      "loss": 0.242,
+      "step": 7388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679364856430459e-06,
+      "loss": 0.899,
+      "step": 7389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676803539399771e-06,
+      "loss": 0.8194,
+      "step": 7390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674242467424394e-06,
+      "loss": 0.8841,
+      "step": 7391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671681640693184e-06,
+      "loss": 0.7917,
+      "step": 7392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669121059394975e-06,
+      "loss": 0.8897,
+      "step": 7393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6665607237185895e-06,
+      "loss": 0.2377,
+      "step": 7394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6640006338528226e-06,
+      "loss": 0.8247,
+      "step": 7395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661440789986467e-06,
+      "loss": 0.8361,
+      "step": 7396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658881192308284e-06,
+      "loss": 0.9627,
+      "step": 7397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656321841007016e-06,
+      "loss": 0.9375,
+      "step": 7398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653762736271399e-06,
+      "loss": 0.9613,
+      "step": 7399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.651203878290139e-06,
+      "loss": 0.8801,
+      "step": 7400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64864526725193e-06,
+      "loss": 0.8836,
+      "step": 7401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646086903345448e-06,
+      "loss": 0.9843,
+      "step": 7402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643528786759343e-06,
+      "loss": 0.9952,
+      "step": 7403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640970917682255e-06,
+      "loss": 0.8715,
+      "step": 7404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638413296302807e-06,
+      "loss": 0.8996,
+      "step": 7405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635855922809598e-06,
+      "loss": 1.0144,
+      "step": 7406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6332987973912075e-06,
+      "loss": 0.9501,
+      "step": 7407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630741920236206e-06,
+      "loss": 0.8715,
+      "step": 7408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6281852915331365e-06,
+      "loss": 0.8758,
+      "step": 7409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625628911470528e-06,
+      "loss": 0.885,
+      "step": 7410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623072780236888e-06,
+      "loss": 0.8369,
+      "step": 7411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620516898020715e-06,
+      "loss": 0.7724,
+      "step": 7412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617961265010469e-06,
+      "loss": 0.7811,
+      "step": 7413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615405881394614e-06,
+      "loss": 0.8794,
+      "step": 7414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612850747361586e-06,
+      "loss": 0.9579,
+      "step": 7415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610295863099796e-06,
+      "loss": 0.2812,
+      "step": 7416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607741228797651e-06,
+      "loss": 0.9276,
+      "step": 7417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60518684464353e-06,
+      "loss": 0.915,
+      "step": 7418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602632710825794e-06,
+      "loss": 0.8268,
+      "step": 7419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600078827532783e-06,
+      "loss": 0.8919,
+      "step": 7420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597525194952832e-06,
+      "loss": 1.0049,
+      "step": 7421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59497181327424e-06,
+      "loss": 0.2327,
+      "step": 7422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592418682685302e-06,
+      "loss": 0.9218,
+      "step": 7423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589865803374283e-06,
+      "loss": 0.2792,
+      "step": 7424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5873131755294325e-06,
+      "loss": 0.8885,
+      "step": 7425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584760799338989e-06,
+      "loss": 0.7713,
+      "step": 7426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582208674991165e-06,
+      "loss": 0.8043,
+      "step": 7427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579656802674156e-06,
+      "loss": 0.9482,
+      "step": 7428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577105182576137e-06,
+      "loss": 0.8619,
+      "step": 7429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57455381488527e-06,
+      "loss": 0.9014,
+      "step": 7430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572002699789694e-06,
+      "loss": 0.9924,
+      "step": 7431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569451837477529e-06,
+      "loss": 0.7758,
+      "step": 7432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5669012281368835e-06,
+      "loss": 0.8764,
+      "step": 7433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564350871955834e-06,
+      "loss": 0.8949,
+      "step": 7434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561800769122448e-06,
+      "loss": 0.8635,
+      "step": 7435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559250919824773e-06,
+      "loss": 0.9208,
+      "step": 7436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556701324250838e-06,
+      "loss": 0.8651,
+      "step": 7437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55415198258865e-06,
+      "loss": 0.9093,
+      "step": 7438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551602895026202e-06,
+      "loss": 0.9357,
+      "step": 7439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549054061751465e-06,
+      "loss": 0.9807,
+      "step": 7440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546505482952392e-06,
+      "loss": 0.8579,
+      "step": 7441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5439571588169186e-06,
+      "loss": 0.9259,
+      "step": 7442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54140908953296e-06,
+      "loss": 0.8528,
+      "step": 7443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538861275288411e-06,
+      "loss": 0.868,
+      "step": 7444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53631371627115e-06,
+      "loss": 0.874,
+      "step": 7445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533766412669038e-06,
+      "loss": 0.8819,
+      "step": 7446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531219364669912e-06,
+      "loss": 0.8098,
+      "step": 7447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528672572461596e-06,
+      "loss": 0.8904,
+      "step": 7448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526126036231895e-06,
+      "loss": 0.9159,
+      "step": 7449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523579756168585e-06,
+      "loss": 1.0213,
+      "step": 7450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521033732459441e-06,
+      "loss": 0.9029,
+      "step": 7451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518487965292202e-06,
+      "loss": 0.2626,
+      "step": 7452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515942454854596e-06,
+      "loss": 0.8245,
+      "step": 7453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.513397201334335e-06,
+      "loss": 0.7755,
+      "step": 7454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510852204919102e-06,
+      "loss": 0.8687,
+      "step": 7455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50830746579657e-06,
+      "loss": 0.908,
+      "step": 7456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505762984154391e-06,
+      "loss": 0.8888,
+      "step": 7457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503218760180199e-06,
+      "loss": 0.8768,
+      "step": 7458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5006747940616014e-06,
+      "loss": 0.9224,
+      "step": 7459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498131085986197e-06,
+      "loss": 0.9102,
+      "step": 7460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49558763614156e-06,
+      "loss": 0.8444,
+      "step": 7461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493044444715244e-06,
+      "loss": 0.7948,
+      "step": 7462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490501511894791e-06,
+      "loss": 0.8793,
+      "step": 7463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487958837867721e-06,
+      "loss": 0.8337,
+      "step": 7464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485416422821522e-06,
+      "loss": 0.9733,
+      "step": 7465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482874266943684e-06,
+      "loss": 0.9157,
+      "step": 7466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480332370421662e-06,
+      "loss": 0.9155,
+      "step": 7467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477790733442897e-06,
+      "loss": 0.8964,
+      "step": 7468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475249356194817e-06,
+      "loss": 0.9067,
+      "step": 7469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472708238864822e-06,
+      "loss": 0.919,
+      "step": 7470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4701673816402956e-06,
+      "loss": 0.8086,
+      "step": 7471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467626784708605e-06,
+      "loss": 0.8354,
+      "step": 7472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465086448257094e-06,
+      "loss": 0.8975,
+      "step": 7473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462546372473092e-06,
+      "loss": 0.257,
+      "step": 7474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460006557543902e-06,
+      "loss": 0.8753,
+      "step": 7475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457467003656814e-06,
+      "loss": 0.9171,
+      "step": 7476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454927710999095e-06,
+      "loss": 0.8274,
+      "step": 7477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452388679757998e-06,
+      "loss": 0.765,
+      "step": 7478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4498499101207535e-06,
+      "loss": 0.8375,
+      "step": 7479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447311402274567e-06,
+      "loss": 0.8795,
+      "step": 7480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444773156406637e-06,
+      "loss": 0.8138,
+      "step": 7481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4422351727041334e-06,
+      "loss": 0.8795,
+      "step": 7482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439697451354208e-06,
+      "loss": 0.9378,
+      "step": 7483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437159992543994e-06,
+      "loss": 0.796,
+      "step": 7484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434622796460613e-06,
+      "loss": 0.9274,
+      "step": 7485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432085863291149e-06,
+      "loss": 0.8365,
+      "step": 7486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4295491932226825e-06,
+      "loss": 0.2505,
+      "step": 7487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427012786442271e-06,
+      "loss": 0.7674,
+      "step": 7488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424476643136948e-06,
+      "loss": 0.9265,
+      "step": 7489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421940763493735e-06,
+      "loss": 0.8364,
+      "step": 7490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419405147699629e-06,
+      "loss": 0.892,
+      "step": 7491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416869795941605e-06,
+      "loss": 0.8343,
+      "step": 7492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414334708406624e-06,
+      "loss": 0.9626,
+      "step": 7493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411799885281628e-06,
+      "loss": 0.8115,
+      "step": 7494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409265326753535e-06,
+      "loss": 0.9333,
+      "step": 7495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406731033009246e-06,
+      "loss": 0.9422,
+      "step": 7496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40419700423564e-06,
+      "loss": 0.9445,
+      "step": 7497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401663240619578e-06,
+      "loss": 0.9158,
+      "step": 7498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399129742347905e-06,
+      "loss": 0.8336,
+      "step": 7499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396596509607441e-06,
+      "loss": 0.862,
+      "step": 7500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394063542584989e-06,
+      "loss": 0.8394,
+      "step": 7501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391530841467333e-06,
+      "loss": 0.7425,
+      "step": 7502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388998406441238e-06,
+      "loss": 0.8347,
+      "step": 7503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386466237693445e-06,
+      "loss": 0.8834,
+      "step": 7504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383934335410682e-06,
+      "loss": 0.8589,
+      "step": 7505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381402699779648e-06,
+      "loss": 0.8571,
+      "step": 7506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37887133098703e-06,
+      "loss": 0.8705,
+      "step": 7507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3763402292194954e-06,
+      "loss": 0.861,
+      "step": 7508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373809394663689e-06,
+      "loss": 0.2283,
+      "step": 7509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371278827506237e-06,
+      "loss": 0.9771,
+      "step": 7510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368748527933741e-06,
+      "loss": 0.8281,
+      "step": 7511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366218496132795e-06,
+      "loss": 0.8533,
+      "step": 7512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363688732289961e-06,
+      "loss": 0.9706,
+      "step": 7513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361159236591784e-06,
+      "loss": 0.7048,
+      "step": 7514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358630009224797e-06,
+      "loss": 0.9067,
+      "step": 7515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356101050375506e-06,
+      "loss": 0.9307,
+      "step": 7516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.353572360230396e-06,
+      "loss": 0.869,
+      "step": 7517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351043938975933e-06,
+      "loss": 0.2163,
+      "step": 7518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348515786798569e-06,
+      "loss": 0.8068,
+      "step": 7519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345987903884729e-06,
+      "loss": 0.8863,
+      "step": 7520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343460290420824e-06,
+      "loss": 0.8404,
+      "step": 7521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340932946593242e-06,
+      "loss": 0.8527,
+      "step": 7522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338405872588349e-06,
+      "loss": 0.8802,
+      "step": 7523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335879068592497e-06,
+      "loss": 0.8423,
+      "step": 7524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333352534792011e-06,
+      "loss": 0.9178,
+      "step": 7525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330826271373205e-06,
+      "loss": 0.8331,
+      "step": 7526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328300278522362e-06,
+      "loss": 0.9449,
+      "step": 7527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3257745564257544e-06,
+      "loss": 0.8817,
+      "step": 7528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323249105269626e-06,
+      "loss": 0.8909,
+      "step": 7529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320723925240213e-06,
+      "loss": 0.8141,
+      "step": 7530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31819901652372e-06,
+      "loss": 0.861,
+      "step": 7531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315674379306336e-06,
+      "loss": 0.2327,
+      "step": 7532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313150013774232e-06,
+      "loss": 0.8639,
+      "step": 7533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310625920113556e-06,
+      "loss": 0.8507,
+      "step": 7534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308102098510436e-06,
+      "loss": 0.8889,
+      "step": 7535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305578549150984e-06,
+      "loss": 1.0366,
+      "step": 7536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303055272221281e-06,
+      "loss": 0.8214,
+      "step": 7537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300532267907401e-06,
+      "loss": 0.8609,
+      "step": 7538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298009536395393e-06,
+      "loss": 0.9316,
+      "step": 7539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295487077871283e-06,
+      "loss": 0.8923,
+      "step": 7540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292964892521079e-06,
+      "loss": 0.8726,
+      "step": 7541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290442980530773e-06,
+      "loss": 0.9124,
+      "step": 7542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2879213420863305e-06,
+      "loss": 0.8432,
+      "step": 7543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2853999773736985e-06,
+      "loss": 0.9749,
+      "step": 7544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2828788865788025e-06,
+      "loss": 0.9423,
+      "step": 7545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280358069887555e-06,
+      "loss": 0.8278,
+      "step": 7546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277837527485844e-06,
+      "loss": 0.9186,
+      "step": 7547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275317259559529e-06,
+      "loss": 0.8111,
+      "step": 7548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272797266294461e-06,
+      "loss": 0.8704,
+      "step": 7549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270277547876465e-06,
+      "loss": 0.898,
+      "step": 7550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26775810449135e-06,
+      "loss": 0.841,
+      "step": 7551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2652389363249e-06,
+      "loss": 0.8055,
+      "step": 7552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262720043562882e-06,
+      "loss": 0.8152,
+      "step": 7553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2602014263910375e-06,
+      "loss": 0.8706,
+      "step": 7554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257683084995095e-06,
+      "loss": 0.7953,
+      "step": 7555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255165019560758e-06,
+      "loss": 0.8907,
+      "step": 7556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252647230273714e-06,
+      "loss": 0.8148,
+      "step": 7557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250129717319622e-06,
+      "loss": 0.8212,
+      "step": 7558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247612480884124e-06,
+      "loss": 0.9045,
+      "step": 7559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2450955211528495e-06,
+      "loss": 0.931,
+      "step": 7560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242578838311398e-06,
+      "loss": 0.894,
+      "step": 7561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240062432545353e-06,
+      "loss": 0.8491,
+      "step": 7562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237546304040272e-06,
+      "loss": 0.8439,
+      "step": 7563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235030452981702e-06,
+      "loss": 0.7727,
+      "step": 7564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232514879555163e-06,
+      "loss": 0.8416,
+      "step": 7565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229999583946151e-06,
+      "loss": 0.8705,
+      "step": 7566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227484566340156e-06,
+      "loss": 0.9118,
+      "step": 7567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224969826922624e-06,
+      "loss": 0.8972,
+      "step": 7568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222455365879003e-06,
+      "loss": 0.9663,
+      "step": 7569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219941183394711e-06,
+      "loss": 0.837,
+      "step": 7570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217427279655144e-06,
+      "loss": 0.8563,
+      "step": 7571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214913654845675e-06,
+      "loss": 0.788,
+      "step": 7572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212400309151671e-06,
+      "loss": 0.8712,
+      "step": 7573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209887242758461e-06,
+      "loss": 0.2807,
+      "step": 7574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207374455851361e-06,
+      "loss": 0.887,
+      "step": 7575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204861948615668e-06,
+      "loss": 0.8395,
+      "step": 7576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202349721236657e-06,
+      "loss": 0.8331,
+      "step": 7577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199837773899582e-06,
+      "loss": 0.8896,
+      "step": 7578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197326106789674e-06,
+      "loss": 0.8609,
+      "step": 7579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194814720092144e-06,
+      "loss": 0.8659,
+      "step": 7580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192303613992186e-06,
+      "loss": 0.2341,
+      "step": 7581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189792788674971e-06,
+      "loss": 0.7781,
+      "step": 7582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18728224432565e-06,
+      "loss": 0.8973,
+      "step": 7583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184771981129352e-06,
+      "loss": 0.7591,
+      "step": 7584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182261999271185e-06,
+      "loss": 0.8179,
+      "step": 7585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179752298936239e-06,
+      "loss": 0.8989,
+      "step": 7586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177242880309582e-06,
+      "loss": 0.9379,
+      "step": 7587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17473374357626e-06,
+      "loss": 0.8528,
+      "step": 7588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172224888921295e-06,
+      "loss": 0.7987,
+      "step": 7589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1697163165296945e-06,
+      "loss": 0.8756,
+      "step": 7590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167208026586445e-06,
+      "loss": 0.903,
+      "step": 7591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164700019276509e-06,
+      "loss": 0.81,
+      "step": 7592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162192294784825e-06,
+      "loss": 0.972,
+      "step": 7593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1596848532963214e-06,
+      "loss": 0.8206,
+      "step": 7594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157177694995895e-06,
+      "loss": 0.8738,
+      "step": 7595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154670820068427e-06,
+      "loss": 0.2529,
+      "step": 7596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152164228698773e-06,
+      "loss": 0.7737,
+      "step": 7597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14965792107178e-06,
+      "loss": 0.8246,
+      "step": 7598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1471518973722546e-06,
+      "loss": 0.8044,
+      "step": 7599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144646157784998e-06,
+      "loss": 0.8369,
+      "step": 7600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142140702494786e-06,
+      "loss": 0.8455,
+      "step": 7601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139635531686372e-06,
+      "loss": 0.8808,
+      "step": 7602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13713064554449e-06,
+      "loss": 0.7889,
+      "step": 7603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134626044253853e-06,
+      "loss": 0.939,
+      "step": 7604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1321217279991514e-06,
+      "loss": 0.9617,
+      "step": 7605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129617696965053e-06,
+      "loss": 0.876,
+      "step": 7606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127113951336214e-06,
+      "loss": 0.8846,
+      "step": 7607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12461049129726e-06,
+      "loss": 0.8609,
+      "step": 7608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122107317032794e-06,
+      "loss": 0.8792,
+      "step": 7609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119604428727406e-06,
+      "loss": 0.8273,
+      "step": 7610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117101826565657e-06,
+      "loss": 0.9091,
+      "step": 7611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114599510732097e-06,
+      "loss": 0.8907,
+      "step": 7612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112097481411246e-06,
+      "loss": 0.9309,
+      "step": 7613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109595738787606e-06,
+      "loss": 0.8814,
+      "step": 7614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107094283045656e-06,
+      "loss": 0.8395,
+      "step": 7615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104593114369859e-06,
+      "loss": 0.9223,
+      "step": 7616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1020922329446494e-06,
+      "loss": 0.8835,
+      "step": 7617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099591638954446e-06,
+      "loss": 0.2498,
+      "step": 7618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097091332583649e-06,
+      "loss": 0.8349,
+      "step": 7619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094591314016624e-06,
+      "loss": 0.8038,
+      "step": 7620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0920915834377315e-06,
+      "loss": 0.86,
+      "step": 7621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0895921410313015e-06,
+      "loss": 0.8522,
+      "step": 7622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0870929869816454e-06,
+      "loss": 0.82,
+      "step": 7623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0845941214730504e-06,
+      "loss": 0.865,
+      "step": 7624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082095544689789e-06,
+      "loss": 0.8432,
+      "step": 7625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0795972568161065e-06,
+      "loss": 0.8454,
+      "step": 7626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077099258036228e-06,
+      "loss": 0.8384,
+      "step": 7627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074601548534358e-06,
+      "loss": 0.9124,
+      "step": 7628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072104128494685e-06,
+      "loss": 0.811,
+      "step": 7629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069606998101364e-06,
+      "loss": 0.8072,
+      "step": 7630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067110157538539e-06,
+      "loss": 0.9298,
+      "step": 7631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064613606990326e-06,
+      "loss": 0.9393,
+      "step": 7632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062117346640824e-06,
+      "loss": 0.9199,
+      "step": 7633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059621376674111e-06,
+      "loss": 0.8879,
+      "step": 7634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057125697274242e-06,
+      "loss": 0.908,
+      "step": 7635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0546303086252465e-06,
+      "loss": 0.8466,
+      "step": 7636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052135210911143e-06,
+      "loss": 0.8202,
+      "step": 7637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049640404315917e-06,
+      "loss": 0.8213,
+      "step": 7638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047145889023543e-06,
+      "loss": 0.9039,
+      "step": 7639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044651665217963e-06,
+      "loss": 0.7909,
+      "step": 7640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042157733083105e-06,
+      "loss": 0.8674,
+      "step": 7641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039664092802872e-06,
+      "loss": 0.812,
+      "step": 7642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03717074456115e-06,
+      "loss": 0.8545,
+      "step": 7643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034677688541802e-06,
+      "loss": 0.8253,
+      "step": 7644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032184924928661e-06,
+      "loss": 0.8509,
+      "step": 7645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0296924539055524e-06,
+      "loss": 0.9399,
+      "step": 7646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027200275656272e-06,
+      "loss": 0.8745,
+      "step": 7647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024708390364593e-06,
+      "loss": 0.8741,
+      "step": 7648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022216798214269e-06,
+      "loss": 0.8162,
+      "step": 7649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01972549938904e-06,
+      "loss": 0.9012,
+      "step": 7650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0172344940726015e-06,
+      "loss": 0.837,
+      "step": 7651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014743782448653e-06,
+      "loss": 0.9828,
+      "step": 7652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012253364700861e-06,
+      "loss": 0.7695,
+      "step": 7653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009763241012866e-06,
+      "loss": 0.901,
+      "step": 7654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0072734115682976e-06,
+      "loss": 0.8133,
+      "step": 7655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004783876550755e-06,
+      "loss": 0.8776,
+      "step": 7656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002294636143818e-06,
+      "loss": 1.0067,
+      "step": 7657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9998056905310454e-06,
+      "loss": 0.8665,
+      "step": 7658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997317039895977e-06,
+      "loss": 0.8845,
+      "step": 7659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994828684422127e-06,
+      "loss": 0.8327,
+      "step": 7660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992340624292987e-06,
+      "loss": 0.9023,
+      "step": 7661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9898528596920295e-06,
+      "loss": 0.9187,
+      "step": 7662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9873653908027014e-06,
+      "loss": 0.9029,
+      "step": 7663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984878217808436e-06,
+      "loss": 0.8557,
+      "step": 7664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982391340892638e-06,
+      "loss": 0.7205,
+      "step": 7665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979904760238691e-06,
+      "loss": 0.8699,
+      "step": 7666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977418476029955e-06,
+      "loss": 0.9115,
+      "step": 7667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974932488449775e-06,
+      "loss": 0.8054,
+      "step": 7668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9724467976814685e-06,
+      "loss": 0.8125,
+      "step": 7669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969961403908334e-06,
+      "loss": 0.9924,
+      "step": 7670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9674763073136425e-06,
+      "loss": 0.8912,
+      "step": 7671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9649915080806475e-06,
+      "loss": 0.903,
+      "step": 7672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962507006392584e-06,
+      "loss": 0.8239,
+      "step": 7673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960022802432659e-06,
+      "loss": 0.9013,
+      "step": 7674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957538896384059e-06,
+      "loss": 0.9034,
+      "step": 7675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955055288429949e-06,
+      "loss": 0.8752,
+      "step": 7676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952571978753477e-06,
+      "loss": 0.2535,
+      "step": 7677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950088967537758e-06,
+      "loss": 0.8598,
+      "step": 7678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947606254965894e-06,
+      "loss": 0.868,
+      "step": 7679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945123841220966e-06,
+      "loss": 0.3041,
+      "step": 7680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9426417264860246e-06,
+      "loss": 0.9658,
+      "step": 7681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9401599109441046e-06,
+      "loss": 0.2464,
+      "step": 7682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937678394778217e-06,
+      "loss": 0.9557,
+      "step": 7683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.935197178171351e-06,
+      "loss": 0.8525,
+      "step": 7684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93271626130647e-06,
+      "loss": 0.2138,
+      "step": 7685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930235644366524e-06,
+      "loss": 0.8195,
+      "step": 7686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927755327534434e-06,
+      "loss": 0.9224,
+      "step": 7687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9252753109931015e-06,
+      "loss": 0.9269,
+      "step": 7688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922795594925403e-06,
+      "loss": 0.9341,
+      "step": 7689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920316179514198e-06,
+      "loss": 0.9085,
+      "step": 7690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917837064942321e-06,
+      "loss": 0.9031,
+      "step": 7691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915358251392579e-06,
+      "loss": 0.8232,
+      "step": 7692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912879739047765e-06,
+      "loss": 0.8744,
+      "step": 7693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910401528090643e-06,
+      "loss": 0.93,
+      "step": 7694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907923618703965e-06,
+      "loss": 0.9477,
+      "step": 7695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905446011070451e-06,
+      "loss": 0.9095,
+      "step": 7696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902968705372798e-06,
+      "loss": 0.9211,
+      "step": 7697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900491701793691e-06,
+      "loss": 1.0185,
+      "step": 7698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898015000515784e-06,
+      "loss": 0.9565,
+      "step": 7699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89553860172171e-06,
+      "loss": 0.8478,
+      "step": 7700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893062505594083e-06,
+      "loss": 0.8545,
+      "step": 7701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890586712315489e-06,
+      "loss": 0.8258,
+      "step": 7702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888111222068495e-06,
+      "loss": 0.8257,
+      "step": 7703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885636035035649e-06,
+      "loss": 0.9371,
+      "step": 7704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883161151399473e-06,
+      "loss": 0.879,
+      "step": 7705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880686571342462e-06,
+      "loss": 0.2858,
+      "step": 7706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878212295047101e-06,
+      "loss": 0.9393,
+      "step": 7707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875738322695841e-06,
+      "loss": 0.846,
+      "step": 7708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8732646544711155e-06,
+      "loss": 0.8949,
+      "step": 7709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870791290555334e-06,
+      "loss": 0.929,
+      "step": 7710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868318231130886e-06,
+      "loss": 0.9324,
+      "step": 7711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865845476380139e-06,
+      "loss": 0.8729,
+      "step": 7712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863373026485433e-06,
+      "loss": 0.9184,
+      "step": 7713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860900881629086e-06,
+      "loss": 0.895,
+      "step": 7714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858429041993399e-06,
+      "loss": 0.8544,
+      "step": 7715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855957507760648e-06,
+      "loss": 0.8387,
+      "step": 7716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853486279113087e-06,
+      "loss": 0.8407,
+      "step": 7717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851015356232944e-06,
+      "loss": 0.7918,
+      "step": 7718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848544739302426e-06,
+      "loss": 0.7561,
+      "step": 7719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846074428503723e-06,
+      "loss": 0.8939,
+      "step": 7720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843604424018995e-06,
+      "loss": 0.8324,
+      "step": 7721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841134726030384e-06,
+      "loss": 0.9596,
+      "step": 7722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838665334720004e-06,
+      "loss": 0.8622,
+      "step": 7723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836196250269952e-06,
+      "loss": 0.7825,
+      "step": 7724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8337274728622985e-06,
+      "loss": 0.9146,
+      "step": 7725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831259002679095e-06,
+      "loss": 0.823,
+      "step": 7726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828790839902372e-06,
+      "loss": 0.8604,
+      "step": 7727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8263229847141275e-06,
+      "loss": 0.879,
+      "step": 7728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823855437296347e-06,
+      "loss": 0.7299,
+      "step": 7729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82138819783099e-06,
+      "loss": 0.877,
+      "step": 7730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81892126649999e-06,
+      "loss": 0.8795,
+      "step": 7731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816454643485265e-06,
+      "loss": 0.8582,
+      "step": 7732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813988328968704e-06,
+      "loss": 0.7911,
+      "step": 7733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811522323132169e-06,
+      "loss": 0.8352,
+      "step": 7734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809056626157512e-06,
+      "loss": 0.8981,
+      "step": 7735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806591238226558e-06,
+      "loss": 0.8569,
+      "step": 7736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804126159521097e-06,
+      "loss": 0.8603,
+      "step": 7737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801661390222913e-06,
+      "loss": 0.899,
+      "step": 7738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799196930513763e-06,
+      "loss": 0.8353,
+      "step": 7739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796732780575371e-06,
+      "loss": 0.2342,
+      "step": 7740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794268940589451e-06,
+      "loss": 0.8694,
+      "step": 7741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791805410737685e-06,
+      "loss": 0.2506,
+      "step": 7742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789342191201739e-06,
+      "loss": 0.8688,
+      "step": 7743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786879282163249e-06,
+      "loss": 0.9433,
+      "step": 7744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784416683803837e-06,
+      "loss": 0.8147,
+      "step": 7745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78195439630509e-06,
+      "loss": 0.8627,
+      "step": 7746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779492419848585e-06,
+      "loss": 0.8411,
+      "step": 7747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777030754615871e-06,
+      "loss": 1.0084,
+      "step": 7748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774569400788468e-06,
+      "loss": 0.9511,
+      "step": 7749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772108358547884e-06,
+      "loss": 0.9804,
+      "step": 7750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769647628075592e-06,
+      "loss": 0.8757,
+      "step": 7751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7671872095530525e-06,
+      "loss": 0.8934,
+      "step": 7752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764727103161703e-06,
+      "loss": 0.9972,
+      "step": 7753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762267309082946e-06,
+      "loss": 0.9822,
+      "step": 7754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75980782749817e-06,
+      "loss": 0.9161,
+      "step": 7755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7573486585887404e-06,
+      "loss": 0.7763,
+      "step": 7756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754889802536003e-06,
+      "loss": 0.9322,
+      "step": 7757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7524312595212685e-06,
+      "loss": 0.8629,
+      "step": 7758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74997302972584e-06,
+      "loss": 0.8755,
+      "step": 7759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7475151133309795e-06,
+      "loss": 0.9194,
+      "step": 7760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745057510517941e-06,
+      "loss": 0.7258,
+      "step": 7761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742600221467954e-06,
+      "loss": 0.8447,
+      "step": 7762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740143246362218e-06,
+      "loss": 0.9025,
+      "step": 7763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7376865853819094e-06,
+      "loss": 0.8962,
+      "step": 7764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735230238708185e-06,
+      "loss": 0.8013,
+      "step": 7765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732774206522185e-06,
+      "loss": 0.857,
+      "step": 7766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730318489005009e-06,
+      "loss": 0.8628,
+      "step": 7767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727863086337753e-06,
+      "loss": 0.98,
+      "step": 7768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7254079987014736e-06,
+      "loss": 0.8928,
+      "step": 7769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7229532262772115e-06,
+      "loss": 0.8469,
+      "step": 7770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72049876924599e-06,
+      "loss": 0.7332,
+      "step": 7771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718044627788797e-06,
+      "loss": 0.2731,
+      "step": 7772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715590802086603e-06,
+      "loss": 0.8824,
+      "step": 7773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713137292320361e-06,
+      "loss": 0.7962,
+      "step": 7774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71068409867099e-06,
+      "loss": 0.7791,
+      "step": 7775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708231221319387e-06,
+      "loss": 0.8488,
+      "step": 7776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7057786604464395e-06,
+      "loss": 0.9329,
+      "step": 7777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703326416232989e-06,
+      "loss": 0.8772,
+      "step": 7778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700874488859874e-06,
+      "loss": 0.9345,
+      "step": 7779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6984228785079035e-06,
+      "loss": 0.8466,
+      "step": 7780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695971585357854e-06,
+      "loss": 0.8742,
+      "step": 7781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693520609590491e-06,
+      "loss": 0.8239,
+      "step": 7782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691069951386553e-06,
+      "loss": 0.8975,
+      "step": 7783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688619610926751e-06,
+      "loss": 0.9385,
+      "step": 7784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686169588391772e-06,
+      "loss": 0.7225,
+      "step": 7785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68371988396229e-06,
+      "loss": 0.7487,
+      "step": 7786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6812704978189396e-06,
+      "loss": 0.8315,
+      "step": 7787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678821430142345e-06,
+      "loss": 0.9498,
+      "step": 7788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676372681113107e-06,
+      "loss": 0.7899,
+      "step": 7789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673924250911791e-06,
+      "loss": 0.9263,
+      "step": 7790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671476139718948e-06,
+      "loss": 1.0146,
+      "step": 7791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66902834771511e-06,
+      "loss": 0.9681,
+      "step": 7792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666580875080769e-06,
+      "loss": 0.879,
+      "step": 7793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664133721996414e-06,
+      "loss": 0.8691,
+      "step": 7794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661686888642497e-06,
+      "loss": 0.8461,
+      "step": 7795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659240375199442e-06,
+      "loss": 0.9329,
+      "step": 7796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656794181847664e-06,
+      "loss": 0.9115,
+      "step": 7797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654348308767549e-06,
+      "loss": 0.8755,
+      "step": 7798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6519027561394516e-06,
+      "loss": 0.9218,
+      "step": 7799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649457524143711e-06,
+      "loss": 0.7861,
+      "step": 7800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647012612960647e-06,
+      "loss": 0.8382,
+      "step": 7801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64456802277054e-06,
+      "loss": 0.801,
+      "step": 7802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642123753753661e-06,
+      "loss": 0.896,
+      "step": 7803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639679806090256e-06,
+      "loss": 0.8492,
+      "step": 7804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637236179960539e-06,
+      "loss": 0.8701,
+      "step": 7805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634792875544701e-06,
+      "loss": 0.8686,
+      "step": 7806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632349893022924e-06,
+      "loss": 0.8268,
+      "step": 7807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629907232575345e-06,
+      "loss": 0.8505,
+      "step": 7808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6274648943820905e-06,
+      "loss": 0.8603,
+      "step": 7809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6250228786232686e-06,
+      "loss": 0.7625,
+      "step": 7810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6225811854789455e-06,
+      "loss": 0.8307,
+      "step": 7811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6201398151291775e-06,
+      "loss": 0.8433,
+      "step": 7812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6176987677539964e-06,
+      "loss": 0.8901,
+      "step": 7813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615258043533402e-06,
+      "loss": 0.8508,
+      "step": 7814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61281764264738e-06,
+      "loss": 0.8697,
+      "step": 7815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610377565275887e-06,
+      "loss": 0.9527,
+      "step": 7816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607937811598849e-06,
+      "loss": 0.2672,
+      "step": 7817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605498381796182e-06,
+      "loss": 0.8341,
+      "step": 7818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603059276047775e-06,
+      "loss": 0.8631,
+      "step": 7819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600620494533481e-06,
+      "loss": 0.9135,
+      "step": 7820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598182037433142e-06,
+      "loss": 0.8027,
+      "step": 7821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595743904926576e-06,
+      "loss": 0.8344,
+      "step": 7822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593306097193565e-06,
+      "loss": 0.9036,
+      "step": 7823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59086861441388e-06,
+      "loss": 0.8968,
+      "step": 7824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588431456767269e-06,
+      "loss": 0.9004,
+      "step": 7825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585994624433437e-06,
+      "loss": 0.8531,
+      "step": 7826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583558117592083e-06,
+      "loss": 0.909,
+      "step": 7827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581121936422882e-06,
+      "loss": 0.9536,
+      "step": 7828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578686081105474e-06,
+      "loss": 0.7529,
+      "step": 7829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57625055181948e-06,
+      "loss": 0.8458,
+      "step": 7830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573815348744508e-06,
+      "loss": 0.8768,
+      "step": 7831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571380472060121e-06,
+      "loss": 0.8176,
+      "step": 7832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568945921945872e-06,
+      "loss": 0.9425,
+      "step": 7833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566511698581294e-06,
+      "loss": 0.6408,
+      "step": 7834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564077802145877e-06,
+      "loss": 0.8386,
+      "step": 7835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561644232819109e-06,
+      "loss": 0.8002,
+      "step": 7836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559210990780438e-06,
+      "loss": 0.2358,
+      "step": 7837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55677807620929e-06,
+      "loss": 0.8782,
+      "step": 7838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5543454892850745e-06,
+      "loss": 0.2774,
+      "step": 7839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551913230187176e-06,
+      "loss": 0.7805,
+      "step": 7840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549481299094943e-06,
+      "loss": 0.8915,
+      "step": 7841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547049696187714e-06,
+      "loss": 0.8502,
+      "step": 7842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5446184216448e-06,
+      "loss": 0.9256,
+      "step": 7843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542187475645475e-06,
+      "loss": 0.8339,
+      "step": 7844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539756858369012e-06,
+      "loss": 0.8213,
+      "step": 7845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537326569994639e-06,
+      "loss": 0.8824,
+      "step": 7846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5348966107015655e-06,
+      "loss": 0.8916,
+      "step": 7847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53246698066898e-06,
+      "loss": 0.8835,
+      "step": 7848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5300376800760524e-06,
+      "loss": 0.8669,
+      "step": 7849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527608709101913e-06,
+      "loss": 0.9069,
+      "step": 7850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525180067925678e-06,
+      "loss": 0.8644,
+      "step": 7851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522751756726444e-06,
+      "loss": 0.9409,
+      "step": 7852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5203237756832674e-06,
+      "loss": 0.8125,
+      "step": 7853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5178961249751976e-06,
+      "loss": 0.8725,
+      "step": 7854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515468804781245e-06,
+      "loss": 0.8384,
+      "step": 7855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513041815280407e-06,
+      "loss": 0.8931,
+      "step": 7856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510615156651648e-06,
+      "loss": 0.8536,
+      "step": 7857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508188829073916e-06,
+      "loss": 0.9401,
+      "step": 7858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505762832726125e-06,
+      "loss": 0.8442,
+      "step": 7859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5033371677871724e-06,
+      "loss": 0.9061,
+      "step": 7860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500911834435936e-06,
+      "loss": 0.875,
+      "step": 7861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498486832851248e-06,
+      "loss": 0.8497,
+      "step": 7862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496062163211945e-06,
+      "loss": 0.838,
+      "step": 7863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493637825696811e-06,
+      "loss": 0.8784,
+      "step": 7864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491213820484625e-06,
+      "loss": 0.9271,
+      "step": 7865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488790147754139e-06,
+      "loss": 0.9008,
+      "step": 7866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4863668076840715e-06,
+      "loss": 0.9186,
+      "step": 7867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4839438004531175e-06,
+      "loss": 0.9292,
+      "step": 7868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481521126239958e-06,
+      "loss": 0.8624,
+      "step": 7869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479098785223246e-06,
+      "loss": 0.7903,
+      "step": 7870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476676777581598e-06,
+      "loss": 0.9122,
+      "step": 7871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474255103493623e-06,
+      "loss": 0.8015,
+      "step": 7872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471833763137891e-06,
+      "loss": 0.8877,
+      "step": 7873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469412756692956e-06,
+      "loss": 0.8702,
+      "step": 7874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466992084337348e-06,
+      "loss": 0.8979,
+      "step": 7875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464571746249564e-06,
+      "loss": 0.8263,
+      "step": 7876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462151742608088e-06,
+      "loss": 0.9838,
+      "step": 7877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459732073591367e-06,
+      "loss": 0.9213,
+      "step": 7878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457312739377836e-06,
+      "loss": 0.8867,
+      "step": 7879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45489374014589e-06,
+      "loss": 0.8406,
+      "step": 7880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4524750760739186e-06,
+      "loss": 0.7856,
+      "step": 7881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450056747340266e-06,
+      "loss": 0.8893,
+      "step": 7882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447638754123265e-06,
+      "loss": 0.8394,
+      "step": 7883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445221096601228e-06,
+      "loss": 0.9097,
+      "step": 7884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4428037749524235e-06,
+      "loss": 0.7856,
+      "step": 7885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4403867893551125e-06,
+      "loss": 0.8235,
+      "step": 7886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43797013998753e-06,
+      "loss": 0.8609,
+      "step": 7887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4355538270278754e-06,
+      "loss": 0.7487,
+      "step": 7888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433137850654328e-06,
+      "loss": 0.869,
+      "step": 7889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430722211045052e-06,
+      "loss": 0.7734,
+      "step": 7890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428306908378169e-06,
+      "loss": 0.793,
+      "step": 7891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425891942831791e-06,
+      "loss": 0.2722,
+      "step": 7892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423477314584003e-06,
+      "loss": 0.824,
+      "step": 7893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421063023812853e-06,
+      "loss": 0.8864,
+      "step": 7894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41864907069638e-06,
+      "loss": 0.8459,
+      "step": 7895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416235455412591e-06,
+      "loss": 0.863,
+      "step": 7896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413822178139465e-06,
+      "loss": 0.8876,
+      "step": 7897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411409239054955e-06,
+      "loss": 0.6748,
+      "step": 7898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408996638337003e-06,
+      "loss": 0.8362,
+      "step": 7899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406584376163506e-06,
+      "loss": 0.8387,
+      "step": 7900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404172452712353e-06,
+      "loss": 0.8067,
+      "step": 7901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4017608681614034e-06,
+      "loss": 0.9349,
+      "step": 7902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399349622688479e-06,
+      "loss": 0.9204,
+      "step": 7903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396938716471396e-06,
+      "loss": 0.9141,
+      "step": 7904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394528149687939e-06,
+      "loss": 0.8882,
+      "step": 7905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392117922515854e-06,
+      "loss": 0.8328,
+      "step": 7906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3897080351328815e-06,
+      "loss": 0.9333,
+      "step": 7907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387298487716734e-06,
+      "loss": 0.8539,
+      "step": 7908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384889280445079e-06,
+      "loss": 0.8306,
+      "step": 7909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382480413495581e-06,
+      "loss": 0.8365,
+      "step": 7910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380071887045876e-06,
+      "loss": 0.9542,
+      "step": 7911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3776637012735626e-06,
+      "loss": 0.9244,
+      "step": 7912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375255856356225e-06,
+      "loss": 0.8131,
+      "step": 7913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372848352471427e-06,
+      "loss": 1.0137,
+      "step": 7914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370441189796689e-06,
+      "loss": 0.7248,
+      "step": 7915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368034368509523e-06,
+      "loss": 0.8928,
+      "step": 7916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365627888787412e-06,
+      "loss": 0.8536,
+      "step": 7917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36322175080781e-06,
+      "loss": 0.9062,
+      "step": 7918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360815954748144e-06,
+      "loss": 0.8068,
+      "step": 7919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3584105007858244e-06,
+      "loss": 0.8736,
+      "step": 7920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356005389098227e-06,
+      "loss": 0.8978,
+      "step": 7921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353600619862708e-06,
+      "loss": 0.7925,
+      "step": 7922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3511961932566025e-06,
+      "loss": 0.8859,
+      "step": 7923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348792109457206e-06,
+      "loss": 0.7692,
+      "step": 7924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3463883686418014e-06,
+      "loss": 0.9171,
+      "step": 7925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34398497098765e-06,
+      "loss": 0.8181,
+      "step": 7926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341581916671967e-06,
+      "loss": 0.8434,
+      "step": 7927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339179205871969e-06,
+      "loss": 0.8084,
+      "step": 7928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336776838764825e-06,
+      "loss": 0.9925,
+      "step": 7929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334374815527688e-06,
+      "loss": 0.9156,
+      "step": 7930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331973136337687e-06,
+      "loss": 0.9037,
+      "step": 7931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3295718013719265e-06,
+      "loss": 0.2232,
+      "step": 7932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327170810807478e-06,
+      "loss": 0.8604,
+      "step": 7933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324770164821394e-06,
+      "loss": 0.809,
+      "step": 7934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322369863590706e-06,
+      "loss": 0.8606,
+      "step": 7935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319969907292405e-06,
+      "loss": 0.9012,
+      "step": 7936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3175702961034705e-06,
+      "loss": 0.9225,
+      "step": 7937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315171030200855e-06,
+      "loss": 0.8479,
+      "step": 7938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312772109761479e-06,
+      "loss": 0.7706,
+      "step": 7939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310373534962238e-06,
+      "loss": 0.877,
+      "step": 7940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307975305980011e-06,
+      "loss": 0.8742,
+      "step": 7941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305577422991639e-06,
+      "loss": 0.9215,
+      "step": 7942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303179886173946e-06,
+      "loss": 0.8707,
+      "step": 7943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300782695703734e-06,
+      "loss": 0.7999,
+      "step": 7944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298385851757765e-06,
+      "loss": 0.9439,
+      "step": 7945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295989354512789e-06,
+      "loss": 0.8547,
+      "step": 7946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29359320414553e-06,
+      "loss": 0.2945,
+      "step": 7947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291197400832674e-06,
+      "loss": 0.876,
+      "step": 7948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288801944750897e-06,
+      "loss": 0.9102,
+      "step": 7949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286406836076836e-06,
+      "loss": 0.7808,
+      "step": 7950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284012074987109e-06,
+      "loss": 0.7928,
+      "step": 7951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281617661658308e-06,
+      "loss": 0.8022,
+      "step": 7952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279223596267005e-06,
+      "loss": 0.8832,
+      "step": 7953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276829878989731e-06,
+      "loss": 0.8641,
+      "step": 7954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274436510003005e-06,
+      "loss": 0.8971,
+      "step": 7955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272043489483321e-06,
+      "loss": 0.2622,
+      "step": 7956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269650817607132e-06,
+      "loss": 0.7781,
+      "step": 7957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2672584945508865e-06,
+      "loss": 0.8964,
+      "step": 7958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264866520490992e-06,
+      "loss": 0.8627,
+      "step": 7959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262474895603829e-06,
+      "loss": 0.9034,
+      "step": 7960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260083620065764e-06,
+      "loss": 0.8897,
+      "step": 7961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257692694053134e-06,
+      "loss": 0.867,
+      "step": 7962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25530211774224e-06,
+      "loss": 0.8601,
+      "step": 7963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252911891309369e-06,
+      "loss": 0.8112,
+      "step": 7964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250522014930785e-06,
+      "loss": 0.7661,
+      "step": 7965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2481324887827075e-06,
+      "loss": 0.8684,
+      "step": 7966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245743313041354e-06,
+      "loss": 0.8506,
+      "step": 7967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243354487882892e-06,
+      "loss": 0.8429,
+      "step": 7968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240966013483484e-06,
+      "loss": 0.8547,
+      "step": 7969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238577890019258e-06,
+      "loss": 0.826,
+      "step": 7970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2361901176663155e-06,
+      "loss": 1.0094,
+      "step": 7971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233802696600728e-06,
+      "loss": 0.8131,
+      "step": 7972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2314156269985495e-06,
+      "loss": 0.8525,
+      "step": 7973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2290289090358074e-06,
+      "loss": 0.9518,
+      "step": 7974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226642542888495e-06,
+      "loss": 0.8522,
+      "step": 7975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2242565287325894e-06,
+      "loss": 0.9445,
+      "step": 7976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221870866744033e-06,
+      "loss": 0.9012,
+      "step": 7977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219485557098749e-06,
+      "loss": 0.8868,
+      "step": 7978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217100599972636e-06,
+      "loss": 0.9229,
+      "step": 7979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214715995541557e-06,
+      "loss": 0.8322,
+      "step": 7980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2123317439813534e-06,
+      "loss": 0.9135,
+      "step": 7981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209947845467846e-06,
+      "loss": 0.8068,
+      "step": 7982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207564300176826e-06,
+      "loss": 0.7504,
+      "step": 7983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2051811082840544e-06,
+      "loss": 0.8434,
+      "step": 7984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202798269965277e-06,
+      "loss": 1.0327,
+      "step": 7985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200415785396196e-06,
+      "loss": 0.8992,
+      "step": 7986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198033654752501e-06,
+      "loss": 0.8588,
+      "step": 7987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195651878209862e-06,
+      "loss": 0.8768,
+      "step": 7988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1932704559439e-06,
+      "loss": 0.8551,
+      "step": 7989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190889388130231e-06,
+      "loss": 0.8379,
+      "step": 7990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188508674944433e-06,
+      "loss": 0.9573,
+      "step": 7991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186128316562066e-06,
+      "loss": 0.8224,
+      "step": 7992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183748313158654e-06,
+      "loss": 0.7928,
+      "step": 7993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1813686649097065e-06,
+      "loss": 0.8668,
+      "step": 7994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178989371990694e-06,
+      "loss": 0.9218,
+      "step": 7995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176610434577072e-06,
+      "loss": 0.8707,
+      "step": 7996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1742318528442685e-06,
+      "loss": 0.8545,
+      "step": 7997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171853626967673e-06,
+      "loss": 0.9401,
+      "step": 7998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169475757122664e-06,
+      "loss": 0.9089,
+      "step": 7999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1670982434845915e-06,
+      "loss": 0.9653,
+      "step": 8000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16472108622877e-06,
+      "loss": 0.7899,
+      "step": 8001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162344285530489e-06,
+      "loss": 0.8591,
+      "step": 8002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159967841565026e-06,
+      "loss": 0.7462,
+      "step": 8003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157591754507612e-06,
+      "loss": 0.8593,
+      "step": 8004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155216024533467e-06,
+      "loss": 0.9393,
+      "step": 8005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152840651817782e-06,
+      "loss": 0.8456,
+      "step": 8006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1504656365357106e-06,
+      "loss": 0.8413,
+      "step": 8007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148090978862395e-06,
+      "loss": 0.8234,
+      "step": 8008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1457166789729455e-06,
+      "loss": 0.7938,
+      "step": 8009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143342737042438e-06,
+      "loss": 0.886,
+      "step": 8010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140969153245938e-06,
+      "loss": 0.8762,
+      "step": 8011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138595927758471e-06,
+      "loss": 0.9001,
+      "step": 8012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136223060755035e-06,
+      "loss": 0.8954,
+      "step": 8013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133850552410616e-06,
+      "loss": 0.8961,
+      "step": 8014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1314784029001615e-06,
+      "loss": 0.8023,
+      "step": 8015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129106612398593e-06,
+      "loss": 0.863,
+      "step": 8016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126735181080813e-06,
+      "loss": 0.904,
+      "step": 8017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124364109121693e-06,
+      "loss": 0.8562,
+      "step": 8018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121993396696072e-06,
+      "loss": 0.8056,
+      "step": 8019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119623043978773e-06,
+      "loss": 0.851,
+      "step": 8020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117253051144594e-06,
+      "loss": 0.775,
+      "step": 8021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114883418368286e-06,
+      "loss": 0.8009,
+      "step": 8022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112514145824594e-06,
+      "loss": 0.8758,
+      "step": 8023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110145233688235e-06,
+      "loss": 0.9867,
+      "step": 8024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107776682133888e-06,
+      "loss": 0.8621,
+      "step": 8025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105408491336214e-06,
+      "loss": 0.8913,
+      "step": 8026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103040661469849e-06,
+      "loss": 0.7923,
+      "step": 8027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100673192709393e-06,
+      "loss": 0.817,
+      "step": 8028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098306085229427e-06,
+      "loss": 0.9084,
+      "step": 8029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09593933920451e-06,
+      "loss": 0.8878,
+      "step": 8030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093572954809158e-06,
+      "loss": 0.7979,
+      "step": 8031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091206932217877e-06,
+      "loss": 0.2769,
+      "step": 8032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088841271605138e-06,
+      "loss": 0.8619,
+      "step": 8033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086475973145382e-06,
+      "loss": 0.9323,
+      "step": 8034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084111037013031e-06,
+      "loss": 0.8888,
+      "step": 8035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081746463382482e-06,
+      "loss": 0.8162,
+      "step": 8036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0793822524280934e-06,
+      "loss": 0.8728,
+      "step": 8037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077018404324209e-06,
+      "loss": 0.9089,
+      "step": 8038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074654919245142e-06,
+      "loss": 0.8989,
+      "step": 8039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072291797365172e-06,
+      "loss": 0.9485,
+      "step": 8040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069929038858561e-06,
+      "loss": 0.8535,
+      "step": 8041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067566643899549e-06,
+      "loss": 0.7717,
+      "step": 8042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0652046126623235e-06,
+      "loss": 0.8035,
+      "step": 8043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062842945321074e-06,
+      "loss": 0.861,
+      "step": 8044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060481642049951e-06,
+      "loss": 0.8424,
+      "step": 8045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058120703023076e-06,
+      "loss": 0.7834,
+      "step": 8046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055760128414548e-06,
+      "loss": 0.8493,
+      "step": 8047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0533999183984414e-06,
+      "loss": 0.8414,
+      "step": 8048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0510400731487945e-06,
+      "loss": 0.827,
+      "step": 8049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048680592839625e-06,
+      "loss": 0.8561,
+      "step": 8050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0463214776449285e-06,
+      "loss": 0.786,
+      "step": 8051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043962727738665e-06,
+      "loss": 0.9141,
+      "step": 8052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041604343294766e-06,
+      "loss": 0.8328,
+      "step": 8053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039246324487148e-06,
+      "loss": 0.8973,
+      "step": 8054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0368886714896885e-06,
+      "loss": 0.87,
+      "step": 8055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034531384476242e-06,
+      "loss": 0.8862,
+      "step": 8056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032174463620642e-06,
+      "loss": 0.8638,
+      "step": 8057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029817909096686e-06,
+      "loss": 0.9196,
+      "step": 8058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027461721078148e-06,
+      "loss": 0.9371,
+      "step": 8059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02510589973878e-06,
+      "loss": 0.8075,
+      "step": 8060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022750445252296e-06,
+      "loss": 0.8926,
+      "step": 8061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0203953577923945e-06,
+      "loss": 0.8004,
+      "step": 8062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0180406375327405e-06,
+      "loss": 0.9102,
+      "step": 8063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015686284646968e-06,
+      "loss": 0.9449,
+      "step": 8064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013332299308692e-06,
+      "loss": 0.8026,
+      "step": 8065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010978681691502e-06,
+      "loss": 0.8392,
+      "step": 8066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0086254319689475e-06,
+      "loss": 0.9014,
+      "step": 8067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006272550314564e-06,
+      "loss": 0.8899,
+      "step": 8068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00392003690186e-06,
+      "loss": 0.9445,
+      "step": 8069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0015678919043e-06,
+      "loss": 0.9218,
+      "step": 8070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999216115495344e-06,
+      "loss": 0.8616,
+      "step": 8071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9968647078484055e-06,
+      "loss": 0.8932,
+      "step": 8072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994513669136887e-06,
+      "loss": 0.7715,
+      "step": 8073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992162999534148e-06,
+      "loss": 0.8902,
+      "step": 8074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9898126992135375e-06,
+      "loss": 0.8871,
+      "step": 8075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98746276834836e-06,
+      "loss": 0.9438,
+      "step": 8076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985113207111906e-06,
+      "loss": 0.8497,
+      "step": 8077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982764015677438e-06,
+      "loss": 0.983,
+      "step": 8078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980415194218179e-06,
+      "loss": 0.9048,
+      "step": 8079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9780667429073405e-06,
+      "loss": 0.8192,
+      "step": 8080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975718661918092e-06,
+      "loss": 0.8097,
+      "step": 8081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9733709514235875e-06,
+      "loss": 0.8966,
+      "step": 8082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971023611596952e-06,
+      "loss": 0.8442,
+      "step": 8083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9686766426112776e-06,
+      "loss": 0.8189,
+      "step": 8084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966330044639627e-06,
+      "loss": 0.9234,
+      "step": 8085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963983817855044e-06,
+      "loss": 0.8007,
+      "step": 8086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9616379624305455e-06,
+      "loss": 0.89,
+      "step": 8087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9592924785391095e-06,
+      "loss": 0.98,
+      "step": 8088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9569473663537015e-06,
+      "loss": 0.9284,
+      "step": 8089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954602626047243e-06,
+      "loss": 0.9474,
+      "step": 8090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952258257792644e-06,
+      "loss": 0.8499,
+      "step": 8091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949914261762782e-06,
+      "loss": 0.8803,
+      "step": 8092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9475706381304976e-06,
+      "loss": 0.8819,
+      "step": 8093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945227387068618e-06,
+      "loss": 0.8986,
+      "step": 8094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942884508749932e-06,
+      "loss": 0.288,
+      "step": 8095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940542003347212e-06,
+      "loss": 0.8897,
+      "step": 8096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938199871033187e-06,
+      "loss": 0.8815,
+      "step": 8097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935858111980576e-06,
+      "loss": 0.8145,
+      "step": 8098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933516726362057e-06,
+      "loss": 0.937,
+      "step": 8099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9311757143502855e-06,
+      "loss": 0.7711,
+      "step": 8100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928835076117897e-06,
+      "loss": 0.8506,
+      "step": 8101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926494811837484e-06,
+      "loss": 0.8316,
+      "step": 8102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924154921681621e-06,
+      "loss": 0.8976,
+      "step": 8103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921815405822859e-06,
+      "loss": 0.8066,
+      "step": 8104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9194762644337134e-06,
+      "loss": 0.9016,
+      "step": 8105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917137497686668e-06,
+      "loss": 0.9849,
+      "step": 8106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914799105754195e-06,
+      "loss": 0.8468,
+      "step": 8107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912461088808721e-06,
+      "loss": 0.8897,
+      "step": 8108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910123447022658e-06,
+      "loss": 0.7628,
+      "step": 8109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907786180568387e-06,
+      "loss": 0.8313,
+      "step": 8110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9054492896182545e-06,
+      "loss": 0.8036,
+      "step": 8111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90311277434459e-06,
+      "loss": 0.9044,
+      "step": 8112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900776634919691e-06,
+      "loss": 0.8286,
+      "step": 8113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898440871515823e-06,
+      "loss": 0.9267,
+      "step": 8114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8961054843052245e-06,
+      "loss": 0.9827,
+      "step": 8115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8937704734601155e-06,
+      "loss": 0.852,
+      "step": 8116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891435839152676e-06,
+      "loss": 0.8874,
+      "step": 8117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889101581555066e-06,
+      "loss": 0.9103,
+      "step": 8118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886767700839419e-06,
+      "loss": 0.8995,
+      "step": 8119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884434197177831e-06,
+      "loss": 0.933,
+      "step": 8120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88210107074238e-06,
+      "loss": 0.7916,
+      "step": 8121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879768321705116e-06,
+      "loss": 0.8631,
+      "step": 8122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87743595023805e-06,
+      "loss": 0.9143,
+      "step": 8123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875103956513183e-06,
+      "loss": 0.9586,
+      "step": 8124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872772340702472e-06,
+      "loss": 0.9079,
+      "step": 8125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870441102977849e-06,
+      "loss": 0.8265,
+      "step": 8126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868110243511225e-06,
+      "loss": 0.8511,
+      "step": 8127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865779762474484e-06,
+      "loss": 0.8491,
+      "step": 8128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86344966003947e-06,
+      "loss": 0.8326,
+      "step": 8129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861119936378011e-06,
+      "loss": 0.8857,
+      "step": 8130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858790591661905e-06,
+      "loss": 0.9534,
+      "step": 8131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856461626062913e-06,
+      "loss": 0.8046,
+      "step": 8132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85413303975278e-06,
+      "loss": 0.8818,
+      "step": 8133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851804832903219e-06,
+      "loss": 0.8271,
+      "step": 8134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849477005685912e-06,
+      "loss": 0.9161,
+      "step": 8135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847149558272512e-06,
+      "loss": 0.8097,
+      "step": 8136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844822490834652e-06,
+      "loss": 0.8584,
+      "step": 8137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842495803543926e-06,
+      "loss": 0.8593,
+      "step": 8138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84016949657191e-06,
+      "loss": 0.8578,
+      "step": 8139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8378435700901505e-06,
+      "loss": 0.834,
+      "step": 8140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835518024270157e-06,
+      "loss": 0.8008,
+      "step": 8141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83319285928342e-06,
+      "loss": 0.8512,
+      "step": 8142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830868075301402e-06,
+      "loss": 0.8905,
+      "step": 8143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828543672495529e-06,
+      "loss": 0.8025,
+      "step": 8144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826219651037211e-06,
+      "loss": 0.8295,
+      "step": 8145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823896011097819e-06,
+      "loss": 0.7693,
+      "step": 8146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821572752848697e-06,
+      "loss": 0.8066,
+      "step": 8147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819249876461168e-06,
+      "loss": 0.8257,
+      "step": 8148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816927382106528e-06,
+      "loss": 0.8745,
+      "step": 8149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814605269956028e-06,
+      "loss": 0.8581,
+      "step": 8150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81228354018091e-06,
+      "loss": 0.9908,
+      "step": 8151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809962192952383e-06,
+      "loss": 0.9172,
+      "step": 8152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807641228441617e-06,
+      "loss": 0.7875,
+      "step": 8153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805320646819766e-06,
+      "loss": 0.8726,
+      "step": 8154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803000448257961e-06,
+      "loss": 0.8155,
+      "step": 8155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800680632927277e-06,
+      "loss": 0.8708,
+      "step": 8156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798361200998789e-06,
+      "loss": 0.8916,
+      "step": 8157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796042152643537e-06,
+      "loss": 0.837,
+      "step": 8158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793723488032521e-06,
+      "loss": 0.8757,
+      "step": 8159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791405207336728e-06,
+      "loss": 0.8064,
+      "step": 8160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7890873107271095e-06,
+      "loss": 0.9428,
+      "step": 8161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7867697983745855e-06,
+      "loss": 0.7762,
+      "step": 8162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7844526704500525e-06,
+      "loss": 0.272,
+      "step": 8163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7821359271243825e-06,
+      "loss": 0.8613,
+      "step": 8164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779819568568407e-06,
+      "loss": 0.8743,
+      "step": 8165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777503594952942e-06,
+      "loss": 0.767,
+      "step": 8166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775188006448768e-06,
+      "loss": 0.8745,
+      "step": 8167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772872803226632e-06,
+      "loss": 0.7829,
+      "step": 8168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770557985457265e-06,
+      "loss": 0.925,
+      "step": 8169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768243553311366e-06,
+      "loss": 0.257,
+      "step": 8170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7659295069595965e-06,
+      "loss": 0.934,
+      "step": 8171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7636158465726e-06,
+      "loss": 0.8814,
+      "step": 8172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761302572320991e-06,
+      "loss": 0.9088,
+      "step": 8173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758989684375344e-06,
+      "loss": 0.2652,
+      "step": 8174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756677182906222e-06,
+      "loss": 0.8705,
+      "step": 8175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754365068084148e-06,
+      "loss": 0.8921,
+      "step": 8176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752053340079613e-06,
+      "loss": 0.9072,
+      "step": 8177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749741999063092e-06,
+      "loss": 0.9075,
+      "step": 8178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747431045205027e-06,
+      "loss": 0.9324,
+      "step": 8179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745120478675824e-06,
+      "loss": 0.9016,
+      "step": 8180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742810299645868e-06,
+      "loss": 0.9073,
+      "step": 8181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740500508285519e-06,
+      "loss": 0.9842,
+      "step": 8182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738191104765094e-06,
+      "loss": 0.8797,
+      "step": 8183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7358820892549e-06,
+      "loss": 0.9659,
+      "step": 8184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733573461925195e-06,
+      "loss": 0.8409,
+      "step": 8185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73126522294623e-06,
+      "loss": 0.8294,
+      "step": 8186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728957372488207e-06,
+      "loss": 0.8499,
+      "step": 8187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726649910721316e-06,
+      "loss": 0.9035,
+      "step": 8188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7243428378157055e-06,
+      "loss": 0.9272,
+      "step": 8189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7220361539415026e-06,
+      "loss": 0.8972,
+      "step": 8190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719729859268809e-06,
+      "loss": 0.8627,
+      "step": 8191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717423953967687e-06,
+      "loss": 0.9562,
+      "step": 8192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7151184382081805e-06,
+      "loss": 0.2422,
+      "step": 8193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712813312160293e-06,
+      "loss": 0.905,
+      "step": 8194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710508575994013e-06,
+      "loss": 0.8327,
+      "step": 8195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708204229879295e-06,
+      "loss": 0.8738,
+      "step": 8196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7059002739860604e-06,
+      "loss": 0.7446,
+      "step": 8197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7035967084842e-06,
+      "loss": 0.8929,
+      "step": 8198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701293533543586e-06,
+      "loss": 0.8563,
+      "step": 8199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69899074933406e-06,
+      "loss": 0.9017,
+      "step": 8200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696688356025424e-06,
+      "loss": 0.9507,
+      "step": 8201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694386353787463e-06,
+      "loss": 0.876,
+      "step": 8202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692084742789925e-06,
+      "loss": 0.8219,
+      "step": 8203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689783523202535e-06,
+      "loss": 0.7687,
+      "step": 8204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68748269519499e-06,
+      "loss": 0.9159,
+      "step": 8205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6851822589369475e-06,
+      "loss": 0.9278,
+      "step": 8206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6828822145980524e-06,
+      "loss": 0.8645,
+      "step": 8207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680582562347904e-06,
+      "loss": 0.8894,
+      "step": 8208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678283302356087e-06,
+      "loss": 0.8862,
+      "step": 8209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675984434792143e-06,
+      "loss": 0.8404,
+      "step": 8210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673685959825602e-06,
+      "loss": 0.8896,
+      "step": 8211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671387877625947e-06,
+      "loss": 0.839,
+      "step": 8212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669090188362644e-06,
+      "loss": 0.8653,
+      "step": 8213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666792892205129e-06,
+      "loss": 0.9418,
+      "step": 8214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664495989322802e-06,
+      "loss": 0.7508,
+      "step": 8215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662199479885041e-06,
+      "loss": 0.8439,
+      "step": 8216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6599033640611956e-06,
+      "loss": 0.7837,
+      "step": 8217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657607642020581e-06,
+      "loss": 0.9217,
+      "step": 8218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655312313932479e-06,
+      "loss": 0.7857,
+      "step": 8219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65301737996616e-06,
+      "loss": 0.8593,
+      "step": 8220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650722840290846e-06,
+      "loss": 0.8441,
+      "step": 8221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64842869507574e-06,
+      "loss": 0.9225,
+      "step": 8222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6461349444900196e-06,
+      "loss": 0.7854,
+      "step": 8223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64384158870282e-06,
+      "loss": 0.932,
+      "step": 8224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64154862788326e-06,
+      "loss": 0.7982,
+      "step": 8225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6392560622004265e-06,
+      "loss": 0.8014,
+      "step": 8226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63696389182337e-06,
+      "loss": 0.8986,
+      "step": 8227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63467211692112e-06,
+      "loss": 0.817,
+      "step": 8228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632380737662676e-06,
+      "loss": 0.8801,
+      "step": 8229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6300897542169995e-06,
+      "loss": 0.9515,
+      "step": 8230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627799166753033e-06,
+      "loss": 0.8385,
+      "step": 8231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625508975439691e-06,
+      "loss": 0.8036,
+      "step": 8232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623219180445846e-06,
+      "loss": 0.8911,
+      "step": 8233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620929781940354e-06,
+      "loss": 0.8029,
+      "step": 8234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6186407800920405e-06,
+      "loss": 0.9013,
+      "step": 8235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616352175069691e-06,
+      "loss": 1.0002,
+      "step": 8236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614063967042072e-06,
+      "loss": 0.8279,
+      "step": 8237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611776156177926e-06,
+      "loss": 0.9097,
+      "step": 8238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609488742645944e-06,
+      "loss": 0.8236,
+      "step": 8239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607201726614808e-06,
+      "loss": 0.23,
+      "step": 8240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604915108253169e-06,
+      "loss": 0.8369,
+      "step": 8241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602628887729637e-06,
+      "loss": 0.7454,
+      "step": 8242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600343065212802e-06,
+      "loss": 0.8711,
+      "step": 8243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598057640871227e-06,
+      "loss": 0.8197,
+      "step": 8244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595772614873436e-06,
+      "loss": 0.9441,
+      "step": 8245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593487987387929e-06,
+      "loss": 0.7643,
+      "step": 8246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591203758583182e-06,
+      "loss": 0.8282,
+      "step": 8247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588919928627631e-06,
+      "loss": 0.8086,
+      "step": 8248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586636497689685e-06,
+      "loss": 0.8698,
+      "step": 8249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584353465937733e-06,
+      "loss": 0.8162,
+      "step": 8250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58207083354012e-06,
+      "loss": 0.7578,
+      "step": 8251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579788600665174e-06,
+      "loss": 0.8491,
+      "step": 8252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577506767481191e-06,
+      "loss": 0.8008,
+      "step": 8253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575225334156429e-06,
+      "loss": 0.8392,
+      "step": 8254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572944300859127e-06,
+      "loss": 0.8279,
+      "step": 8255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570663667757492e-06,
+      "loss": 0.9317,
+      "step": 8256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568383435019695e-06,
+      "loss": 0.8455,
+      "step": 8257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566103602813883e-06,
+      "loss": 0.8811,
+      "step": 8258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563824171308182e-06,
+      "loss": 0.9098,
+      "step": 8259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561545140670665e-06,
+      "loss": 0.8806,
+      "step": 8260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559266511069396e-06,
+      "loss": 0.8615,
+      "step": 8261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556988282672406e-06,
+      "loss": 0.8599,
+      "step": 8262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554710455647688e-06,
+      "loss": 0.8878,
+      "step": 8263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552433030163214e-06,
+      "loss": 0.9146,
+      "step": 8264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550156006386926e-06,
+      "loss": 0.8983,
+      "step": 8265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547879384486728e-06,
+      "loss": 0.8671,
+      "step": 8266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545603164630501e-06,
+      "loss": 0.7992,
+      "step": 8267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5433273469861014e-06,
+      "loss": 0.8198,
+      "step": 8268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541051931721345e-06,
+      "loss": 0.9127,
+      "step": 8269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538776919004021e-06,
+      "loss": 0.8934,
+      "step": 8270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536502309001896e-06,
+      "loss": 0.8375,
+      "step": 8271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534228101882694e-06,
+      "loss": 0.8091,
+      "step": 8272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531954297814123e-06,
+      "loss": 0.8141,
+      "step": 8273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529680896963858e-06,
+      "loss": 0.8285,
+      "step": 8274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527407899499532e-06,
+      "loss": 0.9939,
+      "step": 8275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525135305588764e-06,
+      "loss": 0.9049,
+      "step": 8276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52286311539914e-06,
+      "loss": 0.8355,
+      "step": 8277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520591329098205e-06,
+      "loss": 0.8728,
+      "step": 8278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518319946853491e-06,
+      "loss": 0.7888,
+      "step": 8279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516048968832488e-06,
+      "loss": 0.8338,
+      "step": 8280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513778395202655e-06,
+      "loss": 0.8981,
+      "step": 8281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511508226131431e-06,
+      "loss": 0.8591,
+      "step": 8282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5092384617862215e-06,
+      "loss": 0.8765,
+      "step": 8283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506969102334396e-06,
+      "loss": 0.8894,
+      "step": 8284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5047001479433025e-06,
+      "loss": 0.7349,
+      "step": 8285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502431598780258e-06,
+      "loss": 0.8695,
+      "step": 8286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500163455012541e-06,
+      "loss": 0.8899,
+      "step": 8287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497895716807409e-06,
+      "loss": 0.928,
+      "step": 8288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495628384332091e-06,
+      "loss": 0.8828,
+      "step": 8289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493361457753779e-06,
+      "loss": 0.7915,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491094937239633e-06,
+      "loss": 0.8845,
+      "step": 8291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488828822956796e-06,
+      "loss": 0.9182,
+      "step": 8292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486563115072368e-06,
+      "loss": 0.8036,
+      "step": 8293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484297813753424e-06,
+      "loss": 0.882,
+      "step": 8294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482032919167016e-06,
+      "loss": 0.8483,
+      "step": 8295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47976843148015e-06,
+      "loss": 0.898,
+      "step": 8296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477504350859817e-06,
+      "loss": 0.9422,
+      "step": 8297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475240677472974e-06,
+      "loss": 0.8971,
+      "step": 8298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47297741148654e-06,
+      "loss": 0.9435,
+      "step": 8299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470714553067415e-06,
+      "loss": 0.938,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468452102382465e-06,
+      "loss": 0.9179,
+      "step": 8301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466190059598516e-06,
+      "loss": 0.8808,
+      "step": 8302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463928424882381e-06,
+      "loss": 0.8401,
+      "step": 8303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461667198400836e-06,
+      "loss": 0.8458,
+      "step": 8304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45940638032062e-06,
+      "loss": 0.9249,
+      "step": 8305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457145970808451e-06,
+      "loss": 0.8194,
+      "step": 8306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454885970031015e-06,
+      "loss": 0.8427,
+      "step": 8307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452626378154961e-06,
+      "loss": 0.847,
+      "step": 8308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450367195346922e-06,
+      "loss": 0.8762,
+      "step": 8309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448108421773486e-06,
+      "loss": 0.9124,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4458500576012135e-06,
+      "loss": 0.9246,
+      "step": 8311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443592102996642e-06,
+      "loss": 0.7955,
+      "step": 8312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441334558126279e-06,
+      "loss": 0.8316,
+      "step": 8313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4390774231565905e-06,
+      "loss": 0.9679,
+      "step": 8314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4368206982540226e-06,
+      "loss": 0.8192,
+      "step": 8315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434564383584993e-06,
+      "loss": 0.921,
+      "step": 8316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432308479315875e-06,
+      "loss": 0.8723,
+      "step": 8317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43005298561303e-06,
+      "loss": 0.8223,
+      "step": 8318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427797902642772e-06,
+      "loss": 0.9544,
+      "step": 8319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4255432305713954e-06,
+      "loss": 0.9536,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423288969565167e-06,
+      "loss": 0.9507,
+      "step": 8321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421035119790314e-06,
+      "loss": 0.9338,
+      "step": 8322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418781681413031e-06,
+      "loss": 0.9239,
+      "step": 8323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416528654599496e-06,
+      "loss": 0.9636,
+      "step": 8324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41427603951585e-06,
+      "loss": 0.8932,
+      "step": 8325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412023836328197e-06,
+      "loss": 0.8329,
+      "step": 8326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409772045202623e-06,
+      "loss": 0.8925,
+      "step": 8327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407520666305168e-06,
+      "loss": 0.2421,
+      "step": 8328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405269699801858e-06,
+      "loss": 0.7987,
+      "step": 8329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403019145858682e-06,
+      "loss": 0.8578,
+      "step": 8330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400769004641594e-06,
+      "loss": 0.8441,
+      "step": 8331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398519276316519e-06,
+      "loss": 0.8222,
+      "step": 8332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396269961049356e-06,
+      "loss": 0.8456,
+      "step": 8333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3940210590059775e-06,
+      "loss": 0.8391,
+      "step": 8334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391772570352209e-06,
+      "loss": 0.9453,
+      "step": 8335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3895244952538665e-06,
+      "loss": 0.9073,
+      "step": 8336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387276833876716e-06,
+      "loss": 0.9493,
+      "step": 8337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3850295863865044e-06,
+      "loss": 0.8596,
+      "step": 8338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382782752948951e-06,
+      "loss": 0.9049,
+      "step": 8339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380536333729731e-06,
+      "loss": 0.873,
+      "step": 8340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378290328894506e-06,
+      "loss": 0.2434,
+      "step": 8341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37604473860889e-06,
+      "loss": 0.8689,
+      "step": 8342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373799563038481e-06,
+      "loss": 0.8212,
+      "step": 8343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371554802348834e-06,
+      "loss": 0.9076,
+      "step": 8344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369310456705487e-06,
+      "loss": 0.7018,
+      "step": 8345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367066526273932e-06,
+      "loss": 0.7989,
+      "step": 8346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364823011219642e-06,
+      "loss": 0.9785,
+      "step": 8347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36257991170806e-06,
+      "loss": 0.8696,
+      "step": 8348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360337227904586e-06,
+      "loss": 0.8009,
+      "step": 8349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358094959974603e-06,
+      "loss": 0.855,
+      "step": 8350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355853108083458e-06,
+      "loss": 0.8991,
+      "step": 8351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353611672396465e-06,
+      "loss": 0.892,
+      "step": 8352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351370653078907e-06,
+      "loss": 0.8068,
+      "step": 8353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349130050296044e-06,
+      "loss": 0.9669,
+      "step": 8354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346889864213094e-06,
+      "loss": 0.9048,
+      "step": 8355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344650094995253e-06,
+      "loss": 0.8772,
+      "step": 8356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342410742807688e-06,
+      "loss": 0.9168,
+      "step": 8357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340171807815523e-06,
+      "loss": 0.9971,
+      "step": 8358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337933290183861e-06,
+      "loss": 0.857,
+      "step": 8359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335695190077779e-06,
+      "loss": 0.9117,
+      "step": 8360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333457507662306e-06,
+      "loss": 0.8507,
+      "step": 8361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3312202431024605e-06,
+      "loss": 0.8173,
+      "step": 8362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328983396563214e-06,
+      "loss": 0.8391,
+      "step": 8363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326746968209513e-06,
+      "loss": 0.8192,
+      "step": 8364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324510958206275e-06,
+      "loss": 0.7484,
+      "step": 8365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322275366718388e-06,
+      "loss": 0.8542,
+      "step": 8366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320040193910702e-06,
+      "loss": 0.8473,
+      "step": 8367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317805439948042e-06,
+      "loss": 0.8622,
+      "step": 8368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315571104995205e-06,
+      "loss": 0.7922,
+      "step": 8369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313337189216946e-06,
+      "loss": 0.9032,
+      "step": 8370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3111036927779984e-06,
+      "loss": 0.8194,
+      "step": 8371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308870615843069e-06,
+      "loss": 0.8792,
+      "step": 8372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3066379585768144e-06,
+      "loss": 0.9281,
+      "step": 8373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3044057211438784e-06,
+      "loss": 0.9191,
+      "step": 8374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302173903708872e-06,
+      "loss": 0.8078,
+      "step": 8375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299942506436363e-06,
+      "loss": 0.9432,
+      "step": 8376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297711529490902e-06,
+      "loss": 0.9934,
+      "step": 8377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295480973037006e-06,
+      "loss": 0.8226,
+      "step": 8378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29325083723915e-06,
+      "loss": 0.8605,
+      "step": 8379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291021122261792e-06,
+      "loss": 0.8782,
+      "step": 8380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288791828269353e-06,
+      "loss": 0.8557,
+      "step": 8381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286562955426222e-06,
+      "loss": 0.827,
+      "step": 8382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284334503896754e-06,
+      "loss": 0.7779,
+      "step": 8383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282106473845285e-06,
+      "loss": 0.8758,
+      "step": 8384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279878865436103e-06,
+      "loss": 0.8668,
+      "step": 8385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277651678833478e-06,
+      "loss": 0.8111,
+      "step": 8386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2754249142016465e-06,
+      "loss": 0.8986,
+      "step": 8387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2731985717048085e-06,
+      "loss": 0.8048,
+      "step": 8388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270972651507137e-06,
+      "loss": 0.7759,
+      "step": 8389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2687471537727775e-06,
+      "loss": 0.9012,
+      "step": 8390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2665220786658335e-06,
+      "loss": 0.886,
+      "step": 8391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264297426350387e-06,
+      "loss": 0.8247,
+      "step": 8392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262073196990494e-06,
+      "loss": 0.8523,
+      "step": 8393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2598493907501545e-06,
+      "loss": 0.2343,
+      "step": 8394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257626007793363e-06,
+      "loss": 0.8436,
+      "step": 8395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255403048284076e-06,
+      "loss": 0.7903,
+      "step": 8396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253180512386209e-06,
+      "loss": 0.8789,
+      "step": 8397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.25095840026366e-06,
+      "loss": 0.272,
+      "step": 8398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248736712080289e-06,
+      "loss": 0.8276,
+      "step": 8399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2465154479999204e-06,
+      "loss": 0.8375,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244294608186356e-06,
+      "loss": 0.8663,
+      "step": 8401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242074192803365e-06,
+      "loss": 0.8356,
+      "step": 8402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2398542020146794e-06,
+      "loss": 0.8966,
+      "step": 8403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237634635983999e-06,
+      "loss": 0.8771,
+      "step": 8404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235415494875005e-06,
+      "loss": 0.9104,
+      "step": 8405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2331967788513295e-06,
+      "loss": 0.858,
+      "step": 8406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230978488076588e-06,
+      "loss": 0.8224,
+      "step": 8407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228760622714363e-06,
+      "loss": 0.2385,
+      "step": 8408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2265431829281925e-06,
+      "loss": 0.9396,
+      "step": 8409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224326168881597e-06,
+      "loss": 0.8892,
+      "step": 8410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222109580738067e-06,
+      "loss": 0.9166,
+      "step": 8411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219893418661043e-06,
+      "loss": 0.8048,
+      "step": 8412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2176776828139584e-06,
+      "loss": 0.905,
+      "step": 8413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2154623733601985e-06,
+      "loss": 0.8332,
+      "step": 8414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2132474904631185e-06,
+      "loss": 0.7976,
+      "step": 8415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211033034286048e-06,
+      "loss": 0.8808,
+      "step": 8416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208819004992288e-06,
+      "loss": 0.8228,
+      "step": 8417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206605402745096e-06,
+      "loss": 0.9481,
+      "step": 8418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204392227707708e-06,
+      "loss": 0.8155,
+      "step": 8419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202179480043328e-06,
+      "loss": 0.8303,
+      "step": 8420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1999671599151204e-06,
+      "loss": 0.917,
+      "step": 8421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1977552674862286e-06,
+      "loss": 0.8354,
+      "step": 8422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195543802919755e-06,
+      "loss": 0.7885,
+      "step": 8423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19333276637878e-06,
+      "loss": 0.8632,
+      "step": 8424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191122158026341e-06,
+      "loss": 0.9345,
+      "step": 8425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188911978025456e-06,
+      "loss": 0.9092,
+      "step": 8426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1867022265391e-06,
+      "loss": 0.7883,
+      "step": 8427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184492903730224e-06,
+      "loss": 0.8139,
+      "step": 8428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18228400976175e-06,
+      "loss": 0.8951,
+      "step": 8429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180075544796557e-06,
+      "loss": 0.9257,
+      "step": 8430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1778675089975055e-06,
+      "loss": 0.8406,
+      "step": 8431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17565990252741e-06,
+      "loss": 0.9255,
+      "step": 8432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1734527255490665e-06,
+      "loss": 0.7965,
+      "step": 8433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171245978225237e-06,
+      "loss": 0.7982,
+      "step": 8434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1690396607186445e-06,
+      "loss": 0.8344,
+      "step": 8435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166833773191982e-06,
+      "loss": 0.8739,
+      "step": 8436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164628315807916e-06,
+      "loss": 0.8672,
+      "step": 8437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1624232887290835e-06,
+      "loss": 0.8651,
+      "step": 8438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160218692118079e-06,
+      "loss": 0.2766,
+      "step": 8439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158014526137476e-06,
+      "loss": 0.8451,
+      "step": 8440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155810790949805e-06,
+      "loss": 0.8782,
+      "step": 8441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153607486717576e-06,
+      "loss": 0.8349,
+      "step": 8442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151404613603265e-06,
+      "loss": 0.8784,
+      "step": 8443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149202171769311e-06,
+      "loss": 0.8922,
+      "step": 8444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147000161378121e-06,
+      "loss": 0.9496,
+      "step": 8445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144798582592073e-06,
+      "loss": 0.2719,
+      "step": 8446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142597435573522e-06,
+      "loss": 0.8476,
+      "step": 8447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1403967204847715e-06,
+      "loss": 0.8904,
+      "step": 8448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138196437488112e-06,
+      "loss": 0.8987,
+      "step": 8449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13599658674579e-06,
+      "loss": 0.9038,
+      "step": 8450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1337971684200225e-06,
+      "loss": 0.8769,
+      "step": 8451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131598182673005e-06,
+      "loss": 1.02,
+      "step": 8452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129399629666883e-06,
+      "loss": 0.884,
+      "step": 8453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127201509563783e-06,
+      "loss": 0.8951,
+      "step": 8454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125003822525801e-06,
+      "loss": 0.8273,
+      "step": 8455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122806568714992e-06,
+      "loss": 1.0004,
+      "step": 8456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12060974829338e-06,
+      "loss": 0.8399,
+      "step": 8457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1184133614229675e-06,
+      "loss": 0.8644,
+      "step": 8458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116217408265711e-06,
+      "loss": 0.8319,
+      "step": 8459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114021888983545e-06,
+      "loss": 0.292,
+      "step": 8460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111826803738372e-06,
+      "loss": 0.8517,
+      "step": 8461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109632152692055e-06,
+      "loss": 0.8543,
+      "step": 8462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10743793600643e-06,
+      "loss": 0.906,
+      "step": 8463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105244153843305e-06,
+      "loss": 0.8154,
+      "step": 8464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103050806364448e-06,
+      "loss": 0.7965,
+      "step": 8465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100857893731595e-06,
+      "loss": 0.7952,
+      "step": 8466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0986654161064595e-06,
+      "loss": 0.9404,
+      "step": 8467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09647337365071e-06,
+      "loss": 0.8125,
+      "step": 8468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094281766525992e-06,
+      "loss": 0.8729,
+      "step": 8469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092090594893923e-06,
+      "loss": 0.8371,
+      "step": 8470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089899858916072e-06,
+      "loss": 0.9131,
+      "step": 8471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08770955875399e-06,
+      "loss": 0.8314,
+      "step": 8472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085519694569194e-06,
+      "loss": 0.9703,
+      "step": 8473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083330266523162e-06,
+      "loss": 0.865,
+      "step": 8474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08114127477735e-06,
+      "loss": 0.8208,
+      "step": 8475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078952719493173e-06,
+      "loss": 0.8297,
+      "step": 8476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076764600832011e-06,
+      "loss": 0.8054,
+      "step": 8477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074576918955224e-06,
+      "loss": 0.8193,
+      "step": 8478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072389674024137e-06,
+      "loss": 0.835,
+      "step": 8479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070202866200029e-06,
+      "loss": 0.857,
+      "step": 8480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068016495644164e-06,
+      "loss": 0.7964,
+      "step": 8481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065830562517771e-06,
+      "loss": 0.8353,
+      "step": 8482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0636450669820315e-06,
+      "loss": 0.8279,
+      "step": 8483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061460009198113e-06,
+      "loss": 0.2427,
+      "step": 8484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059275389327145e-06,
+      "loss": 0.8513,
+      "step": 8485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05709120753022e-06,
+      "loss": 0.9042,
+      "step": 8486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054907463968398e-06,
+      "loss": 0.787,
+      "step": 8487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052724158802719e-06,
+      "loss": 0.8763,
+      "step": 8488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050541292194171e-06,
+      "loss": 0.8437,
+      "step": 8489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048358864303727e-06,
+      "loss": 0.7928,
+      "step": 8490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046176875292323e-06,
+      "loss": 0.9352,
+      "step": 8491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043995325320855e-06,
+      "loss": 0.8072,
+      "step": 8492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041814214550196e-06,
+      "loss": 0.8641,
+      "step": 8493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039633543141184e-06,
+      "loss": 0.8881,
+      "step": 8494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0374533112546186e-06,
+      "loss": 0.8182,
+      "step": 8495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035273519051278e-06,
+      "loss": 0.8869,
+      "step": 8496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033094166691899e-06,
+      "loss": 0.2335,
+      "step": 8497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0309152543371855e-06,
+      "loss": 0.7719,
+      "step": 8498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028736782147815e-06,
+      "loss": 0.8054,
+      "step": 8499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026558750284435e-06,
+      "loss": 0.8599,
+      "step": 8500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024381158907646e-06,
+      "loss": 0.9326,
+      "step": 8501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022204008178032e-06,
+      "loss": 0.8074,
+      "step": 8502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020027298256139e-06,
+      "loss": 0.8583,
+      "step": 8503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017851029302472e-06,
+      "loss": 0.8185,
+      "step": 8504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015675201477517e-06,
+      "loss": 0.8683,
+      "step": 8505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013499814941726e-06,
+      "loss": 0.7935,
+      "step": 8506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011324869855502e-06,
+      "loss": 0.8502,
+      "step": 8507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009150366379232e-06,
+      "loss": 0.9041,
+      "step": 8508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0069763046732704e-06,
+      "loss": 0.824,
+      "step": 8509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004802684897927e-06,
+      "loss": 0.8874,
+      "step": 8510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0026295072134904e-06,
+      "loss": 0.8835,
+      "step": 8511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000456771780215e-06,
+      "loss": 0.9076,
+      "step": 8512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998284478758314e-06,
+      "loss": 0.9409,
+      "step": 8513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996112628307979e-06,
+      "loss": 0.7619,
+      "step": 8514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993941220589365e-06,
+      "loss": 0.7486,
+      "step": 8515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991770255762587e-06,
+      "loss": 0.8376,
+      "step": 8516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989599733987741e-06,
+      "loss": 0.9481,
+      "step": 8517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987429655424882e-06,
+      "loss": 0.8383,
+      "step": 8518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985260020234025e-06,
+      "loss": 0.8064,
+      "step": 8519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983090828575169e-06,
+      "loss": 0.8551,
+      "step": 8520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980922080608274e-06,
+      "loss": 0.8349,
+      "step": 8521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978753776493256e-06,
+      "loss": 0.8728,
+      "step": 8522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976585916390014e-06,
+      "loss": 0.8834,
+      "step": 8523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97441850045841e-06,
+      "loss": 0.9274,
+      "step": 8524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972251528858264e-06,
+      "loss": 0.9325,
+      "step": 8525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970085001749379e-06,
+      "loss": 0.9319,
+      "step": 8526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96791891929151e-06,
+      "loss": 0.9496,
+      "step": 8527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965753281644386e-06,
+      "loss": 0.8526,
+      "step": 8528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9635880889677026e-06,
+      "loss": 0.9258,
+      "step": 8529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961423341421129e-06,
+      "loss": 0.8885,
+      "step": 8530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959259039164287e-06,
+      "loss": 0.8678,
+      "step": 8531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957095182356779e-06,
+      "loss": 0.9012,
+      "step": 8532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954931771158173e-06,
+      "loss": 0.9017,
+      "step": 8533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9527688057279925e-06,
+      "loss": 0.819,
+      "step": 8534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950606286225744e-06,
+      "loss": 0.8127,
+      "step": 8535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948444212810886e-06,
+      "loss": 0.8897,
+      "step": 8536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9462825856428595e-06,
+      "loss": 0.9616,
+      "step": 8537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9441214048810585e-06,
+      "loss": 0.8381,
+      "step": 8538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941960670684855e-06,
+      "loss": 0.8857,
+      "step": 8539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939800383213579e-06,
+      "loss": 0.8912,
+      "step": 8540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937640542626533e-06,
+      "loss": 0.9341,
+      "step": 8541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9354811490829906e-06,
+      "loss": 0.8273,
+      "step": 8542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9333222027421795e-06,
+      "loss": 0.8227,
+      "step": 8543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931163703763309e-06,
+      "loss": 0.883,
+      "step": 8544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929005652305542e-06,
+      "loss": 0.9431,
+      "step": 8545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926848048528018e-06,
+      "loss": 0.8309,
+      "step": 8546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924690892589845e-06,
+      "loss": 0.8295,
+      "step": 8547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922534184650091e-06,
+      "loss": 0.8466,
+      "step": 8548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9203779248677875e-06,
+      "loss": 0.7991,
+      "step": 8549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918222113401943e-06,
+      "loss": 0.814,
+      "step": 8550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916066750411534e-06,
+      "loss": 0.7946,
+      "step": 8551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913911836055489e-06,
+      "loss": 0.2564,
+      "step": 8552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911757370492724e-06,
+      "loss": 0.9716,
+      "step": 8553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9096033538821e-06,
+      "loss": 0.8698,
+      "step": 8554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9074497863824625e-06,
+      "loss": 0.8006,
+      "step": 8555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905296668152618e-06,
+      "loss": 0.8414,
+      "step": 8556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9031439993513366e-06,
+      "loss": 0.8524,
+      "step": 8557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900991780137361e-06,
+      "loss": 0.849,
+      "step": 8558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898840010669392e-06,
+      "loss": 0.8097,
+      "step": 8559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89668869110611e-06,
+      "loss": 0.8505,
+      "step": 8560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894537821606149e-06,
+      "loss": 0.8668,
+      "step": 8561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89238740232812e-06,
+      "loss": 0.88,
+      "step": 8562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890237433430594e-06,
+      "loss": 0.8883,
+      "step": 8563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888087915072112e-06,
+      "loss": 0.8483,
+      "step": 8564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885938847411185e-06,
+      "loss": 0.8788,
+      "step": 8565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88379023060628e-06,
+      "loss": 0.8348,
+      "step": 8566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881642064815843e-06,
+      "loss": 0.908,
+      "step": 8567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879494350198284e-06,
+      "loss": 0.8544,
+      "step": 8568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877347086911973e-06,
+      "loss": 0.8326,
+      "step": 8569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875200275115248e-06,
+      "loss": 0.9041,
+      "step": 8570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873053914966426e-06,
+      "loss": 0.7952,
+      "step": 8571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87090800662377e-06,
+      "loss": 0.8712,
+      "step": 8572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868762550245529e-06,
+      "loss": 0.8744,
+      "step": 8573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8666175459899104e-06,
+      "loss": 0.8401,
+      "step": 8574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864472994015084e-06,
+      "loss": 0.7782,
+      "step": 8575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862328894479195e-06,
+      "loss": 0.8246,
+      "step": 8576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860185247540352e-06,
+      "loss": 0.8956,
+      "step": 8577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858042053356624e-06,
+      "loss": 0.8592,
+      "step": 8578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855899312086059e-06,
+      "loss": 0.8879,
+      "step": 8579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8537570238866604e-06,
+      "loss": 0.8708,
+      "step": 8580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8516151889164e-06,
+      "loss": 0.7628,
+      "step": 8581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849473807333221e-06,
+      "loss": 0.7518,
+      "step": 8582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847332879295033e-06,
+      "loss": 0.795,
+      "step": 8583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845192404959707e-06,
+      "loss": 0.8394,
+      "step": 8584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843052384485082e-06,
+      "loss": 0.8465,
+      "step": 8585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8409128180289715e-06,
+      "loss": 0.8678,
+      "step": 8586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838773705749141e-06,
+      "loss": 0.8588,
+      "step": 8587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836635047803332e-06,
+      "loss": 0.9365,
+      "step": 8588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834496844349262e-06,
+      "loss": 0.9461,
+      "step": 8589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832359095544587e-06,
+      "loss": 0.8511,
+      "step": 8590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830221801546954e-06,
+      "loss": 0.9156,
+      "step": 8591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828084962513973e-06,
+      "loss": 0.7853,
+      "step": 8592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825948578603208e-06,
+      "loss": 0.9431,
+      "step": 8593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8238126499722026e-06,
+      "loss": 0.8974,
+      "step": 8594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821677176778465e-06,
+      "loss": 0.7461,
+      "step": 8595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8195421591794615e-06,
+      "loss": 0.9124,
+      "step": 8596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817407597332629e-06,
+      "loss": 0.7894,
+      "step": 8597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815273491395381e-06,
+      "loss": 0.9697,
+      "step": 8598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81313984152508e-06,
+      "loss": 0.8241,
+      "step": 8599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811006647879062e-06,
+      "loss": 0.8467,
+      "step": 8600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8088739106146377e-06,
+      "loss": 0.8531,
+      "step": 8601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806741629889069e-06,
+      "loss": 0.9957,
+      "step": 8602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804609805859596e-06,
+      "loss": 0.9025,
+      "step": 8603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8024784386834255e-06,
+      "loss": 0.9113,
+      "step": 8604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800347528517717e-06,
+      "loss": 0.8887,
+      "step": 8605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7982170755196125e-06,
+      "loss": 0.8834,
+      "step": 8606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7960870798462136e-06,
+      "loss": 0.7823,
+      "step": 8607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7939575416545837e-06,
+      "loss": 0.9245,
+      "step": 8608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7918284611017575e-06,
+      "loss": 0.8906,
+      "step": 8609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789699838344745e-06,
+      "loss": 0.8606,
+      "step": 8610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7875716735404964e-06,
+      "loss": 0.2236,
+      "step": 8611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.785443966845952e-06,
+      "loss": 0.9299,
+      "step": 8612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.783316718418015e-06,
+      "loss": 0.8522,
+      "step": 8613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7811899284135433e-06,
+      "loss": 0.868,
+      "step": 8614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779063596989371e-06,
+      "loss": 0.7838,
+      "step": 8615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7769377243022987e-06,
+      "loss": 0.9286,
+      "step": 8616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774812310509085e-06,
+      "loss": 0.8231,
+      "step": 8617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7726873557664613e-06,
+      "loss": 0.853,
+      "step": 8618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7705628602311272e-06,
+      "loss": 0.9146,
+      "step": 8619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7684388240597425e-06,
+      "loss": 0.9201,
+      "step": 8620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7663152474089303e-06,
+      "loss": 0.9441,
+      "step": 8621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7641921304352935e-06,
+      "loss": 0.9149,
+      "step": 8622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7620694732953854e-06,
+      "loss": 0.8543,
+      "step": 8623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7599472761457356e-06,
+      "loss": 0.8076,
+      "step": 8624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757825539142839e-06,
+      "loss": 0.9053,
+      "step": 8625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7557042624431493e-06,
+      "loss": 0.9013,
+      "step": 8626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753583446203093e-06,
+      "loss": 0.9085,
+      "step": 8627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751463090579066e-06,
+      "loss": 0.8468,
+      "step": 8628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749343195727416e-06,
+      "loss": 0.9155,
+      "step": 8629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747223761804476e-06,
+      "loss": 0.9004,
+      "step": 8630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745104788966528e-06,
+      "loss": 0.8655,
+      "step": 8631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7429862773698257e-06,
+      "loss": 0.8358,
+      "step": 8632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7408682271705934e-06,
+      "loss": 0.9545,
+      "step": 8633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738750638525018e-06,
+      "loss": 0.8953,
+      "step": 8634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7366335115892504e-06,
+      "loss": 0.8019,
+      "step": 8635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7345168465194094e-06,
+      "loss": 0.8572,
+      "step": 8636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732400643471583e-06,
+      "loss": 0.7297,
+      "step": 8637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7302849026018175e-06,
+      "loss": 0.8302,
+      "step": 8638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7281696240661334e-06,
+      "loss": 0.2604,
+      "step": 8639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7260548080205085e-06,
+      "loss": 0.8479,
+      "step": 8640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7239404546208968e-06,
+      "loss": 0.9594,
+      "step": 8641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721826564023205e-06,
+      "loss": 0.9214,
+      "step": 8642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7197131363833226e-06,
+      "loss": 0.8911,
+      "step": 8643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717600171857085e-06,
+      "loss": 0.7874,
+      "step": 8644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7154876706003107e-06,
+      "loss": 0.815,
+      "step": 8645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71337563276878e-06,
+      "loss": 0.8122,
+      "step": 8646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711264058518228e-06,
+      "loss": 0.9032,
+      "step": 8647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.709152948004372e-06,
+      "loss": 0.8497,
+      "step": 8648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7070423013828795e-06,
+      "loss": 0.7976,
+      "step": 8649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7049321188093956e-06,
+      "loss": 0.7838,
+      "step": 8650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7028224004395308e-06,
+      "loss": 0.8219,
+      "step": 8651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7007131464288524e-06,
+      "loss": 0.8259,
+      "step": 8652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698604356932897e-06,
+      "loss": 0.8895,
+      "step": 8653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6964960321071697e-06,
+      "loss": 0.8303,
+      "step": 8654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694388172107145e-06,
+      "loss": 0.8423,
+      "step": 8655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6922807770882518e-06,
+      "loss": 0.8974,
+      "step": 8656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690173847205898e-06,
+      "loss": 0.8794,
+      "step": 8657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6880673826154435e-06,
+      "loss": 0.9084,
+      "step": 8658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685961383472222e-06,
+      "loss": 0.8674,
+      "step": 8659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683855849931538e-06,
+      "loss": 0.804,
+      "step": 8660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681750782148651e-06,
+      "loss": 0.8989,
+      "step": 8661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679646180278786e-06,
+      "loss": 0.2305,
+      "step": 8662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677542044477144e-06,
+      "loss": 0.8063,
+      "step": 8663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6754383748988854e-06,
+      "loss": 0.8333,
+      "step": 8664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6733351716991328e-06,
+      "loss": 0.8463,
+      "step": 8665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671232435032984e-06,
+      "loss": 0.957,
+      "step": 8666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.66913016505549e-06,
+      "loss": 0.2385,
+      "step": 8667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6670283619216774e-06,
+      "loss": 0.8692,
+      "step": 8668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.664927025786538e-06,
+      "loss": 0.8086,
+      "step": 8669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6628261568050183e-06,
+      "loss": 0.8952,
+      "step": 8670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6607257551320472e-06,
+      "loss": 0.8611,
+      "step": 8671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6586258209225024e-06,
+      "loss": 0.8555,
+      "step": 8672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6565263543312413e-06,
+      "loss": 0.8922,
+      "step": 8673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6544273555130728e-06,
+      "loss": 0.9951,
+      "step": 8674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6523288246227874e-06,
+      "loss": 0.9167,
+      "step": 8675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6502307618151234e-06,
+      "loss": 0.8316,
+      "step": 8676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6481331672447993e-06,
+      "loss": 0.8315,
+      "step": 8677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6460360410664964e-06,
+      "loss": 0.8107,
+      "step": 8678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6439393834348512e-06,
+      "loss": 0.7714,
+      "step": 8679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6418431945044774e-06,
+      "loss": 0.9261,
+      "step": 8680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639747474429951e-06,
+      "loss": 0.9209,
+      "step": 8681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.637652223365812e-06,
+      "loss": 0.7875,
+      "step": 8682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6355574414665594e-06,
+      "loss": 0.8354,
+      "step": 8683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633463128886673e-06,
+      "loss": 0.9442,
+      "step": 8684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6313692857805806e-06,
+      "loss": 0.8057,
+      "step": 8685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6292759123026898e-06,
+      "loss": 0.8216,
+      "step": 8686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6271830086073687e-06,
+      "loss": 0.7678,
+      "step": 8687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6250905748489453e-06,
+      "loss": 0.9226,
+      "step": 8688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6229986111817182e-06,
+      "loss": 0.9341,
+      "step": 8689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6209071177599565e-06,
+      "loss": 0.8788,
+      "step": 8690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61881609473788e-06,
+      "loss": 0.8748,
+      "step": 8691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616725542269691e-06,
+      "loss": 0.8409,
+      "step": 8692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6146354605095436e-06,
+      "loss": 0.8189,
+      "step": 8693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6125458496115594e-06,
+      "loss": 0.869,
+      "step": 8694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6104567097298315e-06,
+      "loss": 0.9169,
+      "step": 8695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6083680410184184e-06,
+      "loss": 0.8412,
+      "step": 8696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.606279843631334e-06,
+      "loss": 0.8452,
+      "step": 8697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604192117722566e-06,
+      "loss": 0.8682,
+      "step": 8698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602104863446069e-06,
+      "loss": 0.9486,
+      "step": 8699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6000180809557515e-06,
+      "loss": 0.9207,
+      "step": 8700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5979317704054994e-06,
+      "loss": 0.9084,
+      "step": 8701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5958459319491646e-06,
+      "loss": 0.8426,
+      "step": 8702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593760565740545e-06,
+      "loss": 0.9552,
+      "step": 8703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5916756719334246e-06,
+      "loss": 0.8548,
+      "step": 8704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589591250681549e-06,
+      "loss": 0.9082,
+      "step": 8705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5875073021386173e-06,
+      "loss": 0.7713,
+      "step": 8706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5854238264583052e-06,
+      "loss": 0.8569,
+      "step": 8707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5833408237942548e-06,
+      "loss": 0.7296,
+      "step": 8708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5812582943000598e-06,
+      "loss": 0.2451,
+      "step": 8709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5791762381292915e-06,
+      "loss": 0.8995,
+      "step": 8710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577094655435487e-06,
+      "loss": 0.8214,
+      "step": 8711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5750135463721358e-06,
+      "loss": 0.7512,
+      "step": 8712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572932911092708e-06,
+      "loss": 0.806,
+      "step": 8713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5708527497506283e-06,
+      "loss": 0.7993,
+      "step": 8714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5687730624992865e-06,
+      "loss": 0.8915,
+      "step": 8715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5666938494920424e-06,
+      "loss": 0.8102,
+      "step": 8716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5646151108822235e-06,
+      "loss": 0.8199,
+      "step": 8717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5625368468231104e-06,
+      "loss": 0.8041,
+      "step": 8718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56045905746796e-06,
+      "loss": 0.8512,
+      "step": 8719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558381742969993e-06,
+      "loss": 0.7258,
+      "step": 8720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556304903482386e-06,
+      "loss": 0.9001,
+      "step": 8721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.554228539158291e-06,
+      "loss": 0.8791,
+      "step": 8722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5521526501508253e-06,
+      "loss": 0.2523,
+      "step": 8723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5500772366130555e-06,
+      "loss": 0.8389,
+      "step": 8724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54800229869803e-06,
+      "loss": 0.9147,
+      "step": 8725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54592783655876e-06,
+      "loss": 0.9088,
+      "step": 8726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543853850348211e-06,
+      "loss": 0.8782,
+      "step": 8727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541780340219325e-06,
+      "loss": 0.8072,
+      "step": 8728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5397073063250076e-06,
+      "loss": 0.7667,
+      "step": 8729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5376347488181174e-06,
+      "loss": 0.8604,
+      "step": 8730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535562667851492e-06,
+      "loss": 0.8911,
+      "step": 8731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5334910635779306e-06,
+      "loss": 0.8087,
+      "step": 8732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5314199361501934e-06,
+      "loss": 0.7875,
+      "step": 8733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529349285721001e-06,
+      "loss": 0.9035,
+      "step": 8734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527279112443054e-06,
+      "loss": 0.838,
+      "step": 8735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525209416469e-06,
+      "loss": 0.9329,
+      "step": 8736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5231401979514656e-06,
+      "loss": 0.8561,
+      "step": 8737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5210714570430383e-06,
+      "loss": 0.93,
+      "step": 8738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519003193896263e-06,
+      "loss": 0.8303,
+      "step": 8739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51693540866366e-06,
+      "loss": 0.8278,
+      "step": 8740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514868101497709e-06,
+      "loss": 0.929,
+      "step": 8741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5128012725508522e-06,
+      "loss": 0.8411,
+      "step": 8742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5107349219755025e-06,
+      "loss": 0.8418,
+      "step": 8743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508669049924034e-06,
+      "loss": 0.8619,
+      "step": 8744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5066036565487814e-06,
+      "loss": 0.8764,
+      "step": 8745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5045387420020517e-06,
+      "loss": 0.8018,
+      "step": 8746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5024743064361166e-06,
+      "loss": 0.7846,
+      "step": 8747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5004103500032027e-06,
+      "loss": 0.8705,
+      "step": 8748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4983468728555116e-06,
+      "loss": 0.8016,
+      "step": 8749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496283875145209e-06,
+      "loss": 0.8707,
+      "step": 8750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4942213570244155e-06,
+      "loss": 0.8652,
+      "step": 8751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4921593186452297e-06,
+      "loss": 0.8481,
+      "step": 8752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490097760159702e-06,
+      "loss": 0.9165,
+      "step": 8753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4880366817198584e-06,
+      "loss": 0.7748,
+      "step": 8754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485976083477679e-06,
+      "loss": 0.892,
+      "step": 8755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4839159655851206e-06,
+      "loss": 0.889,
+      "step": 8756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481856328194091e-06,
+      "loss": 0.8998,
+      "step": 8757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479797171456474e-06,
+      "loss": 0.8183,
+      "step": 8758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4777384955241158e-06,
+      "loss": 0.2979,
+      "step": 8759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475680300548818e-06,
+      "loss": 0.8914,
+      "step": 8760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4736225866823613e-06,
+      "loss": 0.885,
+      "step": 8761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471565354076477e-06,
+      "loss": 0.962,
+      "step": 8762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4695086028828684e-06,
+      "loss": 0.8591,
+      "step": 8763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4674523332532074e-06,
+      "loss": 0.8464,
+      "step": 8764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4653965453391215e-06,
+      "loss": 0.8806,
+      "step": 8765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4633412392922015e-06,
+      "loss": 0.91,
+      "step": 8766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4612864152640114e-06,
+      "loss": 0.8362,
+      "step": 8767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4592320734060803e-06,
+      "loss": 0.8841,
+      "step": 8768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.457178213869888e-06,
+      "loss": 0.8737,
+      "step": 8769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4551248368068968e-06,
+      "loss": 0.8501,
+      "step": 8770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4530719423685166e-06,
+      "loss": 0.8765,
+      "step": 8771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4510195307061332e-06,
+      "loss": 0.9719,
+      "step": 8772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448967601971097e-06,
+      "loss": 0.7909,
+      "step": 8773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4469161563147114e-06,
+      "loss": 0.8891,
+      "step": 8774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444865193888258e-06,
+      "loss": 0.9256,
+      "step": 8775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4428147148429715e-06,
+      "loss": 0.873,
+      "step": 8776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440764719330062e-06,
+      "loss": 0.9081,
+      "step": 8777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43871520750069e-06,
+      "loss": 0.8644,
+      "step": 8778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4366661795059932e-06,
+      "loss": 0.9044,
+      "step": 8779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4346176354970705e-06,
+      "loss": 0.8848,
+      "step": 8780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4325695756249776e-06,
+      "loss": 0.2632,
+      "step": 8781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4305220000407467e-06,
+      "loss": 0.8533,
+      "step": 8782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4284749088953607e-06,
+      "loss": 0.9265,
+      "step": 8783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4264283023397783e-06,
+      "loss": 0.8354,
+      "step": 8784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42438218052492e-06,
+      "loss": 0.8675,
+      "step": 8785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4223365436016652e-06,
+      "loss": 0.938,
+      "step": 8786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4202913917208592e-06,
+      "loss": 0.8532,
+      "step": 8787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418246725033314e-06,
+      "loss": 0.9724,
+      "step": 8788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4162025436898105e-06,
+      "loss": 0.8426,
+      "step": 8789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4141588478410814e-06,
+      "loss": 0.8112,
+      "step": 8790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4121156376378363e-06,
+      "loss": 0.9133,
+      "step": 8791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4100729132307365e-06,
+      "loss": 0.8331,
+      "step": 8792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4080306747704196e-06,
+      "loss": 0.8531,
+      "step": 8793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4059889224074825e-06,
+      "loss": 0.9491,
+      "step": 8794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4039476562924845e-06,
+      "loss": 0.8097,
+      "step": 8795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4019068765759455e-06,
+      "loss": 0.9974,
+      "step": 8796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3998665834083587e-06,
+      "loss": 0.8802,
+      "step": 8797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3978267769401808e-06,
+      "loss": 0.959,
+      "step": 8798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3957874573218207e-06,
+      "loss": 0.8169,
+      "step": 8799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3937486247036667e-06,
+      "loss": 0.7654,
+      "step": 8800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3917102792360577e-06,
+      "loss": 0.7946,
+      "step": 8801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3896724210693065e-06,
+      "loss": 0.8074,
+      "step": 8802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3876350503536903e-06,
+      "loss": 0.929,
+      "step": 8803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3855981672394376e-06,
+      "loss": 0.8173,
+      "step": 8804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3835617718767554e-06,
+      "loss": 0.8353,
+      "step": 8805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3815258644158122e-06,
+      "loss": 0.7782,
+      "step": 8806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3794904450067347e-06,
+      "loss": 0.8087,
+      "step": 8807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3774555137996103e-06,
+      "loss": 0.8882,
+      "step": 8808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375421070944507e-06,
+      "loss": 0.8541,
+      "step": 8809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3733871165914365e-06,
+      "loss": 0.8507,
+      "step": 8810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.37135365089039e-06,
+      "loss": 0.857,
+      "step": 8811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3693206739913187e-06,
+      "loss": 0.8672,
+      "step": 8812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3672881860441296e-06,
+      "loss": 0.8549,
+      "step": 8813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3652561871987044e-06,
+      "loss": 0.8909,
+      "step": 8814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3632246776048873e-06,
+      "loss": 0.2672,
+      "step": 8815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3611936574124803e-06,
+      "loss": 0.8868,
+      "step": 8816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359163126771248e-06,
+      "loss": 0.9488,
+      "step": 8817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3571330858309313e-06,
+      "loss": 0.8941,
+      "step": 8818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355103534741221e-06,
+      "loss": 0.8433,
+      "step": 8819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3530744736517793e-06,
+      "loss": 0.8611,
+      "step": 8820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3510459027122366e-06,
+      "loss": 0.8486,
+      "step": 8821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3490178220721724e-06,
+      "loss": 0.8669,
+      "step": 8822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3469902318811444e-06,
+      "loss": 0.9051,
+      "step": 8823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.344963132288671e-06,
+      "loss": 0.9173,
+      "step": 8824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342936523444226e-06,
+      "loss": 0.9041,
+      "step": 8825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3409104054972596e-06,
+      "loss": 0.81,
+      "step": 8826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3388847785971756e-06,
+      "loss": 0.7956,
+      "step": 8827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336859642893344e-06,
+      "loss": 0.7971,
+      "step": 8828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3348349985351016e-06,
+      "loss": 0.2432,
+      "step": 8829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3328108456717513e-06,
+      "loss": 0.8718,
+      "step": 8830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3307871844525486e-06,
+      "loss": 0.219,
+      "step": 8831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328764015026724e-06,
+      "loss": 0.7362,
+      "step": 8832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326741337543472e-06,
+      "loss": 0.8795,
+      "step": 8833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324719152151937e-06,
+      "loss": 0.9135,
+      "step": 8834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322697459001244e-06,
+      "loss": 0.915,
+      "step": 8835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320676258240474e-06,
+      "loss": 0.7413,
+      "step": 8836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3186555500186703e-06,
+      "loss": 0.8391,
+      "step": 8837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316635334484838e-06,
+      "loss": 0.913,
+      "step": 8838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.314615611787957e-06,
+      "loss": 0.9666,
+      "step": 8839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3125963820769547e-06,
+      "loss": 0.9021,
+      "step": 8840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3105776455007365e-06,
+      "loss": 0.9066,
+      "step": 8841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308559402208168e-06,
+      "loss": 0.885,
+      "step": 8842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30654165234807e-06,
+      "loss": 0.8066,
+      "step": 8843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304524396069234e-06,
+      "loss": 0.2618,
+      "step": 8844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302507633520421e-06,
+      "loss": 0.7932,
+      "step": 8845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30049136485034e-06,
+      "loss": 0.2596,
+      "step": 8846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2984755902076802e-06,
+      "loss": 0.9744,
+      "step": 8847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296460309741082e-06,
+      "loss": 0.8015,
+      "step": 8848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.294445523599151e-06,
+      "loss": 0.8538,
+      "step": 8849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292431231930463e-06,
+      "loss": 0.8449,
+      "step": 8850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2904174348835572e-06,
+      "loss": 0.8209,
+      "step": 8851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288404132606925e-06,
+      "loss": 0.8489,
+      "step": 8852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286391325249032e-06,
+      "loss": 0.7895,
+      "step": 8853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2843790129583085e-06,
+      "loss": 0.8837,
+      "step": 8854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2823671958831384e-06,
+      "loss": 0.8445,
+      "step": 8855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2803558741718755e-06,
+      "loss": 0.8641,
+      "step": 8856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2783450479728453e-06,
+      "loss": 0.8412,
+      "step": 8857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276334717434314e-06,
+      "loss": 0.8252,
+      "step": 8858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2743248827045315e-06,
+      "loss": 0.8419,
+      "step": 8859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2723155439317067e-06,
+      "loss": 0.9056,
+      "step": 8860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2703067012640065e-06,
+      "loss": 0.9267,
+      "step": 8861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268298354849565e-06,
+      "loss": 0.8746,
+      "step": 8862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2662905048364834e-06,
+      "loss": 0.8023,
+      "step": 8863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2642831513728167e-06,
+      "loss": 0.8593,
+      "step": 8864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262276294606591e-06,
+      "loss": 0.7856,
+      "step": 8865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260269934685798e-06,
+      "loss": 0.2587,
+      "step": 8866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2582640717583812e-06,
+      "loss": 0.8853,
+      "step": 8867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2562587059722605e-06,
+      "loss": 0.9271,
+      "step": 8868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254253837475311e-06,
+      "loss": 0.8339,
+      "step": 8869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2522494664153693e-06,
+      "loss": 0.9399,
+      "step": 8870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250245592940243e-06,
+      "loss": 0.8663,
+      "step": 8871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2482422171977037e-06,
+      "loss": 0.9445,
+      "step": 8872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246239339335474e-06,
+      "loss": 0.792,
+      "step": 8873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2442369595012514e-06,
+      "loss": 0.8841,
+      "step": 8874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.242235077842697e-06,
+      "loss": 0.7971,
+      "step": 8875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2402336945074244e-06,
+      "loss": 0.8812,
+      "step": 8876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2382328096430226e-06,
+      "loss": 0.7905,
+      "step": 8877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236232423397038e-06,
+      "loss": 0.8688,
+      "step": 8878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2342325359169744e-06,
+      "loss": 0.9359,
+      "step": 8879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232233147350311e-06,
+      "loss": 0.8492,
+      "step": 8880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230234257844487e-06,
+      "loss": 0.8612,
+      "step": 8881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228235867546894e-06,
+      "loss": 0.8579,
+      "step": 8882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2262379766049e-06,
+      "loss": 0.8725,
+      "step": 8883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2242405851658344e-06,
+      "loss": 0.8511,
+      "step": 8884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2222436933769795e-06,
+      "loss": 0.7688,
+      "step": 8885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.220247301385594e-06,
+      "loss": 0.8152,
+      "step": 8886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2182514093388873e-06,
+      "loss": 0.8351,
+      "step": 8887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216256017384045e-06,
+      "loss": 0.805,
+      "step": 8888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214261125668202e-06,
+      "loss": 0.2601,
+      "step": 8889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2122667343384705e-06,
+      "loss": 0.9015,
+      "step": 8890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2102728435419116e-06,
+      "loss": 0.903,
+      "step": 8891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.20827945342556e-06,
+      "loss": 0.8217,
+      "step": 8892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2062865641364126e-06,
+      "loss": 0.8247,
+      "step": 8893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204294175821421e-06,
+      "loss": 0.8245,
+      "step": 8894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2023022886275113e-06,
+      "loss": 0.8744,
+      "step": 8895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200310902701561e-06,
+      "loss": 0.8701,
+      "step": 8896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1983200181904193e-06,
+      "loss": 0.8232,
+      "step": 8897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1963296352409002e-06,
+      "loss": 0.8494,
+      "step": 8898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194339753999771e-06,
+      "loss": 0.8768,
+      "step": 8899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1923503746137663e-06,
+      "loss": 0.8606,
+      "step": 8900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190361497229586e-06,
+      "loss": 0.8224,
+      "step": 8901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1883731219938962e-06,
+      "loss": 0.2524,
+      "step": 8902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186385249053313e-06,
+      "loss": 0.905,
+      "step": 8903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1843978785544327e-06,
+      "loss": 0.8375,
+      "step": 8904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182411010643797e-06,
+      "loss": 0.8726,
+      "step": 8905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1804246454679243e-06,
+      "loss": 0.829,
+      "step": 8906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178438783173292e-06,
+      "loss": 0.8659,
+      "step": 8907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.176453423906335e-06,
+      "loss": 0.833,
+      "step": 8908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174468567813461e-06,
+      "loss": 0.9403,
+      "step": 8909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1724842150410275e-06,
+      "loss": 0.8057,
+      "step": 8910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1705003657353706e-06,
+      "loss": 0.8774,
+      "step": 8911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1685170200427728e-06,
+      "loss": 0.8195,
+      "step": 8912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166534178109495e-06,
+      "loss": 0.8895,
+      "step": 8913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1645518400817475e-06,
+      "loss": 0.822,
+      "step": 8914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162570006105713e-06,
+      "loss": 0.8862,
+      "step": 8915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1605886763275352e-06,
+      "loss": 0.8384,
+      "step": 8916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1586078508933136e-06,
+      "loss": 0.84,
+      "step": 8917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15662752994912e-06,
+      "loss": 0.7855,
+      "step": 8918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154647713640987e-06,
+      "loss": 0.274,
+      "step": 8919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152668402114906e-06,
+      "loss": 0.8218,
+      "step": 8920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150689595516827e-06,
+      "loss": 0.9222,
+      "step": 8921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148711293992679e-06,
+      "loss": 0.8834,
+      "step": 8922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1467334976883347e-06,
+      "loss": 0.8985,
+      "step": 8923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144756206749643e-06,
+      "loss": 0.8932,
+      "step": 8924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142779421322413e-06,
+      "loss": 0.83,
+      "step": 8925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1408031415524085e-06,
+      "loss": 0.9131,
+      "step": 8926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1388273675853665e-06,
+      "loss": 0.8445,
+      "step": 8927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136852099566984e-06,
+      "loss": 0.8669,
+      "step": 8928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1348773376429155e-06,
+      "loss": 0.8882,
+      "step": 8929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132903081958779e-06,
+      "loss": 0.9122,
+      "step": 8930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1309293326601653e-06,
+      "loss": 0.8714,
+      "step": 8931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128956089892613e-06,
+      "loss": 0.8381,
+      "step": 8932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1269833538016324e-06,
+      "loss": 0.7892,
+      "step": 8933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125011124532701e-06,
+      "loss": 0.8641,
+      "step": 8934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123039402231244e-06,
+      "loss": 0.8979,
+      "step": 8935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1210681870426608e-06,
+      "loss": 0.7516,
+      "step": 8936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119097479112315e-06,
+      "loss": 0.9001,
+      "step": 8937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117127278585521e-06,
+      "loss": 0.8645,
+      "step": 8938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115157585607568e-06,
+      "loss": 0.8777,
+      "step": 8939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1131884003237055e-06,
+      "loss": 0.322,
+      "step": 8940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111219722879134e-06,
+      "loss": 0.9275,
+      "step": 8941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1092515534190305e-06,
+      "loss": 0.8722,
+      "step": 8942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1072838920885318e-06,
+      "loss": 0.788,
+      "step": 8943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105316739032729e-06,
+      "loss": 0.9737,
+      "step": 8944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1033500943966853e-06,
+      "loss": 0.9162,
+      "step": 8945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1013839583254255e-06,
+      "loss": 0.8886,
+      "step": 8946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0994183309639283e-06,
+      "loss": 0.8442,
+      "step": 8947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097453212457142e-06,
+      "loss": 0.8337,
+      "step": 8948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095488602949982e-06,
+      "loss": 0.9678,
+      "step": 8949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093524502587315e-06,
+      "loss": 0.88,
+      "step": 8950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091560911513973e-06,
+      "loss": 0.8764,
+      "step": 8951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0895978298747607e-06,
+      "loss": 0.8892,
+      "step": 8952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0876352578144287e-06,
+      "loss": 0.8654,
+      "step": 8953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085673195477702e-06,
+      "loss": 0.8939,
+      "step": 8954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0837116430092697e-06,
+      "loss": 0.8409,
+      "step": 8955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.08175060055377e-06,
+      "loss": 0.8319,
+      "step": 8956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0797900682558168e-06,
+      "loss": 0.8545,
+      "step": 8957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077830046259984e-06,
+      "loss": 0.834,
+      "step": 8958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075870534710799e-06,
+      "loss": 0.8858,
+      "step": 8959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0739115337527626e-06,
+      "loss": 0.8594,
+      "step": 8960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.071953043530331e-06,
+      "loss": 0.2565,
+      "step": 8961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0699950641879237e-06,
+      "loss": 0.924,
+      "step": 8962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068037595869924e-06,
+      "loss": 0.2471,
+      "step": 8963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066080638720682e-06,
+      "loss": 0.8629,
+      "step": 8964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064124192884499e-06,
+      "loss": 0.8671,
+      "step": 8965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0621682585056477e-06,
+      "loss": 0.8092,
+      "step": 8966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0602128357283632e-06,
+      "loss": 0.814,
+      "step": 8967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058257924696835e-06,
+      "loss": 0.8167,
+      "step": 8968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056303525555222e-06,
+      "loss": 0.9079,
+      "step": 8969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0543496384476455e-06,
+      "loss": 0.8013,
+      "step": 8970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052396263518186e-06,
+      "loss": 0.8386,
+      "step": 8971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050443400910882e-06,
+      "loss": 0.9202,
+      "step": 8972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0484910507697463e-06,
+      "loss": 0.2692,
+      "step": 8973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0465392132387407e-06,
+      "loss": 0.8421,
+      "step": 8974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0445878884617974e-06,
+      "loss": 0.8436,
+      "step": 8975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042637076582814e-06,
+      "loss": 0.8773,
+      "step": 8976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040686777745636e-06,
+      "loss": 0.9706,
+      "step": 8977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0387369920940856e-06,
+      "loss": 0.8233,
+      "step": 8978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036787719771943e-06,
+      "loss": 0.8848,
+      "step": 8979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0348389609229432e-06,
+      "loss": 0.7566,
+      "step": 8980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032890715690797e-06,
+      "loss": 0.8604,
+      "step": 8981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.030942984219166e-06,
+      "loss": 0.8237,
+      "step": 8982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0289957666516735e-06,
+      "loss": 0.9324,
+      "step": 8983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027049063131913e-06,
+      "loss": 0.8858,
+      "step": 8984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025102873803438e-06,
+      "loss": 0.8845,
+      "step": 8985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0231571988097576e-06,
+      "loss": 0.8159,
+      "step": 8986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021212038294349e-06,
+      "loss": 0.8326,
+      "step": 8987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0192673924006546e-06,
+      "loss": 0.8701,
+      "step": 8988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017323261272066e-06,
+      "loss": 0.2959,
+      "step": 8989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0153796450519533e-06,
+      "loss": 0.8775,
+      "step": 8990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0134365438836365e-06,
+      "loss": 0.8847,
+      "step": 8991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0114939579103985e-06,
+      "loss": 0.8511,
+      "step": 8992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0095518872754903e-06,
+      "loss": 0.2337,
+      "step": 8993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0076103321221252e-06,
+      "loss": 0.8224,
+      "step": 8994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0056692925934684e-06,
+      "loss": 0.9447,
+      "step": 8995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0037287688326566e-06,
+      "loss": 0.8127,
+      "step": 8996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001788760982789e-06,
+      "loss": 0.9082,
+      "step": 8997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9998492691869185e-06,
+      "loss": 0.9284,
+      "step": 8998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99791029358807e-06,
+      "loss": 0.8917,
+      "step": 8999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995971834329218e-06,
+      "loss": 0.7458,
+      "step": 9000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9940338915533106e-06,
+      "loss": 0.7867,
+      "step": 9001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992096465403257e-06,
+      "loss": 0.9237,
+      "step": 9002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99015955602192e-06,
+      "loss": 0.7627,
+      "step": 9003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988223163552125e-06,
+      "loss": 0.8157,
+      "step": 9004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.98628728813667e-06,
+      "loss": 0.9421,
+      "step": 9005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9843519299183076e-06,
+      "loss": 0.9329,
+      "step": 9006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982417089039749e-06,
+      "loss": 0.8332,
+      "step": 9007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9804827656436753e-06,
+      "loss": 0.8253,
+      "step": 9008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9785489598727203e-06,
+      "loss": 0.8665,
+      "step": 9009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976615671869486e-06,
+      "loss": 0.8392,
+      "step": 9010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.97468290177654e-06,
+      "loss": 0.8917,
+      "step": 9011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9727506497364013e-06,
+      "loss": 0.2476,
+      "step": 9012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9708189158915535e-06,
+      "loss": 0.9287,
+      "step": 9013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9688877003844474e-06,
+      "loss": 0.8158,
+      "step": 9014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966957003357496e-06,
+      "loss": 0.8812,
+      "step": 9015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9650268249530636e-06,
+      "loss": 0.8088,
+      "step": 9016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.963097165313489e-06,
+      "loss": 0.8369,
+      "step": 9017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9611680245810627e-06,
+      "loss": 0.9383,
+      "step": 9018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959239402898042e-06,
+      "loss": 0.8456,
+      "step": 9019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957311300406649e-06,
+      "loss": 0.9239,
+      "step": 9020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9553837172490576e-06,
+      "loss": 0.8557,
+      "step": 9021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953456653567416e-06,
+      "loss": 0.9338,
+      "step": 9022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9515301095038206e-06,
+      "loss": 0.8092,
+      "step": 9023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9496040852003428e-06,
+      "loss": 0.2512,
+      "step": 9024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9476785807990028e-06,
+      "loss": 0.8174,
+      "step": 9025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.945753596441796e-06,
+      "loss": 0.9391,
+      "step": 9026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9438291322706646e-06,
+      "loss": 0.7883,
+      "step": 9027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941905188427525e-06,
+      "loss": 0.8802,
+      "step": 9028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939981765054253e-06,
+      "loss": 0.8818,
+      "step": 9029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9380588622926765e-06,
+      "loss": 0.8349,
+      "step": 9030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9361364802845958e-06,
+      "loss": 0.9792,
+      "step": 9031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9342146191717723e-06,
+      "loss": 0.8911,
+      "step": 9032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9322932790959224e-06,
+      "loss": 0.8658,
+      "step": 9033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9303724601987236e-06,
+      "loss": 0.8358,
+      "step": 9034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9284521626218254e-06,
+      "loss": 0.8723,
+      "step": 9035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926532386506825e-06,
+      "loss": 0.8313,
+      "step": 9036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924613131995293e-06,
+      "loss": 0.772,
+      "step": 9037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9226943992287604e-06,
+      "loss": 0.2747,
+      "step": 9038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9207761883487074e-06,
+      "loss": 0.903,
+      "step": 9039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9188584994965896e-06,
+      "loss": 0.8654,
+      "step": 9040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9169413328138208e-06,
+      "loss": 0.8315,
+      "step": 9041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91502468844177e-06,
+      "loss": 0.8467,
+      "step": 9042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913108566521776e-06,
+      "loss": 0.8548,
+      "step": 9043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9111929671951346e-06,
+      "loss": 0.8575,
+      "step": 9044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909277890603099e-06,
+      "loss": 0.8847,
+      "step": 9045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9073633368868925e-06,
+      "loss": 0.9257,
+      "step": 9046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9054493061876988e-06,
+      "loss": 0.7474,
+      "step": 9047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9035357986466528e-06,
+      "loss": 0.9277,
+      "step": 9048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901622814404863e-06,
+      "loss": 0.8083,
+      "step": 9049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8997103536033976e-06,
+      "loss": 0.8503,
+      "step": 9050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897798416383275e-06,
+      "loss": 0.287,
+      "step": 9051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.895887002885488e-06,
+      "loss": 0.8335,
+      "step": 9052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8939761132509913e-06,
+      "loss": 0.8069,
+      "step": 9053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8920657476206826e-06,
+      "loss": 0.851,
+      "step": 9054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8901559061354412e-06,
+      "loss": 0.8063,
+      "step": 9055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888246588936102e-06,
+      "loss": 0.8723,
+      "step": 9056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8863377961634554e-06,
+      "loss": 0.8311,
+      "step": 9057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884429527958259e-06,
+      "loss": 0.8511,
+      "step": 9058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882521784461234e-06,
+      "loss": 0.8357,
+      "step": 9059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8806145658130523e-06,
+      "loss": 0.8762,
+      "step": 9060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878707872154357e-06,
+      "loss": 0.8266,
+      "step": 9061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8768017036257524e-06,
+      "loss": 0.2506,
+      "step": 9062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8748960603677956e-06,
+      "loss": 0.8213,
+      "step": 9063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8729909425210157e-06,
+      "loss": 0.8762,
+      "step": 9064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8710863502258933e-06,
+      "loss": 0.7762,
+      "step": 9065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8691822836228744e-06,
+      "loss": 0.3024,
+      "step": 9066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8672787428523686e-06,
+      "loss": 0.8908,
+      "step": 9067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8653757280547456e-06,
+      "loss": 0.8632,
+      "step": 9068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8634732393703323e-06,
+      "loss": 0.9141,
+      "step": 9069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861571276939421e-06,
+      "loss": 0.8549,
+      "step": 9070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8596698409022673e-06,
+      "loss": 0.8022,
+      "step": 9071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8577689313990788e-06,
+      "loss": 0.8159,
+      "step": 9072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8558685485700323e-06,
+      "loss": 0.8252,
+      "step": 9073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8539686925552725e-06,
+      "loss": 0.923,
+      "step": 9074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8520693634948804e-06,
+      "loss": 0.8357,
+      "step": 9075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850170561528923e-06,
+      "loss": 0.9017,
+      "step": 9076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848272286797421e-06,
+      "loss": 0.7924,
+      "step": 9077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846374539440351e-06,
+      "loss": 0.781,
+      "step": 9078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844477319597655e-06,
+      "loss": 0.8488,
+      "step": 9079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8425806274092396e-06,
+      "loss": 0.8716,
+      "step": 9080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8406844630149622e-06,
+      "loss": 0.9483,
+      "step": 9081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8387888265546505e-06,
+      "loss": 0.8107,
+      "step": 9082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8368937181680943e-06,
+      "loss": 0.8013,
+      "step": 9083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834999137995036e-06,
+      "loss": 0.8306,
+      "step": 9084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8331050861751817e-06,
+      "loss": 0.8141,
+      "step": 9085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831211562848206e-06,
+      "loss": 0.7941,
+      "step": 9086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8293185681537337e-06,
+      "loss": 0.8059,
+      "step": 9087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8274261022313576e-06,
+      "loss": 0.9368,
+      "step": 9088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8255341652206326e-06,
+      "loss": 0.2368,
+      "step": 9089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823642757261067e-06,
+      "loss": 0.9564,
+      "step": 9090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821751878492137e-06,
+      "loss": 0.8749,
+      "step": 9091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8198615290532805e-06,
+      "loss": 0.8586,
+      "step": 9092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8179717090838877e-06,
+      "loss": 0.7473,
+      "step": 9093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8160824187233217e-06,
+      "loss": 0.8555,
+      "step": 9094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8141936581108975e-06,
+      "loss": 0.8362,
+      "step": 9095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8123054273858896e-06,
+      "loss": 0.8629,
+      "step": 9096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8104177266875434e-06,
+      "loss": 0.8387,
+      "step": 9097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8085305561550593e-06,
+      "loss": 0.9106,
+      "step": 9098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066439159275948e-06,
+      "loss": 0.8749,
+      "step": 9099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804757806144275e-06,
+      "loss": 0.8539,
+      "step": 9100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8028722269441856e-06,
+      "loss": 0.8397,
+      "step": 9101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8009871784663656e-06,
+      "loss": 0.23,
+      "step": 9102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7991026608498264e-06,
+      "loss": 0.8574,
+      "step": 9103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7972186742335273e-06,
+      "loss": 0.7672,
+      "step": 9104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795335218756401e-06,
+      "loss": 0.8608,
+      "step": 9105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793452294557328e-06,
+      "loss": 0.8007,
+      "step": 9106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791569901775166e-06,
+      "loss": 0.7706,
+      "step": 9107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7896880405487147e-06,
+      "loss": 0.713,
+      "step": 9108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7878067110167493e-06,
+      "loss": 0.8118,
+      "step": 9109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7859259133180026e-06,
+      "loss": 0.9332,
+      "step": 9110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7840456475911604e-06,
+      "loss": 0.7448,
+      "step": 9111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7821659139748813e-06,
+      "loss": 0.9131,
+      "step": 9112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7802867126077728e-06,
+      "loss": 0.788,
+      "step": 9113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77840804362841e-06,
+      "loss": 0.8702,
+      "step": 9114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776529907175334e-06,
+      "loss": 0.8212,
+      "step": 9115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7746523033870333e-06,
+      "loss": 0.9011,
+      "step": 9116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772775232401963e-06,
+      "loss": 0.8896,
+      "step": 9117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7708986943585424e-06,
+      "loss": 0.6985,
+      "step": 9118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7690226893951533e-06,
+      "loss": 0.8693,
+      "step": 9119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7671472176501255e-06,
+      "loss": 0.8395,
+      "step": 9120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7652722792617657e-06,
+      "loss": 0.845,
+      "step": 9121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763397874368328e-06,
+      "loss": 0.7968,
+      "step": 9122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7615240031080336e-06,
+      "loss": 0.7957,
+      "step": 9123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.759650665619067e-06,
+      "loss": 0.8522,
+      "step": 9124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7577778620395645e-06,
+      "loss": 0.9115,
+      "step": 9125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755905592507633e-06,
+      "loss": 0.8824,
+      "step": 9126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540338571613302e-06,
+      "loss": 0.8999,
+      "step": 9127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752162656138685e-06,
+      "loss": 0.8108,
+      "step": 9128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750291989577676e-06,
+      "loss": 0.8315,
+      "step": 9129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7484218576162523e-06,
+      "loss": 0.7985,
+      "step": 9130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7465522603923145e-06,
+      "loss": 0.8837,
+      "step": 9131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7446831980437306e-06,
+      "loss": 0.8184,
+      "step": 9132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7428146707083303e-06,
+      "loss": 0.7002,
+      "step": 9133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7409466785238937e-06,
+      "loss": 0.8163,
+      "step": 9134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7390792216281716e-06,
+      "loss": 0.8838,
+      "step": 9135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7372123001588745e-06,
+      "loss": 0.823,
+      "step": 9136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7353459142536675e-06,
+      "loss": 0.8506,
+      "step": 9137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733480064050178e-06,
+      "loss": 0.8958,
+      "step": 9138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731614749685999e-06,
+      "loss": 0.8248,
+      "step": 9139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7297499712986765e-06,
+      "loss": 0.827,
+      "step": 9140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727885729025723e-06,
+      "loss": 0.8622,
+      "step": 9141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726022023004612e-06,
+      "loss": 0.8772,
+      "step": 9142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7241588533727693e-06,
+      "loss": 0.7934,
+      "step": 9143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7222962202675883e-06,
+      "loss": 0.8129,
+      "step": 9144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7204341238264264e-06,
+      "loss": 0.8528,
+      "step": 9145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7185725641865913e-06,
+      "loss": 0.8689,
+      "step": 9146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716711541485354e-06,
+      "loss": 0.7664,
+      "step": 9147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.714851055859953e-06,
+      "loss": 0.9179,
+      "step": 9148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7129911074475766e-06,
+      "loss": 0.86,
+      "step": 9149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7111316963853817e-06,
+      "loss": 0.8945,
+      "step": 9150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709272822810486e-06,
+      "loss": 0.871,
+      "step": 9151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7074144868599595e-06,
+      "loss": 0.8393,
+      "step": 9152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7055566886708382e-06,
+      "loss": 0.2518,
+      "step": 9153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703699428380123e-06,
+      "loss": 0.8716,
+      "step": 9154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7018427061247623e-06,
+      "loss": 0.8101,
+      "step": 9155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6999865220416753e-06,
+      "loss": 0.8669,
+      "step": 9156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698130876267746e-06,
+      "loss": 0.7501,
+      "step": 9157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696275768939798e-06,
+      "loss": 0.8842,
+      "step": 9158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6944212001946344e-06,
+      "loss": 0.8643,
+      "step": 9159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692567170169017e-06,
+      "loss": 0.9205,
+      "step": 9160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6907136789996558e-06,
+      "loss": 0.2351,
+      "step": 9161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6888607268232337e-06,
+      "loss": 0.9244,
+      "step": 9162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6870083137763902e-06,
+      "loss": 0.8698,
+      "step": 9163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6851564399957188e-06,
+      "loss": 0.983,
+      "step": 9164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6833051056177805e-06,
+      "loss": 0.8973,
+      "step": 9165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681454310779098e-06,
+      "loss": 0.8663,
+      "step": 9166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6796040556161484e-06,
+      "loss": 0.2497,
+      "step": 9167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677754340265365e-06,
+      "loss": 0.8728,
+      "step": 9168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6759051648631564e-06,
+      "loss": 0.8572,
+      "step": 9169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674056529545875e-06,
+      "loss": 0.7451,
+      "step": 9170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6722084344498434e-06,
+      "loss": 0.8448,
+      "step": 9171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6703608797113454e-06,
+      "loss": 0.7747,
+      "step": 9172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6685138654666154e-06,
+      "loss": 0.8666,
+      "step": 9173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666667391851857e-06,
+      "loss": 0.9565,
+      "step": 9174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664821459003233e-06,
+      "loss": 0.8741,
+      "step": 9175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6629760670568574e-06,
+      "loss": 0.8381,
+      "step": 9176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6611312161488177e-06,
+      "loss": 0.9271,
+      "step": 9177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6592869064151527e-06,
+      "loss": 0.8239,
+      "step": 9178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6574431379918585e-06,
+      "loss": 0.8619,
+      "step": 9179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6555999110149012e-06,
+      "loss": 0.8542,
+      "step": 9180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6537572256202027e-06,
+      "loss": 0.75,
+      "step": 9181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65191508194364e-06,
+      "loss": 0.8103,
+      "step": 9182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650073480121057e-06,
+      "loss": 0.7812,
+      "step": 9183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648232420288257e-06,
+      "loss": 0.9452,
+      "step": 9184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646391902580997e-06,
+      "loss": 0.9498,
+      "step": 9185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6445519271349983e-06,
+      "loss": 0.8048,
+      "step": 9186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6427124940859517e-06,
+      "loss": 0.8239,
+      "step": 9187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640873603569485e-06,
+      "loss": 0.8746,
+      "step": 9188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639035255721205e-06,
+      "loss": 0.8205,
+      "step": 9189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6371974506766775e-06,
+      "loss": 0.8521,
+      "step": 9190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6353601885714154e-06,
+      "loss": 0.9402,
+      "step": 9191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6335234695409053e-06,
+      "loss": 0.847,
+      "step": 9192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6316872937205905e-06,
+      "loss": 1.0086,
+      "step": 9193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629851661245867e-06,
+      "loss": 0.2756,
+      "step": 9194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6280165722520965e-06,
+      "loss": 0.75,
+      "step": 9195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626182026874606e-06,
+      "loss": 0.8003,
+      "step": 9196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6243480252486686e-06,
+      "loss": 0.7624,
+      "step": 9197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622514567509531e-06,
+      "loss": 0.804,
+      "step": 9198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6206816537923917e-06,
+      "loss": 0.9692,
+      "step": 9199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6188492842324088e-06,
+      "loss": 0.7924,
+      "step": 9200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6170174589647046e-06,
+      "loss": 0.9162,
+      "step": 9201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6151861781243627e-06,
+      "loss": 0.807,
+      "step": 9202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613355441846418e-06,
+      "loss": 0.8474,
+      "step": 9203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611525250265873e-06,
+      "loss": 0.9572,
+      "step": 9204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60969560351769e-06,
+      "loss": 0.8722,
+      "step": 9205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6078665017367844e-06,
+      "loss": 0.8175,
+      "step": 9206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6060379450580396e-06,
+      "loss": 0.877,
+      "step": 9207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604209933616293e-06,
+      "loss": 0.7975,
+      "step": 9208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6023824675463404e-06,
+      "loss": 0.7723,
+      "step": 9209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6005555469829445e-06,
+      "loss": 0.8636,
+      "step": 9210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5987291720608264e-06,
+      "loss": 0.726,
+      "step": 9211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5969033429146585e-06,
+      "loss": 0.9105,
+      "step": 9212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5950780596790816e-06,
+      "loss": 0.8441,
+      "step": 9213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593253322488697e-06,
+      "loss": 0.6489,
+      "step": 9214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591429131478057e-06,
+      "loss": 0.8498,
+      "step": 9215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5896054867816835e-06,
+      "loss": 0.8723,
+      "step": 9216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5877823885340493e-06,
+      "loss": 0.876,
+      "step": 9217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585959836869595e-06,
+      "loss": 0.8383,
+      "step": 9218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584137831922714e-06,
+      "loss": 0.8812,
+      "step": 9219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5823163738277657e-06,
+      "loss": 0.8789,
+      "step": 9220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5804954627190627e-06,
+      "loss": 0.8187,
+      "step": 9221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5786750987308807e-06,
+      "loss": 0.8504,
+      "step": 9222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576855281997459e-06,
+      "loss": 0.7903,
+      "step": 9223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575036012652987e-06,
+      "loss": 0.8233,
+      "step": 9224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5732172908316254e-06,
+      "loss": 0.8904,
+      "step": 9225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5713991166674813e-06,
+      "loss": 0.7504,
+      "step": 9226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5695814902946314e-06,
+      "loss": 0.9417,
+      "step": 9227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5677644118471125e-06,
+      "loss": 0.7874,
+      "step": 9228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5659478814589146e-06,
+      "loss": 0.8751,
+      "step": 9229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5641318992639862e-06,
+      "loss": 0.8083,
+      "step": 9230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562316465396244e-06,
+      "loss": 0.8493,
+      "step": 9231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5605015799895617e-06,
+      "loss": 0.8209,
+      "step": 9232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5586872431777644e-06,
+      "loss": 0.887,
+      "step": 9233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5568734550946483e-06,
+      "loss": 0.8693,
+      "step": 9234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.555060215873958e-06,
+      "loss": 0.8538,
+      "step": 9235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5532475256494073e-06,
+      "loss": 0.886,
+      "step": 9236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551435384554668e-06,
+      "loss": 0.93,
+      "step": 9237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5496237927233624e-06,
+      "loss": 0.8481,
+      "step": 9238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547812750289085e-06,
+      "loss": 0.7588,
+      "step": 9239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546002257385377e-06,
+      "loss": 0.7434,
+      "step": 9240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5441923141457536e-06,
+      "loss": 0.8371,
+      "step": 9241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5423829207036733e-06,
+      "loss": 0.8437,
+      "step": 9242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54057407719257e-06,
+      "loss": 0.2716,
+      "step": 9243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5387657837458225e-06,
+      "loss": 0.933,
+      "step": 9244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5369580404967787e-06,
+      "loss": 0.9711,
+      "step": 9245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5351508475787467e-06,
+      "loss": 0.8883,
+      "step": 9246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5333442051249836e-06,
+      "loss": 0.833,
+      "step": 9247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5315381132687177e-06,
+      "loss": 0.8367,
+      "step": 9248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.529732572143132e-06,
+      "loss": 0.8495,
+      "step": 9249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5279275818813674e-06,
+      "loss": 0.8685,
+      "step": 9250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5261231426165224e-06,
+      "loss": 0.8252,
+      "step": 9251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524319254481663e-06,
+      "loss": 0.813,
+      "step": 9252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5225159176098036e-06,
+      "loss": 0.8307,
+      "step": 9253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5207131321339274e-06,
+      "loss": 0.9025,
+      "step": 9254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518910898186976e-06,
+      "loss": 0.7291,
+      "step": 9255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5171092159018416e-06,
+      "loss": 0.9386,
+      "step": 9256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5153080854113856e-06,
+      "loss": 0.8733,
+      "step": 9257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513507506848426e-06,
+      "loss": 0.8782,
+      "step": 9258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5117074803457343e-06,
+      "loss": 0.211,
+      "step": 9259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5099080060360526e-06,
+      "loss": 0.8794,
+      "step": 9260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5081090840520706e-06,
+      "loss": 0.9162,
+      "step": 9261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506310714526441e-06,
+      "loss": 0.8495,
+      "step": 9262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50451289759178e-06,
+      "loss": 0.9222,
+      "step": 9263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502715633380662e-06,
+      "loss": 0.8121,
+      "step": 9264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5009189220256137e-06,
+      "loss": 0.919,
+      "step": 9265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4991227636591297e-06,
+      "loss": 0.8386,
+      "step": 9266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4973271584136615e-06,
+      "loss": 0.8941,
+      "step": 9267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4955321064216143e-06,
+      "loss": 0.8973,
+      "step": 9268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4937376078153587e-06,
+      "loss": 0.8556,
+      "step": 9269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4919436627272286e-06,
+      "loss": 0.8774,
+      "step": 9270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4901502712894985e-06,
+      "loss": 0.9326,
+      "step": 9271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488357433634423e-06,
+      "loss": 0.8626,
+      "step": 9272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4865651498942077e-06,
+      "loss": 0.9,
+      "step": 9273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484773420201012e-06,
+      "loss": 0.9592,
+      "step": 9274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4829822446869634e-06,
+      "loss": 0.9185,
+      "step": 9275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481191623484147e-06,
+      "loss": 0.8637,
+      "step": 9276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4794015567245986e-06,
+      "loss": 0.8029,
+      "step": 9277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4776120445403218e-06,
+      "loss": 0.9376,
+      "step": 9278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4758230870632815e-06,
+      "loss": 0.8351,
+      "step": 9279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4740346844253914e-06,
+      "loss": 0.9448,
+      "step": 9280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4722468367585293e-06,
+      "loss": 0.7694,
+      "step": 9281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4704595441945377e-06,
+      "loss": 0.8195,
+      "step": 9282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4686728068652065e-06,
+      "loss": 0.826,
+      "step": 9283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4668866249022936e-06,
+      "loss": 0.9323,
+      "step": 9284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4651009984375183e-06,
+      "loss": 0.9402,
+      "step": 9285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4633159276025475e-06,
+      "loss": 0.8665,
+      "step": 9286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4615314125290166e-06,
+      "loss": 0.8723,
+      "step": 9287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459747453348521e-06,
+      "loss": 0.2671,
+      "step": 9288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4579640501926037e-06,
+      "loss": 0.7925,
+      "step": 9289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45618120319278e-06,
+      "loss": 0.8234,
+      "step": 9290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4543989124805214e-06,
+      "loss": 0.8805,
+      "step": 9291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4526171781872454e-06,
+      "loss": 0.7821,
+      "step": 9292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450836000444344e-06,
+      "loss": 0.8531,
+      "step": 9293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4490553793831672e-06,
+      "loss": 0.81,
+      "step": 9294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4472753151350113e-06,
+      "loss": 0.8776,
+      "step": 9295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.445495807831144e-06,
+      "loss": 0.8058,
+      "step": 9296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4437168576027915e-06,
+      "loss": 0.8117,
+      "step": 9297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441938464581126e-06,
+      "loss": 0.2358,
+      "step": 9298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4401606288972945e-06,
+      "loss": 0.9319,
+      "step": 9299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438383350682396e-06,
+      "loss": 0.87,
+      "step": 9300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436606630067487e-06,
+      "loss": 0.9346,
+      "step": 9301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434830467183582e-06,
+      "loss": 0.7449,
+      "step": 9302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4330548621616613e-06,
+      "loss": 0.8582,
+      "step": 9303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4312798151326543e-06,
+      "loss": 0.8726,
+      "step": 9304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429505326227457e-06,
+      "loss": 0.8035,
+      "step": 9305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4277313955769257e-06,
+      "loss": 0.7141,
+      "step": 9306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4259580233118653e-06,
+      "loss": 0.8838,
+      "step": 9307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424185209563047e-06,
+      "loss": 0.9261,
+      "step": 9308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422412954461204e-06,
+      "loss": 0.8847,
+      "step": 9309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4206412581370187e-06,
+      "loss": 0.8149,
+      "step": 9310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4188701207211417e-06,
+      "loss": 0.9213,
+      "step": 9311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417099542344177e-06,
+      "loss": 0.9443,
+      "step": 9312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4153295231366836e-06,
+      "loss": 0.9111,
+      "step": 9313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413560063229189e-06,
+      "loss": 0.8624,
+      "step": 9314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411791162752176e-06,
+      "loss": 0.9227,
+      "step": 9315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410022821836081e-06,
+      "loss": 0.8566,
+      "step": 9316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4082550406113035e-06,
+      "loss": 0.9635,
+      "step": 9317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4064878192082055e-06,
+      "loss": 0.9278,
+      "step": 9318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404721157757097e-06,
+      "loss": 0.9338,
+      "step": 9319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4029550563882576e-06,
+      "loss": 0.8434,
+      "step": 9320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4011895152319208e-06,
+      "loss": 0.2783,
+      "step": 9321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39942453441828e-06,
+      "loss": 0.2241,
+      "step": 9322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39766011407748e-06,
+      "loss": 0.7573,
+      "step": 9323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395896254339639e-06,
+      "loss": 0.8016,
+      "step": 9324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394132955334818e-06,
+      "loss": 0.8143,
+      "step": 9325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3923702171930486e-06,
+      "loss": 0.9417,
+      "step": 9326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3906080400443176e-06,
+      "loss": 0.795,
+      "step": 9327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388846424018566e-06,
+      "loss": 0.8601,
+      "step": 9328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3870853692456973e-06,
+      "loss": 0.2852,
+      "step": 9329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3853248758555782e-06,
+      "loss": 0.7821,
+      "step": 9330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383564943978022e-06,
+      "loss": 0.8432,
+      "step": 9331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3818055737428136e-06,
+      "loss": 0.7861,
+      "step": 9332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3800467652796867e-06,
+      "loss": 0.8598,
+      "step": 9333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378288518718336e-06,
+      "loss": 0.8737,
+      "step": 9334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3765308341884173e-06,
+      "loss": 0.7205,
+      "step": 9335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374773711819549e-06,
+      "loss": 0.2516,
+      "step": 9336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373017151741295e-06,
+      "loss": 0.8565,
+      "step": 9337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3712611540831888e-06,
+      "loss": 0.8297,
+      "step": 9338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3695057189747216e-06,
+      "loss": 0.9372,
+      "step": 9339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3677508465453357e-06,
+      "loss": 0.866,
+      "step": 9340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365996536924443e-06,
+      "loss": 0.8784,
+      "step": 9341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3642427902414046e-06,
+      "loss": 0.943,
+      "step": 9342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3624896066255385e-06,
+      "loss": 0.8843,
+      "step": 9343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3607369862061312e-06,
+      "loss": 0.8945,
+      "step": 9344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3589849291124255e-06,
+      "loss": 0.848,
+      "step": 9345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3572334354736116e-06,
+      "loss": 0.8433,
+      "step": 9346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3554825054188503e-06,
+      "loss": 0.8473,
+      "step": 9347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3537321390772607e-06,
+      "loss": 0.2307,
+      "step": 9348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351982336577908e-06,
+      "loss": 0.832,
+      "step": 9349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3502330980498313e-06,
+      "loss": 0.2267,
+      "step": 9350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348484423622015e-06,
+      "loss": 0.7705,
+      "step": 9351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3467363134234113e-06,
+      "loss": 0.9933,
+      "step": 9352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3449887675829285e-06,
+      "loss": 0.2586,
+      "step": 9353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343241786229431e-06,
+      "loss": 0.8762,
+      "step": 9354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.34149536949174e-06,
+      "loss": 0.8777,
+      "step": 9355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3397495174986383e-06,
+      "loss": 0.7469,
+      "step": 9356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338004230378872e-06,
+      "loss": 0.9052,
+      "step": 9357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336259508261132e-06,
+      "loss": 0.7956,
+      "step": 9358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3345153512740826e-06,
+      "loss": 0.8496,
+      "step": 9359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3327717595463338e-06,
+      "loss": 0.8325,
+      "step": 9360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3310287332064627e-06,
+      "loss": 0.8343,
+      "step": 9361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3292862723830034e-06,
+      "loss": 0.8997,
+      "step": 9362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3275443772044438e-06,
+      "loss": 0.83,
+      "step": 9363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3258030477992298e-06,
+      "loss": 0.8553,
+      "step": 9364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3240622842957728e-06,
+      "loss": 0.808,
+      "step": 9365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322322086822438e-06,
+      "loss": 0.8604,
+      "step": 9366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3205824555075475e-06,
+      "loss": 0.8423,
+      "step": 9367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318843390479385e-06,
+      "loss": 0.8713,
+      "step": 9368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3171048918661863e-06,
+      "loss": 0.8713,
+      "step": 9369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3153669597961525e-06,
+      "loss": 0.8509,
+      "step": 9370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313629594397444e-06,
+      "loss": 0.7993,
+      "step": 9371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31189279579817e-06,
+      "loss": 0.9216,
+      "step": 9372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310156564126407e-06,
+      "loss": 0.8472,
+      "step": 9373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308420899510182e-06,
+      "loss": 0.8245,
+      "step": 9374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3066858020774906e-06,
+      "loss": 0.8226,
+      "step": 9375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3049512719562726e-06,
+      "loss": 0.7751,
+      "step": 9376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3032173092744427e-06,
+      "loss": 0.7939,
+      "step": 9377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3014839141598557e-06,
+      "loss": 0.8044,
+      "step": 9378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2997510867403384e-06,
+      "loss": 0.8309,
+      "step": 9379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2980188271436734e-06,
+      "loss": 0.265,
+      "step": 9380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2962871354975937e-06,
+      "loss": 0.7855,
+      "step": 9381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294556011929797e-06,
+      "loss": 0.898,
+      "step": 9382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292825456567942e-06,
+      "loss": 0.9089,
+      "step": 9383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2910954695396383e-06,
+      "loss": 0.8029,
+      "step": 9384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2893660509724535e-06,
+      "loss": 0.8675,
+      "step": 9385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287637200993923e-06,
+      "loss": 0.8267,
+      "step": 9386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2859089197315263e-06,
+      "loss": 0.8782,
+      "step": 9387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2841812073127125e-06,
+      "loss": 0.2604,
+      "step": 9388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2824540638648872e-06,
+      "loss": 0.7871,
+      "step": 9389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2807274895154053e-06,
+      "loss": 0.783,
+      "step": 9390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279001484391589e-06,
+      "loss": 0.8181,
+      "step": 9391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277276048620717e-06,
+      "loss": 0.8078,
+      "step": 9392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27555118233002e-06,
+      "loss": 0.9401,
+      "step": 9393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2738268856466984e-06,
+      "loss": 0.832,
+      "step": 9394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272103158697897e-06,
+      "loss": 0.9184,
+      "step": 9395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2703800016107235e-06,
+      "loss": 0.8002,
+      "step": 9396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26865741451225e-06,
+      "loss": 0.9263,
+      "step": 9397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2669353975295004e-06,
+      "loss": 0.8428,
+      "step": 9398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2652139507894545e-06,
+      "loss": 0.8751,
+      "step": 9399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2634930744190552e-06,
+      "loss": 0.8789,
+      "step": 9400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261772768545205e-06,
+      "loss": 0.9067,
+      "step": 9401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2600530332947547e-06,
+      "loss": 0.7269,
+      "step": 9402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2583338687945212e-06,
+      "loss": 0.8699,
+      "step": 9403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2566152751712834e-06,
+      "loss": 0.9247,
+      "step": 9404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25489725255176e-06,
+      "loss": 0.8157,
+      "step": 9405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253179801062646e-06,
+      "loss": 0.774,
+      "step": 9406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2514629208305905e-06,
+      "loss": 0.8174,
+      "step": 9407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249746611982191e-06,
+      "loss": 0.8193,
+      "step": 9408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248030874644014e-06,
+      "loss": 0.9793,
+      "step": 9409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.24631570894258e-06,
+      "loss": 0.8727,
+      "step": 9410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2446011150043635e-06,
+      "loss": 0.868,
+      "step": 9411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2428870929558012e-06,
+      "loss": 0.8685,
+      "step": 9412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.24117364292329e-06,
+      "loss": 0.8394,
+      "step": 9413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2394607650331755e-06,
+      "loss": 0.8229,
+      "step": 9414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2377484594117714e-06,
+      "loss": 0.9039,
+      "step": 9415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2360367261853442e-06,
+      "loss": 0.8085,
+      "step": 9416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234325565480114e-06,
+      "loss": 0.8637,
+      "step": 9417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232614977422266e-06,
+      "loss": 0.9116,
+      "step": 9418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2309049621379432e-06,
+      "loss": 0.8417,
+      "step": 9419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229195519753239e-06,
+      "loss": 0.9288,
+      "step": 9420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2274866503942103e-06,
+      "loss": 0.8328,
+      "step": 9421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225778354186875e-06,
+      "loss": 0.8635,
+      "step": 9422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224070631257198e-06,
+      "loss": 0.8413,
+      "step": 9423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2223634817311102e-06,
+      "loss": 0.261,
+      "step": 9424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220656905734505e-06,
+      "loss": 0.8238,
+      "step": 9425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2189509033932157e-06,
+      "loss": 0.9358,
+      "step": 9426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.21724547483305e-06,
+      "loss": 0.7968,
+      "step": 9427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215540620179768e-06,
+      "loss": 0.8755,
+      "step": 9428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2138363395590846e-06,
+      "loss": 0.8383,
+      "step": 9429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212132633096676e-06,
+      "loss": 0.2754,
+      "step": 9430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2104295009181785e-06,
+      "loss": 0.8512,
+      "step": 9431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2087269431491775e-06,
+      "loss": 0.8445,
+      "step": 9432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2070249599152214e-06,
+      "loss": 0.8969,
+      "step": 9433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2053235513418213e-06,
+      "loss": 0.8888,
+      "step": 9434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2036227175544365e-06,
+      "loss": 0.8902,
+      "step": 9435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201922458678486e-06,
+      "loss": 0.8339,
+      "step": 9436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2002227748393524e-06,
+      "loss": 0.8159,
+      "step": 9437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1985236661623668e-06,
+      "loss": 0.8733,
+      "step": 9438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196825132772826e-06,
+      "loss": 0.8372,
+      "step": 9439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1951271747959845e-06,
+      "loss": 0.842,
+      "step": 9440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193429792357045e-06,
+      "loss": 0.7585,
+      "step": 9441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1917329855811762e-06,
+      "loss": 0.976,
+      "step": 9442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1900367545935066e-06,
+      "loss": 0.8516,
+      "step": 9443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18834109951911e-06,
+      "loss": 0.8899,
+      "step": 9444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186646020483032e-06,
+      "loss": 0.8605,
+      "step": 9445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1849515176102665e-06,
+      "loss": 0.8112,
+      "step": 9446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183257591025766e-06,
+      "loss": 0.8458,
+      "step": 9447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1815642408544414e-06,
+      "loss": 0.8776,
+      "step": 9448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1798714672211675e-06,
+      "loss": 0.9929,
+      "step": 9449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1781792702507642e-06,
+      "loss": 0.8626,
+      "step": 9450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17648765006802e-06,
+      "loss": 0.8812,
+      "step": 9451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1747966067976756e-06,
+      "loss": 0.808,
+      "step": 9452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1731061405644283e-06,
+      "loss": 0.8609,
+      "step": 9453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171416251492937e-06,
+      "loss": 0.8842,
+      "step": 9454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1697269397078127e-06,
+      "loss": 0.8888,
+      "step": 9455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168038205333629e-06,
+      "loss": 0.9608,
+      "step": 9456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166350048494912e-06,
+      "loss": 0.8361,
+      "step": 9457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164662469316152e-06,
+      "loss": 0.856,
+      "step": 9458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1629754679217873e-06,
+      "loss": 0.7969,
+      "step": 9459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1612890444362224e-06,
+      "loss": 0.8759,
+      "step": 9460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159603198983816e-06,
+      "loss": 0.8073,
+      "step": 9461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1579179316888797e-06,
+      "loss": 0.8243,
+      "step": 9462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1562332426756925e-06,
+      "loss": 0.2491,
+      "step": 9463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1545491320684786e-06,
+      "loss": 0.8538,
+      "step": 9464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1528655999914296e-06,
+      "loss": 0.8129,
+      "step": 9465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1511826465686913e-06,
+      "loss": 0.8073,
+      "step": 9466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1495002719243643e-06,
+      "loss": 0.8799,
+      "step": 9467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147818476182506e-06,
+      "loss": 0.9424,
+      "step": 9468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146137259467136e-06,
+      "loss": 0.9285,
+      "step": 9469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1444566219022324e-06,
+      "loss": 0.9325,
+      "step": 9470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1427765636117194e-06,
+      "loss": 0.7713,
+      "step": 9471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141097084719491e-06,
+      "loss": 0.8098,
+      "step": 9472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1394181853493914e-06,
+      "loss": 0.8238,
+      "step": 9473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1377398656252225e-06,
+      "loss": 0.9291,
+      "step": 9474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1360621256707513e-06,
+      "loss": 0.8341,
+      "step": 9475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1343849656096903e-06,
+      "loss": 0.9089,
+      "step": 9476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132708385565714e-06,
+      "loss": 0.9512,
+      "step": 9477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131032385662457e-06,
+      "loss": 0.9296,
+      "step": 9478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1293569660235114e-06,
+      "loss": 0.9239,
+      "step": 9479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127682126772418e-06,
+      "loss": 0.9398,
+      "step": 9480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126007868032687e-06,
+      "loss": 0.9615,
+      "step": 9481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1243341899277746e-06,
+      "loss": 0.9704,
+      "step": 9482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122661092581101e-06,
+      "loss": 0.807,
+      "step": 9483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1209885761160454e-06,
+      "loss": 0.9063,
+      "step": 9484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119316640655934e-06,
+      "loss": 0.8318,
+      "step": 9485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1176452863240617e-06,
+      "loss": 0.9092,
+      "step": 9486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115974513243676e-06,
+      "loss": 0.8016,
+      "step": 9487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1143043215379787e-06,
+      "loss": 0.7737,
+      "step": 9488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112634711330129e-06,
+      "loss": 0.8807,
+      "step": 9489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110965682743251e-06,
+      "loss": 0.8947,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1092972359004148e-06,
+      "loss": 0.8119,
+      "step": 9491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1076293709246554e-06,
+      "loss": 0.9553,
+      "step": 9492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1059620879389666e-06,
+      "loss": 0.8594,
+      "step": 9493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1042953870662873e-06,
+      "loss": 0.8873,
+      "step": 9494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1026292684295256e-06,
+      "loss": 0.786,
+      "step": 9495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1009637321515452e-06,
+      "loss": 0.9385,
+      "step": 9496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099298778355161e-06,
+      "loss": 0.8193,
+      "step": 9497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097634407163147e-06,
+      "loss": 0.7497,
+      "step": 9498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0959706186982388e-06,
+      "loss": 0.8301,
+      "step": 9499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0943074130831207e-06,
+      "loss": 0.9041,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0926447904404425e-06,
+      "loss": 0.8196,
+      "step": 9501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090982750892808e-06,
+      "loss": 0.857,
+      "step": 9502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0893212945627738e-06,
+      "loss": 0.9214,
+      "step": 9503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08766042157286e-06,
+      "loss": 0.7932,
+      "step": 9504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0860001320455425e-06,
+      "loss": 0.7902,
+      "step": 9505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0843404261032474e-06,
+      "loss": 0.9062,
+      "step": 9506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082681303868368e-06,
+      "loss": 0.8591,
+      "step": 9507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0810227654632465e-06,
+      "loss": 0.8459,
+      "step": 9508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079364811010183e-06,
+      "loss": 0.83,
+      "step": 9509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077707440631439e-06,
+      "loss": 0.8906,
+      "step": 9510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0760506544492322e-06,
+      "loss": 0.8969,
+      "step": 9511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0743944525857307e-06,
+      "loss": 0.7893,
+      "step": 9512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0727388351630674e-06,
+      "loss": 0.9033,
+      "step": 9513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071083802303331e-06,
+      "loss": 0.8542,
+      "step": 9514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0694293541285593e-06,
+      "loss": 0.8729,
+      "step": 9515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067775490760756e-06,
+      "loss": 0.7906,
+      "step": 9516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0661222123218807e-06,
+      "loss": 0.959,
+      "step": 9517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064469518933845e-06,
+      "loss": 0.8484,
+      "step": 9518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.062817410718517e-06,
+      "loss": 0.7778,
+      "step": 9519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0611658877977313e-06,
+      "loss": 0.8083,
+      "step": 9520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0595149502932654e-06,
+      "loss": 0.8701,
+      "step": 9521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057864598326864e-06,
+      "loss": 0.2728,
+      "step": 9522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056214832020228e-06,
+      "loss": 0.8639,
+      "step": 9523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0545656514950084e-06,
+      "loss": 0.7873,
+      "step": 9524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0529170568728186e-06,
+      "loss": 0.2343,
+      "step": 9525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0512690482752308e-06,
+      "loss": 0.8832,
+      "step": 9526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0496216258237635e-06,
+      "loss": 0.6729,
+      "step": 9527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047974789639907e-06,
+      "loss": 0.8797,
+      "step": 9528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046328539845096e-06,
+      "loss": 0.8053,
+      "step": 9529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044682876560724e-06,
+      "loss": 0.8517,
+      "step": 9530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043037799908146e-06,
+      "loss": 0.8678,
+      "step": 9531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0413933100086746e-06,
+      "loss": 0.8655,
+      "step": 9532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03974940698357e-06,
+      "loss": 0.8266,
+      "step": 9533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0381060909540584e-06,
+      "loss": 0.809,
+      "step": 9534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036463362041321e-06,
+      "loss": 0.8307,
+      "step": 9535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03482122036649e-06,
+      "loss": 0.8217,
+      "step": 9536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0331796660506598e-06,
+      "loss": 0.8406,
+      "step": 9537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0315386992148857e-06,
+      "loss": 0.1875,
+      "step": 9538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0298983199801626e-06,
+      "loss": 0.2715,
+      "step": 9539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0282585284674606e-06,
+      "loss": 0.8477,
+      "step": 9540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0266193247976996e-06,
+      "loss": 0.8137,
+      "step": 9541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0249807090917526e-06,
+      "loss": 0.8762,
+      "step": 9542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0233426814704547e-06,
+      "loss": 0.8393,
+      "step": 9543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021705242054598e-06,
+      "loss": 0.9061,
+      "step": 9544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020068390964922e-06,
+      "loss": 0.8816,
+      "step": 9545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0184321283221344e-06,
+      "loss": 0.7916,
+      "step": 9546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0167964542468952e-06,
+      "loss": 0.9318,
+      "step": 9547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0151613688598158e-06,
+      "loss": 0.9155,
+      "step": 9548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013526872281475e-06,
+      "loss": 0.8839,
+      "step": 9549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0118929646323982e-06,
+      "loss": 0.9136,
+      "step": 9550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0102596460330694e-06,
+      "loss": 0.8694,
+      "step": 9551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008626916603932e-06,
+      "loss": 0.815,
+      "step": 9552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0069947764653884e-06,
+      "loss": 0.8309,
+      "step": 9553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0053632257377896e-06,
+      "loss": 0.7958,
+      "step": 9554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003732264541449e-06,
+      "loss": 0.7503,
+      "step": 9555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0021018929966384e-06,
+      "loss": 0.792,
+      "step": 9556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0004721112235768e-06,
+      "loss": 0.8406,
+      "step": 9557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998842919342451e-06,
+      "loss": 0.8558,
+      "step": 9558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9972143174733973e-06,
+      "loss": 0.844,
+      "step": 9559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9955863057365068e-06,
+      "loss": 0.9132,
+      "step": 9560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993958884251832e-06,
+      "loss": 0.7933,
+      "step": 9561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9923320531393853e-06,
+      "loss": 0.7982,
+      "step": 9562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990705812519124e-06,
+      "loss": 0.8426,
+      "step": 9563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9890801625109713e-06,
+      "loss": 0.7798,
+      "step": 9564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9874551032348068e-06,
+      "loss": 0.9118,
+      "step": 9565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9858306348104573e-06,
+      "loss": 0.8267,
+      "step": 9566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9842067573577196e-06,
+      "loss": 0.2429,
+      "step": 9567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9825834709963334e-06,
+      "loss": 0.7762,
+      "step": 9568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980960775846006e-06,
+      "loss": 0.9138,
+      "step": 9569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979338672026393e-06,
+      "loss": 0.8824,
+      "step": 9570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977717159657112e-06,
+      "loss": 0.8612,
+      "step": 9571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9760962388577332e-06,
+      "loss": 0.786,
+      "step": 9572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974475909747784e-06,
+      "loss": 0.8869,
+      "step": 9573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972856172446753e-06,
+      "loss": 0.8708,
+      "step": 9574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9712370270740756e-06,
+      "loss": 0.8671,
+      "step": 9575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969618473749153e-06,
+      "loss": 0.8414,
+      "step": 9576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968000512591336e-06,
+      "loss": 0.9052,
+      "step": 9577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9663831437199343e-06,
+      "loss": 0.251,
+      "step": 9578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964766367254218e-06,
+      "loss": 0.8315,
+      "step": 9579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963150183313408e-06,
+      "loss": 0.2644,
+      "step": 9580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9615345920166784e-06,
+      "loss": 0.8107,
+      "step": 9581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9599195934831682e-06,
+      "loss": 0.9378,
+      "step": 9582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9583051878319704e-06,
+      "loss": 0.9197,
+      "step": 9583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956691375182128e-06,
+      "loss": 0.8744,
+      "step": 9584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95507815565265e-06,
+      "loss": 0.8238,
+      "step": 9585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9534655293624914e-06,
+      "loss": 0.8343,
+      "step": 9586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951853496430571e-06,
+      "loss": 0.7915,
+      "step": 9587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950242056975765e-06,
+      "loss": 0.8996,
+      "step": 9588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948631211116897e-06,
+      "loss": 0.8174,
+      "step": 9589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9470209589727552e-06,
+      "loss": 0.8389,
+      "step": 9590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9454113006620777e-06,
+      "loss": 0.9058,
+      "step": 9591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943802236303568e-06,
+      "loss": 0.8084,
+      "step": 9592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942193766015872e-06,
+      "loss": 0.8432,
+      "step": 9593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940585889917608e-06,
+      "loss": 0.8572,
+      "step": 9594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938978608127333e-06,
+      "loss": 0.8293,
+      "step": 9595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9373719207635766e-06,
+      "loss": 0.9201,
+      "step": 9596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9357658279448165e-06,
+      "loss": 0.7995,
+      "step": 9597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934160329789483e-06,
+      "loss": 0.9013,
+      "step": 9598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9325554264159696e-06,
+      "loss": 0.8792,
+      "step": 9599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9309511179426264e-06,
+      "loss": 0.8249,
+      "step": 9600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9293474044877524e-06,
+      "loss": 0.8845,
+      "step": 9601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927744286169606e-06,
+      "loss": 0.2838,
+      "step": 9602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926141763106406e-06,
+      "loss": 0.851,
+      "step": 9603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9245398354163202e-06,
+      "loss": 0.885,
+      "step": 9604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9229385032174773e-06,
+      "loss": 0.877,
+      "step": 9605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9213377666279654e-06,
+      "loss": 0.8447,
+      "step": 9606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919737625765816e-06,
+      "loss": 0.9096,
+      "step": 9607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9181380807490304e-06,
+      "loss": 0.8596,
+      "step": 9608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9165391316955605e-06,
+      "loss": 0.848,
+      "step": 9609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914940778723311e-06,
+      "loss": 0.9414,
+      "step": 9610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9133430219501504e-06,
+      "loss": 0.8517,
+      "step": 9611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9117458614938944e-06,
+      "loss": 0.9021,
+      "step": 9612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910149297472319e-06,
+      "loss": 0.9016,
+      "step": 9613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908553330003158e-06,
+      "loss": 0.8536,
+      "step": 9614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9069579592041e-06,
+      "loss": 0.219,
+      "step": 9615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9053631851927857e-06,
+      "loss": 0.9155,
+      "step": 9616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9037690080868176e-06,
+      "loss": 0.8467,
+      "step": 9617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902175428003754e-06,
+      "loss": 0.8885,
+      "step": 9618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900582445061101e-06,
+      "loss": 0.8252,
+      "step": 9619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.89899005937633e-06,
+      "loss": 0.8308,
+      "step": 9620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8973982710668703e-06,
+      "loss": 0.9058,
+      "step": 9621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8958070802500905e-06,
+      "loss": 0.8324,
+      "step": 9622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894216487043331e-06,
+      "loss": 0.8481,
+      "step": 9623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8926264915638882e-06,
+      "loss": 0.7603,
+      "step": 9624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8910370939290024e-06,
+      "loss": 0.8124,
+      "step": 9625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.88944829425588e-06,
+      "loss": 0.8083,
+      "step": 9626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8878600926616842e-06,
+      "loss": 0.8791,
+      "step": 9627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8862724892635243e-06,
+      "loss": 0.8093,
+      "step": 9628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8846854841784745e-06,
+      "loss": 0.9446,
+      "step": 9629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8830990775235648e-06,
+      "loss": 0.8321,
+      "step": 9630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8815132694157745e-06,
+      "loss": 0.8209,
+      "step": 9631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8799280599720404e-06,
+      "loss": 0.9545,
+      "step": 9632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8783434493092624e-06,
+      "loss": 0.7921,
+      "step": 9633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8767594375442855e-06,
+      "loss": 0.9299,
+      "step": 9634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8751760247939188e-06,
+      "loss": 0.936,
+      "step": 9635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8735932111749277e-06,
+      "loss": 0.8261,
+      "step": 9636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8720109968040245e-06,
+      "loss": 0.8578,
+      "step": 9637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8704293817978859e-06,
+      "loss": 0.8767,
+      "step": 9638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8688483662731426e-06,
+      "loss": 0.9076,
+      "step": 9639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8672679503463775e-06,
+      "loss": 0.9029,
+      "step": 9640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8656881341341327e-06,
+      "loss": 0.856,
+      "step": 9641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8641089177529104e-06,
+      "loss": 0.9041,
+      "step": 9642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8625303013191532e-06,
+      "loss": 0.8149,
+      "step": 9643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8609522849492755e-06,
+      "loss": 0.9293,
+      "step": 9644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8593748687596435e-06,
+      "loss": 0.8644,
+      "step": 9645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8577980528665718e-06,
+      "loss": 0.7476,
+      "step": 9646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8562218373863383e-06,
+      "loss": 0.8782,
+      "step": 9647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854646222435178e-06,
+      "loss": 0.8666,
+      "step": 9648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8530712081292734e-06,
+      "loss": 0.8739,
+      "step": 9649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8514967945847685e-06,
+      "loss": 0.8273,
+      "step": 9650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8499229819177643e-06,
+      "loss": 0.8957,
+      "step": 9651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8483497702443142e-06,
+      "loss": 0.7846,
+      "step": 9652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846777159680425e-06,
+      "loss": 0.8498,
+      "step": 9653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8452051503420677e-06,
+      "loss": 0.8707,
+      "step": 9654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8436337423451566e-06,
+      "loss": 0.8835,
+      "step": 9655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.842062935805573e-06,
+      "loss": 0.8855,
+      "step": 9656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8404927308391518e-06,
+      "loss": 0.8927,
+      "step": 9657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8389231275616758e-06,
+      "loss": 0.8347,
+      "step": 9658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8373541260888917e-06,
+      "loss": 0.8963,
+      "step": 9659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8357857265365009e-06,
+      "loss": 0.8573,
+      "step": 9660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342179290201546e-06,
+      "loss": 0.9257,
+      "step": 9661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8326507336554677e-06,
+      "loss": 0.8056,
+      "step": 9662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8310841405580037e-06,
+      "loss": 0.8447,
+      "step": 9663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8295181498432824e-06,
+      "loss": 0.7439,
+      "step": 9664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8279527616267844e-06,
+      "loss": 0.8188,
+      "step": 9665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8263879760239446e-06,
+      "loss": 0.946,
+      "step": 9666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8248237931501466e-06,
+      "loss": 0.8439,
+      "step": 9667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8232602131207377e-06,
+      "loss": 0.8405,
+      "step": 9668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8216972360510187e-06,
+      "loss": 0.911,
+      "step": 9669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8201348620562409e-06,
+      "loss": 0.8986,
+      "step": 9670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8185730912516198e-06,
+      "loss": 0.9111,
+      "step": 9671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8170119237523176e-06,
+      "loss": 0.8451,
+      "step": 9672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8154513596734603e-06,
+      "loss": 0.8183,
+      "step": 9673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8138913991301188e-06,
+      "loss": 0.7868,
+      "step": 9674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8123320422373335e-06,
+      "loss": 0.8086,
+      "step": 9675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810773289110086e-06,
+      "loss": 0.8104,
+      "step": 9676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8092151398633239e-06,
+      "loss": 0.9225,
+      "step": 9677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8076575946119478e-06,
+      "loss": 0.8177,
+      "step": 9678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806100653470807e-06,
+      "loss": 0.2585,
+      "step": 9679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8045443165547184e-06,
+      "loss": 0.8475,
+      "step": 9680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8029885839784411e-06,
+      "loss": 0.2845,
+      "step": 9681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8014334558566992e-06,
+      "loss": 0.8904,
+      "step": 9682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799878932304171e-06,
+      "loss": 0.833,
+      "step": 9683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7983250134354867e-06,
+      "loss": 0.9281,
+      "step": 9684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7967716993652317e-06,
+      "loss": 0.2968,
+      "step": 9685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952189902079498e-06,
+      "loss": 0.865,
+      "step": 9686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936668860781414e-06,
+      "loss": 0.2556,
+      "step": 9687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7921153870902564e-06,
+      "loss": 0.7537,
+      "step": 9688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905644933587073e-06,
+      "loss": 0.8792,
+      "step": 9689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7890142049978544e-06,
+      "loss": 0.8705,
+      "step": 9690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7874645221220198e-06,
+      "loss": 0.826,
+      "step": 9691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7859154448454795e-06,
+      "loss": 0.7276,
+      "step": 9692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7843669732824632e-06,
+      "loss": 0.8575,
+      "step": 9693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7828191075471524e-06,
+      "loss": 0.8853,
+      "step": 9694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7812718477536906e-06,
+      "loss": 0.9009,
+      "step": 9695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7797251940161775e-06,
+      "loss": 0.7725,
+      "step": 9696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7781791464486597e-06,
+      "loss": 0.9432,
+      "step": 9697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7766337051651471e-06,
+      "loss": 0.8128,
+      "step": 9698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7750888702795999e-06,
+      "loss": 0.8071,
+      "step": 9699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735446419059355e-06,
+      "loss": 0.2476,
+      "step": 9700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7720010201580308e-06,
+      "loss": 0.8543,
+      "step": 9701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7704580051497078e-06,
+      "loss": 0.88,
+      "step": 9702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7689155969947546e-06,
+      "loss": 0.854,
+      "step": 9703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767373795806905e-06,
+      "loss": 0.9099,
+      "step": 9704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7658326016998583e-06,
+      "loss": 0.9719,
+      "step": 9705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7642920147872566e-06,
+      "loss": 0.8814,
+      "step": 9706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762752035182711e-06,
+      "loss": 0.7949,
+      "step": 9707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761212662999776e-06,
+      "loss": 0.8468,
+      "step": 9708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7596738983519668e-06,
+      "loss": 0.8456,
+      "step": 9709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7581357413527578e-06,
+      "loss": 0.2742,
+      "step": 9710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7565981921155684e-06,
+      "loss": 0.8603,
+      "step": 9711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75506125075378e-06,
+      "loss": 0.7996,
+      "step": 9712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7535249173807312e-06,
+      "loss": 0.8641,
+      "step": 9713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751989192109711e-06,
+      "loss": 0.9247,
+      "step": 9714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.750454075053961e-06,
+      "loss": 0.9461,
+      "step": 9715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7489195663266878e-06,
+      "loss": 0.8022,
+      "step": 9716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7473856660410437e-06,
+      "loss": 0.8622,
+      "step": 9717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7458523743101398e-06,
+      "loss": 0.8875,
+      "step": 9718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7443196912470462e-06,
+      "loss": 0.8955,
+      "step": 9719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7427876169647795e-06,
+      "loss": 0.7974,
+      "step": 9720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741256151576317e-06,
+      "loss": 0.8111,
+      "step": 9721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7397252951945952e-06,
+      "loss": 0.8467,
+      "step": 9722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7381950479324938e-06,
+      "loss": 0.8766,
+      "step": 9723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.73666540990286e-06,
+      "loss": 0.8296,
+      "step": 9724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7351363812184897e-06,
+      "loss": 0.8564,
+      "step": 9725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7336079619921297e-06,
+      "loss": 0.9,
+      "step": 9726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7320801523364905e-06,
+      "loss": 0.8547,
+      "step": 9727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7305529523642373e-06,
+      "loss": 0.8629,
+      "step": 9728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729026362187981e-06,
+      "loss": 0.2647,
+      "step": 9729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7275003819202962e-06,
+      "loss": 0.8856,
+      "step": 9730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7259750116737139e-06,
+      "loss": 0.8387,
+      "step": 9731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7244502515607098e-06,
+      "loss": 0.8369,
+      "step": 9732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7229261016937238e-06,
+      "loss": 0.8025,
+      "step": 9733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7214025621851526e-06,
+      "loss": 0.8717,
+      "step": 9734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7198796331473344e-06,
+      "loss": 0.8582,
+      "step": 9735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7183573146925757e-06,
+      "loss": 0.8794,
+      "step": 9736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716835606933136e-06,
+      "loss": 0.8235,
+      "step": 9737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.715314509981222e-06,
+      "loss": 0.8645,
+      "step": 9738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137940239490037e-06,
+      "loss": 0.8159,
+      "step": 9739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.712274148948605e-06,
+      "loss": 0.7753,
+      "step": 9740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7107548850920974e-06,
+      "loss": 0.8642,
+      "step": 9741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709236232491517e-06,
+      "loss": 0.9361,
+      "step": 9742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077181912588514e-06,
+      "loss": 0.8619,
+      "step": 9743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062007615060382e-06,
+      "loss": 0.9227,
+      "step": 9744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704683943344978e-06,
+      "loss": 0.7957,
+      "step": 9745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70316773688752e-06,
+      "loss": 0.8899,
+      "step": 9746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701652142245469e-06,
+      "loss": 0.7464,
+      "step": 9747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7001371595305872e-06,
+      "loss": 0.8978,
+      "step": 9748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986227888545948e-06,
+      "loss": 0.9063,
+      "step": 9749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697109030329156e-06,
+      "loss": 0.8168,
+      "step": 9750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6955958840659015e-06,
+      "loss": 0.8586,
+      "step": 9751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694083350176412e-06,
+      "loss": 0.9067,
+      "step": 9752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6925714287722194e-06,
+      "loss": 0.8233,
+      "step": 9753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6910601199648157e-06,
+      "loss": 0.8781,
+      "step": 9754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6895494238656517e-06,
+      "loss": 0.8699,
+      "step": 9755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6880393405861184e-06,
+      "loss": 0.7982,
+      "step": 9756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6865298702375743e-06,
+      "loss": 0.8105,
+      "step": 9757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6850210129313315e-06,
+      "loss": 0.9421,
+      "step": 9758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683512768778649e-06,
+      "loss": 0.8218,
+      "step": 9759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6820051378907499e-06,
+      "loss": 0.8607,
+      "step": 9760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6804981203788107e-06,
+      "loss": 0.8205,
+      "step": 9761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789917163539526e-06,
+      "loss": 0.8229,
+      "step": 9762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6774859259272647e-06,
+      "loss": 0.8554,
+      "step": 9763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6759807492097868e-06,
+      "loss": 0.847,
+      "step": 9764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6744761863125082e-06,
+      "loss": 0.8823,
+      "step": 9765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6729722373463752e-06,
+      "loss": 0.8068,
+      "step": 9766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714689024222951e-06,
+      "loss": 0.8707,
+      "step": 9767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6699661816511203e-06,
+      "loss": 0.8539,
+      "step": 9768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6684640751436642e-06,
+      "loss": 0.8259,
+      "step": 9769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6669625830106972e-06,
+      "loss": 0.8098,
+      "step": 9770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665461705362935e-06,
+      "loss": 0.8767,
+      "step": 9771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.663961442311056e-06,
+      "loss": 0.7782,
+      "step": 9772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662461793965694e-06,
+      "loss": 0.7929,
+      "step": 9773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6609627604374278e-06,
+      "loss": 0.9015,
+      "step": 9774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6594643418368051e-06,
+      "loss": 0.8437,
+      "step": 9775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6579665382743148e-06,
+      "loss": 0.8306,
+      "step": 9776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6564693498604057e-06,
+      "loss": 0.7926,
+      "step": 9777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6549727767054835e-06,
+      "loss": 0.8478,
+      "step": 9778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6534768189199103e-06,
+      "loss": 0.8587,
+      "step": 9779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651981476613993e-06,
+      "loss": 0.8922,
+      "step": 9780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6504867498980038e-06,
+      "loss": 0.8906,
+      "step": 9781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6489926388821643e-06,
+      "loss": 0.8787,
+      "step": 9782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6474991436766496e-06,
+      "loss": 0.8007,
+      "step": 9783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6460062643915952e-06,
+      "loss": 0.867,
+      "step": 9784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6445140011370819e-06,
+      "loss": 0.9169,
+      "step": 9785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643022354023156e-06,
+      "loss": 0.7804,
+      "step": 9786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6415313231598073e-06,
+      "loss": 0.8552,
+      "step": 9787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6400409086569913e-06,
+      "loss": 0.7779,
+      "step": 9788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6385511106246078e-06,
+      "loss": 0.8353,
+      "step": 9789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6370619291725165e-06,
+      "loss": 0.8341,
+      "step": 9790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6355733644105355e-06,
+      "loss": 0.9126,
+      "step": 9791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6340854164484254e-06,
+      "loss": 0.8984,
+      "step": 9792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6325980853959157e-06,
+      "loss": 0.7462,
+      "step": 9793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6311113713626781e-06,
+      "loss": 0.8824,
+      "step": 9794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6296252744583463e-06,
+      "loss": 0.7773,
+      "step": 9795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6281397947925093e-06,
+      "loss": 0.8716,
+      "step": 9796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626654932474705e-06,
+      "loss": 0.7651,
+      "step": 9797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6251706876144246e-06,
+      "loss": 0.7951,
+      "step": 9798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623687060321122e-06,
+      "loss": 0.8933,
+      "step": 9799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6222040507042024e-06,
+      "loss": 0.8459,
+      "step": 9800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6207216588730201e-06,
+      "loss": 0.9098,
+      "step": 9801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6192398849368919e-06,
+      "loss": 0.7743,
+      "step": 9802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617758729005081e-06,
+      "loss": 0.7775,
+      "step": 9803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162781911868109e-06,
+      "loss": 0.9452,
+      "step": 9804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6147982715912603e-06,
+      "loss": 0.8132,
+      "step": 9805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6133189703275553e-06,
+      "loss": 0.8333,
+      "step": 9806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6118402875047845e-06,
+      "loss": 0.7722,
+      "step": 9807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6103622232319838e-06,
+      "loss": 0.8866,
+      "step": 9808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6088847776181516e-06,
+      "loss": 0.7836,
+      "step": 9809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6074079507722294e-06,
+      "loss": 0.8751,
+      "step": 9810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6059317428031273e-06,
+      "loss": 0.8652,
+      "step": 9811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6044561538196945e-06,
+      "loss": 0.8213,
+      "step": 9812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6029811839307462e-06,
+      "loss": 0.868,
+      "step": 9813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6015068332450502e-06,
+      "loss": 0.8327,
+      "step": 9814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600033101871321e-06,
+      "loss": 0.7797,
+      "step": 9815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598559989918237e-06,
+      "loss": 0.2199,
+      "step": 9816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5970874974944262e-06,
+      "loss": 0.8179,
+      "step": 9817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5956156247084708e-06,
+      "loss": 0.8546,
+      "step": 9818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5941443716689043e-06,
+      "loss": 0.8092,
+      "step": 9819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592673738484225e-06,
+      "loss": 0.7823,
+      "step": 9820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.591203725262872e-06,
+      "loss": 0.8391,
+      "step": 9821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5897343321132496e-06,
+      "loss": 0.8949,
+      "step": 9822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5882655591437125e-06,
+      "loss": 0.8411,
+      "step": 9823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5867974064625646e-06,
+      "loss": 0.8345,
+      "step": 9824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5853298741780733e-06,
+      "loss": 0.7782,
+      "step": 9825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5838629623984558e-06,
+      "loss": 0.8962,
+      "step": 9826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5823966712318816e-06,
+      "loss": 0.8373,
+      "step": 9827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5809310007864743e-06,
+      "loss": 0.8106,
+      "step": 9828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5794659511703147e-06,
+      "loss": 0.9189,
+      "step": 9829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5780015224914413e-06,
+      "loss": 0.8324,
+      "step": 9830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5765377148578365e-06,
+      "loss": 0.8701,
+      "step": 9831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575074528377447e-06,
+      "loss": 0.9248,
+      "step": 9832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573611963158166e-06,
+      "loss": 0.9167,
+      "step": 9833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5721500193078455e-06,
+      "loss": 0.7787,
+      "step": 9834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706886969342927e-06,
+      "loss": 0.7991,
+      "step": 9835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5692279961452628e-06,
+      "loss": 0.791,
+      "step": 9836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5677679170484704e-06,
+      "loss": 0.9275,
+      "step": 9837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5663084597515865e-06,
+      "loss": 0.8996,
+      "step": 9838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5648496243622303e-06,
+      "loss": 0.8773,
+      "step": 9839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5633914109879745e-06,
+      "loss": 0.933,
+      "step": 9840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5619338197363533e-06,
+      "loss": 0.9327,
+      "step": 9841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5604768507148471e-06,
+      "loss": 0.9203,
+      "step": 9842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.559020504030897e-06,
+      "loss": 0.9067,
+      "step": 9843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5575647797918947e-06,
+      "loss": 0.8624,
+      "step": 9844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5561096781051854e-06,
+      "loss": 0.8581,
+      "step": 9845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554655199078069e-06,
+      "loss": 0.9197,
+      "step": 9846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.553201342817804e-06,
+      "loss": 0.857,
+      "step": 9847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5517481094315955e-06,
+      "loss": 0.8898,
+      "step": 9848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5502954990266062e-06,
+      "loss": 0.8545,
+      "step": 9849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5488435117099543e-06,
+      "loss": 0.2414,
+      "step": 9850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5473921475887078e-06,
+      "loss": 0.8502,
+      "step": 9851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.545941406769893e-06,
+      "loss": 0.7932,
+      "step": 9852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5444912893604913e-06,
+      "loss": 0.8362,
+      "step": 9853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5430417954674315e-06,
+      "loss": 0.8286,
+      "step": 9854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5415929251976025e-06,
+      "loss": 0.8483,
+      "step": 9855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5401446786578479e-06,
+      "loss": 0.8191,
+      "step": 9856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5386970559549564e-06,
+      "loss": 0.8163,
+      "step": 9857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.537250057195684e-06,
+      "loss": 0.8363,
+      "step": 9858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5358036824867295e-06,
+      "loss": 0.8505,
+      "step": 9859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534357931934748e-06,
+      "loss": 0.2529,
+      "step": 9860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5329128056463515e-06,
+      "loss": 0.871,
+      "step": 9861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5314683037281097e-06,
+      "loss": 0.8676,
+      "step": 9862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530024426286535e-06,
+      "loss": 0.2692,
+      "step": 9863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528581173428103e-06,
+      "loss": 0.8675,
+      "step": 9864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5271385452592424e-06,
+      "loss": 0.8768,
+      "step": 9865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.525696541886329e-06,
+      "loss": 0.8453,
+      "step": 9866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5242551634156999e-06,
+      "loss": 0.7928,
+      "step": 9867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5228144099536468e-06,
+      "loss": 0.7403,
+      "step": 9868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213742816064082e-06,
+      "loss": 0.2614,
+      "step": 9869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5199347784801787e-06,
+      "loss": 0.9008,
+      "step": 9870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518495900681114e-06,
+      "loss": 0.8061,
+      "step": 9871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5170576483153122e-06,
+      "loss": 0.837,
+      "step": 9872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.515620021488834e-06,
+      "loss": 0.8597,
+      "step": 9873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5141830203076935e-06,
+      "loss": 0.861,
+      "step": 9874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5127466448778515e-06,
+      "loss": 0.8192,
+      "step": 9875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5113108953052303e-06,
+      "loss": 0.829,
+      "step": 9876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5098757716957057e-06,
+      "loss": 0.8557,
+      "step": 9877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5084412741551002e-06,
+      "loss": 0.7515,
+      "step": 9878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5070074027892002e-06,
+      "loss": 0.8437,
+      "step": 9879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5055741577037364e-06,
+      "loss": 0.8259,
+      "step": 9880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5041415390043967e-06,
+      "loss": 0.8197,
+      "step": 9881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5027095467968244e-06,
+      "loss": 0.8339,
+      "step": 9882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5012781811866207e-06,
+      "loss": 0.8758,
+      "step": 9883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4998474422793286e-06,
+      "loss": 0.795,
+      "step": 9884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498417330180455e-06,
+      "loss": 0.7823,
+      "step": 9885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.496987844995461e-06,
+      "loss": 0.9724,
+      "step": 9886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4955589868297526e-06,
+      "loss": 0.8374,
+      "step": 9887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4941307557886964e-06,
+      "loss": 0.2678,
+      "step": 9888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4927031519776181e-06,
+      "loss": 0.8906,
+      "step": 9889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491276175501779e-06,
+      "loss": 0.8273,
+      "step": 9890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4898498264664107e-06,
+      "loss": 0.8909,
+      "step": 9891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4884241049766967e-06,
+      "loss": 0.7073,
+      "step": 9892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4869990111377663e-06,
+      "loss": 0.7086,
+      "step": 9893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855745450547087e-06,
+      "loss": 0.7589,
+      "step": 9894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4841507068325667e-06,
+      "loss": 0.9332,
+      "step": 9895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4827274965763327e-06,
+      "loss": 0.7684,
+      "step": 9896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4813049143909563e-06,
+      "loss": 0.7858,
+      "step": 9897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4798829603813425e-06,
+      "loss": 0.8725,
+      "step": 9898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478461634652343e-06,
+      "loss": 0.911,
+      "step": 9899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4770409373087712e-06,
+      "loss": 0.8583,
+      "step": 9900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4756208684553907e-06,
+      "loss": 0.8826,
+      "step": 9901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474201428196913e-06,
+      "loss": 0.9162,
+      "step": 9902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727826166380144e-06,
+      "loss": 0.2751,
+      "step": 9903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4713644338833188e-06,
+      "loss": 0.933,
+      "step": 9904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4699468800374017e-06,
+      "loss": 0.9626,
+      "step": 9905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468529955204795e-06,
+      "loss": 0.8305,
+      "step": 9906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4671136594899882e-06,
+      "loss": 0.7963,
+      "step": 9907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4656979929974136e-06,
+      "loss": 0.8508,
+      "step": 9908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4642829558314698e-06,
+      "loss": 0.9063,
+      "step": 9909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4628685480965e-06,
+      "loss": 0.8625,
+      "step": 9910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4614547698968019e-06,
+      "loss": 0.8436,
+      "step": 9911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4600416213366298e-06,
+      "loss": 0.8565,
+      "step": 9912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.458629102520195e-06,
+      "loss": 0.8023,
+      "step": 9913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4572172135516504e-06,
+      "loss": 0.9224,
+      "step": 9914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4558059545351144e-06,
+      "loss": 0.9083,
+      "step": 9915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4543953255746546e-06,
+      "loss": 0.8235,
+      "step": 9916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45298532677429e-06,
+      "loss": 0.7976,
+      "step": 9917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4515759582379985e-06,
+      "loss": 0.8123,
+      "step": 9918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501672200697025e-06,
+      "loss": 0.9287,
+      "step": 9919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4487591123732903e-06,
+      "loss": 0.8896,
+      "step": 9920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4473516352525896e-06,
+      "loss": 0.8588,
+      "step": 9921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4459447888113964e-06,
+      "loss": 0.8829,
+      "step": 9922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4445385731534457e-06,
+      "loss": 0.8392,
+      "step": 9923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4431329883824375e-06,
+      "loss": 0.8055,
+      "step": 9924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4417280346020212e-06,
+      "loss": 0.8327,
+      "step": 9925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4403237119157954e-06,
+      "loss": 0.9155,
+      "step": 9926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4389200204273212e-06,
+      "loss": 0.8677,
+      "step": 9927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4375169602401028e-06,
+      "loss": 0.8862,
+      "step": 9928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4361145314576064e-06,
+      "loss": 0.8265,
+      "step": 9929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347127341832478e-06,
+      "loss": 0.8387,
+      "step": 9930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.433311568520398e-06,
+      "loss": 1.0022,
+      "step": 9931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4319110345723753e-06,
+      "loss": 0.8746,
+      "step": 9932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4305111324424591e-06,
+      "loss": 0.8318,
+      "step": 9933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4291118622338828e-06,
+      "loss": 0.8932,
+      "step": 9934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4277132240498238e-06,
+      "loss": 0.8745,
+      "step": 9935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4263152179934236e-06,
+      "loss": 0.8297,
+      "step": 9936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4249178441677691e-06,
+      "loss": 0.859,
+      "step": 9937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4235211026759045e-06,
+      "loss": 0.9604,
+      "step": 9938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221249936208292e-06,
+      "loss": 0.9158,
+      "step": 9939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4207295171054902e-06,
+      "loss": 0.7968,
+      "step": 9940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4193346732327951e-06,
+      "loss": 0.8584,
+      "step": 9941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4179404621055948e-06,
+      "loss": 0.9569,
+      "step": 9942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416546883826706e-06,
+      "loss": 0.8868,
+      "step": 9943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4151539384988865e-06,
+      "loss": 0.8539,
+      "step": 9944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4137616262248587e-06,
+      "loss": 0.8659,
+      "step": 9945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4123699471072883e-06,
+      "loss": 0.8441,
+      "step": 9946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4109789012488017e-06,
+      "loss": 0.8763,
+      "step": 9947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4095884887519761e-06,
+      "loss": 0.8686,
+      "step": 9948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4081987097193383e-06,
+      "loss": 0.7969,
+      "step": 9949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4068095642533753e-06,
+      "loss": 0.8057,
+      "step": 9950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4054210524565237e-06,
+      "loss": 0.7443,
+      "step": 9951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404033174431173e-06,
+      "loss": 0.8517,
+      "step": 9952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4026459302796647e-06,
+      "loss": 0.954,
+      "step": 9953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.401259320104298e-06,
+      "loss": 0.8547,
+      "step": 9954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3998733440073186e-06,
+      "loss": 0.8653,
+      "step": 9955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3984880020909331e-06,
+      "loss": 0.8995,
+      "step": 9956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3971032944572983e-06,
+      "loss": 0.2798,
+      "step": 9957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3957192212085213e-06,
+      "loss": 0.8681,
+      "step": 9958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3943357824466652e-06,
+      "loss": 0.7898,
+      "step": 9959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3929529782737493e-06,
+      "loss": 0.891,
+      "step": 9960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.391570808791739e-06,
+      "loss": 0.8243,
+      "step": 9961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390189274102557e-06,
+      "loss": 0.8478,
+      "step": 9962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3888083743080816e-06,
+      "loss": 0.8367,
+      "step": 9963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.387428109510136e-06,
+      "loss": 0.2226,
+      "step": 9964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3860484798105067e-06,
+      "loss": 0.7955,
+      "step": 9965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.38466948531093e-06,
+      "loss": 0.7852,
+      "step": 9966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3832911261130888e-06,
+      "loss": 0.9106,
+      "step": 9967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3819134023186275e-06,
+      "loss": 0.9796,
+      "step": 9968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805363140291438e-06,
+      "loss": 0.7937,
+      "step": 9969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3791598613461788e-06,
+      "loss": 0.7726,
+      "step": 9970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3777840443712364e-06,
+      "loss": 0.7791,
+      "step": 9971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376408863205776e-06,
+      "loss": 0.2354,
+      "step": 9972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.375034317951195e-06,
+      "loss": 0.9132,
+      "step": 9973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3736604087088567e-06,
+      "loss": 0.7924,
+      "step": 9974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3722871355800782e-06,
+      "loss": 0.8895,
+      "step": 9975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3709144986661217e-06,
+      "loss": 0.8806,
+      "step": 9976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.369542498068207e-06,
+      "loss": 0.908,
+      "step": 9977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3681711338875104e-06,
+      "loss": 0.8576,
+      "step": 9978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3668004062251517e-06,
+      "loss": 0.9583,
+      "step": 9979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3654303151822123e-06,
+      "loss": 0.8372,
+      "step": 9980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3640608608597273e-06,
+      "loss": 0.7914,
+      "step": 9981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3626920433586765e-06,
+      "loss": 0.8452,
+      "step": 9982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3613238627799985e-06,
+      "loss": 0.8804,
+      "step": 9983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359956319224587e-06,
+      "loss": 0.7479,
+      "step": 9984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3585894127932808e-06,
+      "loss": 0.7773,
+      "step": 9985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3572231435868798e-06,
+      "loss": 0.9065,
+      "step": 9986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3558575117061357e-06,
+      "loss": 0.8851,
+      "step": 9987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3544925172517475e-06,
+      "loss": 0.836,
+      "step": 9988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3531281603243728e-06,
+      "loss": 0.7603,
+      "step": 9989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3517644410246222e-06,
+      "loss": 0.812,
+      "step": 9990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350401359453054e-06,
+      "loss": 0.878,
+      "step": 9991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3490389157101869e-06,
+      "loss": 0.8684,
+      "step": 9992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3476771098964869e-06,
+      "loss": 0.779,
+      "step": 9993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3463159421123718e-06,
+      "loss": 0.8024,
+      "step": 9994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3449554124582176e-06,
+      "loss": 0.8607,
+      "step": 9995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3435955210343532e-06,
+      "loss": 0.7737,
+      "step": 9996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3422362679410538e-06,
+      "loss": 0.8728,
+      "step": 9997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3408776532785539e-06,
+      "loss": 0.8293,
+      "step": 9998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339519677147042e-06,
+      "loss": 0.8077,
+      "step": 9999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3381623396466514e-06,
+      "loss": 0.8484,
+      "step": 10000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3368056408774742e-06,
+      "loss": 0.246,
+      "step": 10001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3354495809395584e-06,
+      "loss": 0.8927,
+      "step": 10002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3340941599328994e-06,
+      "loss": 0.8765,
+      "step": 10003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3327393779574426e-06,
+      "loss": 0.8194,
+      "step": 10004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3313852351130973e-06,
+      "loss": 0.8555,
+      "step": 10005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3300317314997136e-06,
+      "loss": 0.8007,
+      "step": 10006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3286788672171013e-06,
+      "loss": 0.9021,
+      "step": 10007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273266423650266e-06,
+      "loss": 0.8336,
+      "step": 10008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3259750570431973e-06,
+      "loss": 0.8898,
+      "step": 10009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3246241113512825e-06,
+      "loss": 0.8682,
+      "step": 10010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3232738053889049e-06,
+      "loss": 0.7608,
+      "step": 10011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3219241392556337e-06,
+      "loss": 0.9387,
+      "step": 10012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.320575113050997e-06,
+      "loss": 0.7775,
+      "step": 10013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319226726874472e-06,
+      "loss": 0.8607,
+      "step": 10014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3178789808254877e-06,
+      "loss": 0.9696,
+      "step": 10015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3165318750034296e-06,
+      "loss": 0.7716,
+      "step": 10016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3151854095076377e-06,
+      "loss": 0.8626,
+      "step": 10017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138395844373964e-06,
+      "loss": 0.2607,
+      "step": 10018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3124943998919492e-06,
+      "loss": 0.899,
+      "step": 10019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311149855970495e-06,
+      "loss": 0.9034,
+      "step": 10020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3098059527721774e-06,
+      "loss": 0.8414,
+      "step": 10021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084626903960995e-06,
+      "loss": 0.7943,
+      "step": 10022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071200689413133e-06,
+      "loss": 0.8185,
+      "step": 10023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3057780885068228e-06,
+      "loss": 0.2291,
+      "step": 10024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3044367491915889e-06,
+      "loss": 0.906,
+      "step": 10025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030960510945257e-06,
+      "loss": 0.8595,
+      "step": 10026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301755994314492e-06,
+      "loss": 0.2233,
+      "step": 10027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3004165789503075e-06,
+      "loss": 0.8442,
+      "step": 10028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2990778051007447e-06,
+      "loss": 0.9101,
+      "step": 10029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2977396728645208e-06,
+      "loss": 0.8037,
+      "step": 10030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296402182340315e-06,
+      "loss": 0.8456,
+      "step": 10031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2950653336267504e-06,
+      "loss": 0.8562,
+      "step": 10032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937291268224105e-06,
+      "loss": 0.8479,
+      "step": 10033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29239356202583e-06,
+      "loss": 0.8954,
+      "step": 10034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2910586393354928e-06,
+      "loss": 0.8741,
+      "step": 10035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2897243588498332e-06,
+      "loss": 0.8396,
+      "step": 10036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2883907206672475e-06,
+      "loss": 0.7813,
+      "step": 10037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2870577248860783e-06,
+      "loss": 0.8812,
+      "step": 10038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2857253716046192e-06,
+      "loss": 0.2553,
+      "step": 10039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2843936609211238e-06,
+      "loss": 0.7686,
+      "step": 10040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.283062592933788e-06,
+      "loss": 0.7793,
+      "step": 10041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2817321677407691e-06,
+      "loss": 0.2922,
+      "step": 10042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2804023854401748e-06,
+      "loss": 0.8984,
+      "step": 10043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2790732461300637e-06,
+      "loss": 0.9222,
+      "step": 10044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277744749908444e-06,
+      "loss": 0.8215,
+      "step": 10045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2764168968732826e-06,
+      "loss": 0.8332,
+      "step": 10046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2750896871225005e-06,
+      "loss": 0.9155,
+      "step": 10047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2737631207539602e-06,
+      "loss": 0.8457,
+      "step": 10048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2724371978654892e-06,
+      "loss": 0.7821,
+      "step": 10049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271111918554858e-06,
+      "loss": 0.8199,
+      "step": 10050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697872829197966e-06,
+      "loss": 0.8777,
+      "step": 10051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2684632910579863e-06,
+      "loss": 0.8722,
+      "step": 10052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267139943067055e-06,
+      "loss": 0.8835,
+      "step": 10053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2658172390445922e-06,
+      "loss": 0.8884,
+      "step": 10054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2644951790881299e-06,
+      "loss": 0.7977,
+      "step": 10055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2631737632951635e-06,
+      "loss": 0.8144,
+      "step": 10056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2618529917631294e-06,
+      "loss": 0.964,
+      "step": 10057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2605328645894289e-06,
+      "loss": 0.7902,
+      "step": 10058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2592133818714037e-06,
+      "loss": 0.9179,
+      "step": 10059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.257894543706355e-06,
+      "loss": 0.8255,
+      "step": 10060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565763501915384e-06,
+      "loss": 0.841,
+      "step": 10061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2552588014241552e-06,
+      "loss": 0.7903,
+      "step": 10062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2539418975013618e-06,
+      "loss": 0.9183,
+      "step": 10063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2526256385202717e-06,
+      "loss": 0.8729,
+      "step": 10064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2513100245779453e-06,
+      "loss": 0.8952,
+      "step": 10065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2499950557713947e-06,
+      "loss": 0.9263,
+      "step": 10066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2486807321975903e-06,
+      "loss": 0.8735,
+      "step": 10067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2473670539534467e-06,
+      "loss": 0.9579,
+      "step": 10068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2460540211358397e-06,
+      "loss": 0.9088,
+      "step": 10069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244741633841594e-06,
+      "loss": 0.8719,
+      "step": 10070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2434298921674826e-06,
+      "loss": 0.8633,
+      "step": 10071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2421187962102365e-06,
+      "loss": 0.8435,
+      "step": 10072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2408083460665387e-06,
+      "loss": 0.9319,
+      "step": 10073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2394985418330197e-06,
+      "loss": 0.8555,
+      "step": 10074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2381893836062687e-06,
+      "loss": 0.8871,
+      "step": 10075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2368808714828229e-06,
+      "loss": 0.8547,
+      "step": 10076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2355730055591697e-06,
+      "loss": 0.8989,
+      "step": 10077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2342657859317554e-06,
+      "loss": 0.8807,
+      "step": 10078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232959212696977e-06,
+      "loss": 0.8129,
+      "step": 10079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2316532859511787e-06,
+      "loss": 0.9011,
+      "step": 10080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2303480057906626e-06,
+      "loss": 0.8813,
+      "step": 10081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290433723116823e-06,
+      "loss": 0.7959,
+      "step": 10082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2277393856104403e-06,
+      "loss": 0.2652,
+      "step": 10083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2264360457830936e-06,
+      "loss": 0.8124,
+      "step": 10084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225133352925758e-06,
+      "loss": 0.8859,
+      "step": 10085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2238313071344853e-06,
+      "loss": 0.8409,
+      "step": 10086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2225299085052933e-06,
+      "loss": 0.8642,
+      "step": 10087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2212291571341516e-06,
+      "loss": 0.8172,
+      "step": 10088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2199290531169738e-06,
+      "loss": 0.8488,
+      "step": 10089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2186295965496332e-06,
+      "loss": 0.2884,
+      "step": 10090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2173307875279539e-06,
+      "loss": 0.9087,
+      "step": 10091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2160326261477084e-06,
+      "loss": 0.9019,
+      "step": 10092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2147351125046246e-06,
+      "loss": 0.8447,
+      "step": 10093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213438246694386e-06,
+      "loss": 0.8816,
+      "step": 10094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2121420288126207e-06,
+      "loss": 0.8543,
+      "step": 10095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2108464589549152e-06,
+      "loss": 0.8027,
+      "step": 10096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2095515372168043e-06,
+      "loss": 0.2674,
+      "step": 10097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.208257263693775e-06,
+      "loss": 0.8513,
+      "step": 10098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20696363848127e-06,
+      "loss": 0.8964,
+      "step": 10099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.205670661674686e-06,
+      "loss": 0.799,
+      "step": 10100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2043783333693616e-06,
+      "loss": 0.9252,
+      "step": 10101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2030866536605967e-06,
+      "loss": 0.8357,
+      "step": 10102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2017956226436445e-06,
+      "loss": 0.784,
+      "step": 10103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2005052404137008e-06,
+      "loss": 0.8404,
+      "step": 10104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1992155070659228e-06,
+      "loss": 0.9377,
+      "step": 10105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979264226954201e-06,
+      "loss": 0.8591,
+      "step": 10106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1966379873972423e-06,
+      "loss": 0.8563,
+      "step": 10107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1953502012664043e-06,
+      "loss": 0.8551,
+      "step": 10108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1940630643978713e-06,
+      "loss": 0.8471,
+      "step": 10109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1927765768865519e-06,
+      "loss": 0.243,
+      "step": 10110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1914907388273168e-06,
+      "loss": 0.9071,
+      "step": 10111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1902055503149856e-06,
+      "loss": 0.9202,
+      "step": 10112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188921011444325e-06,
+      "loss": 0.7783,
+      "step": 10113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1876371223100603e-06,
+      "loss": 0.8391,
+      "step": 10114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1863538830068689e-06,
+      "loss": 0.8949,
+      "step": 10115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1850712936293762e-06,
+      "loss": 0.8197,
+      "step": 10116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183789354272159e-06,
+      "loss": 0.8716,
+      "step": 10117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1825080650297527e-06,
+      "loss": 0.8229,
+      "step": 10118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.181227425996635e-06,
+      "loss": 0.8299,
+      "step": 10119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1799474372672458e-06,
+      "loss": 0.95,
+      "step": 10120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1786680989359745e-06,
+      "loss": 0.8826,
+      "step": 10121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177389411097154e-06,
+      "loss": 0.2732,
+      "step": 10122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1761113738450802e-06,
+      "loss": 0.8335,
+      "step": 10123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1748339872739977e-06,
+      "loss": 0.7976,
+      "step": 10124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173557251478099e-06,
+      "loss": 0.8224,
+      "step": 10125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1722811665515343e-06,
+      "loss": 0.8883,
+      "step": 10126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1710057325884017e-06,
+      "loss": 0.8314,
+      "step": 10127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1697309496827513e-06,
+      "loss": 0.8153,
+      "step": 10128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1684568179285883e-06,
+      "loss": 0.8684,
+      "step": 10129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1671833374198694e-06,
+      "loss": 0.8727,
+      "step": 10130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1659105082504995e-06,
+      "loss": 0.8217,
+      "step": 10131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1646383305143393e-06,
+      "loss": 0.808,
+      "step": 10132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1633668043052027e-06,
+      "loss": 0.8539,
+      "step": 10133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1620959297168489e-06,
+      "loss": 0.809,
+      "step": 10134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1608257068429973e-06,
+      "loss": 0.7782,
+      "step": 10135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159556135777311e-06,
+      "loss": 0.7964,
+      "step": 10136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158287216613414e-06,
+      "loss": 0.8817,
+      "step": 10137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1570189494448724e-06,
+      "loss": 0.8431,
+      "step": 10138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1557513343652138e-06,
+      "loss": 0.8969,
+      "step": 10139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.15448437146791e-06,
+      "loss": 0.791,
+      "step": 10140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532180608463883e-06,
+      "loss": 0.8397,
+      "step": 10141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519524025940298e-06,
+      "loss": 0.8527,
+      "step": 10142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1506873968041632e-06,
+      "loss": 0.9023,
+      "step": 10143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494230435700727e-06,
+      "loss": 0.2579,
+      "step": 10144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481593429849892e-06,
+      "loss": 0.9375,
+      "step": 10145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468962951421014e-06,
+      "loss": 0.8626,
+      "step": 10146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1456339001345506e-06,
+      "loss": 0.2674,
+      "step": 10147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1443721580554235e-06,
+      "loss": 0.9249,
+      "step": 10148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143111068997761e-06,
+      "loss": 0.7824,
+      "step": 10149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1418506330545576e-06,
+      "loss": 0.882,
+      "step": 10150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405908503187624e-06,
+      "loss": 0.8864,
+      "step": 10151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1393317208832677e-06,
+      "loss": 0.8313,
+      "step": 10152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138073244840926e-06,
+      "loss": 0.8256,
+      "step": 10153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1368154222845373e-06,
+      "loss": 0.9059,
+      "step": 10154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1355582533068532e-06,
+      "loss": 0.8713,
+      "step": 10155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1343017380005828e-06,
+      "loss": 0.8623,
+      "step": 10156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330458764583774e-06,
+      "loss": 0.8161,
+      "step": 10157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317906687728497e-06,
+      "loss": 0.7842,
+      "step": 10158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1305361150365557e-06,
+      "loss": 0.8292,
+      "step": 10159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1292822153420113e-06,
+      "loss": 0.7469,
+      "step": 10160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1280289697816759e-06,
+      "loss": 0.9095,
+      "step": 10161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126776378447969e-06,
+      "loss": 0.7846,
+      "step": 10162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1255244414332545e-06,
+      "loss": 0.9135,
+      "step": 10163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1242731588298517e-06,
+      "loss": 0.9013,
+      "step": 10164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1230225307300347e-06,
+      "loss": 0.876,
+      "step": 10165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1217725572260207e-06,
+      "loss": 0.2645,
+      "step": 10166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1205232384099863e-06,
+      "loss": 0.8621,
+      "step": 10167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192745743740596e-06,
+      "loss": 0.8295,
+      "step": 10168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1180265652103162e-06,
+      "loss": 0.841,
+      "step": 10169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1167792110107833e-06,
+      "loss": 0.823,
+      "step": 10170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1155325118674454e-06,
+      "loss": 0.2287,
+      "step": 10171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114286467872231e-06,
+      "loss": 0.8112,
+      "step": 10172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1130410791170265e-06,
+      "loss": 0.9781,
+      "step": 10173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.111796345693671e-06,
+      "loss": 0.9118,
+      "step": 10174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1105522676939474e-06,
+      "loss": 0.8594,
+      "step": 10175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1093088452095967e-06,
+      "loss": 0.8978,
+      "step": 10176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1080660783323126e-06,
+      "loss": 0.9501,
+      "step": 10177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1068239671537352e-06,
+      "loss": 0.7471,
+      "step": 10178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1055825117654574e-06,
+      "loss": 0.8747,
+      "step": 10179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1043417122590295e-06,
+      "loss": 0.8347,
+      "step": 10180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1031015687259429e-06,
+      "loss": 0.8481,
+      "step": 10181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.101862081257652e-06,
+      "loss": 0.8785,
+      "step": 10182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1006232499455571e-06,
+      "loss": 0.8983,
+      "step": 10183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0993850748810075e-06,
+      "loss": 0.9144,
+      "step": 10184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0981475561553101e-06,
+      "loss": 0.8109,
+      "step": 10185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0969106938597206e-06,
+      "loss": 0.7979,
+      "step": 10186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0956744880854452e-06,
+      "loss": 0.8319,
+      "step": 10187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094438938923642e-06,
+      "loss": 0.8326,
+      "step": 10188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093204046465427e-06,
+      "loss": 0.8196,
+      "step": 10189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091969810801854e-06,
+      "loss": 0.8831,
+      "step": 10190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.090736232023939e-06,
+      "loss": 0.9272,
+      "step": 10191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0895033102226516e-06,
+      "loss": 0.7921,
+      "step": 10192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0882710454889033e-06,
+      "loss": 0.7766,
+      "step": 10193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087039437913564e-06,
+      "loss": 0.8244,
+      "step": 10194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0858084875874563e-06,
+      "loss": 0.8397,
+      "step": 10195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084578194601348e-06,
+      "loss": 0.8774,
+      "step": 10196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0833485590459614e-06,
+      "loss": 0.9142,
+      "step": 10197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0821195810119755e-06,
+      "loss": 0.8441,
+      "step": 10198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0808912605900124e-06,
+      "loss": 0.7616,
+      "step": 10199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0796635978706482e-06,
+      "loss": 0.8359,
+      "step": 10200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0784365929444162e-06,
+      "loss": 0.9268,
+      "step": 10201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.077210245901793e-06,
+      "loss": 0.9286,
+      "step": 10202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0759845568332117e-06,
+      "loss": 0.8595,
+      "step": 10203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0747595258290577e-06,
+      "loss": 0.8752,
+      "step": 10204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0735351529796612e-06,
+      "loss": 0.9269,
+      "step": 10205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072311438375312e-06,
+      "loss": 0.2615,
+      "step": 10206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.071088382106249e-06,
+      "loss": 0.766,
+      "step": 10207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069865984262658e-06,
+      "loss": 0.8333,
+      "step": 10208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0686442449346822e-06,
+      "loss": 0.8156,
+      "step": 10209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0674231642124133e-06,
+      "loss": 0.7993,
+      "step": 10210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0662027421858912e-06,
+      "loss": 0.8638,
+      "step": 10211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064982978945115e-06,
+      "loss": 0.8415,
+      "step": 10212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0637638745800305e-06,
+      "loss": 0.8763,
+      "step": 10213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0625454291805338e-06,
+      "loss": 0.8667,
+      "step": 10214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061327642836475e-06,
+      "loss": 0.9373,
+      "step": 10215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0601105156376567e-06,
+      "loss": 0.9736,
+      "step": 10216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0588940476738275e-06,
+      "loss": 0.2157,
+      "step": 10217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0576782390346917e-06,
+      "loss": 0.8531,
+      "step": 10218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0564630898099094e-06,
+      "loss": 0.8965,
+      "step": 10219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0552486000890771e-06,
+      "loss": 0.8564,
+      "step": 10220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054034769961758e-06,
+      "loss": 0.2529,
+      "step": 10221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0528215995174617e-06,
+      "loss": 0.8601,
+      "step": 10222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0516090888456455e-06,
+      "loss": 0.8912,
+      "step": 10223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0503972380357207e-06,
+      "loss": 0.89,
+      "step": 10224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0491860471770554e-06,
+      "loss": 0.8795,
+      "step": 10225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0479755163589578e-06,
+      "loss": 0.8852,
+      "step": 10226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0467656456706954e-06,
+      "loss": 0.8599,
+      "step": 10227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045556435201488e-06,
+      "loss": 0.8649,
+      "step": 10228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0443478850404997e-06,
+      "loss": 0.7784,
+      "step": 10229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431399952768529e-06,
+      "loss": 0.821,
+      "step": 10230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0419327659996182e-06,
+      "loss": 0.9136,
+      "step": 10231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0407261972978133e-06,
+      "loss": 0.8764,
+      "step": 10232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395202892604162e-06,
+      "loss": 0.8558,
+      "step": 10233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0383150419763521e-06,
+      "loss": 0.8522,
+      "step": 10234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371104555344936e-06,
+      "loss": 0.7206,
+      "step": 10235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035906530023668e-06,
+      "loss": 0.8401,
+      "step": 10236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0347032655326595e-06,
+      "loss": 0.8948,
+      "step": 10237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0335006621501897e-06,
+      "loss": 0.8006,
+      "step": 10238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032298719964947e-06,
+      "loss": 0.8809,
+      "step": 10239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0310974390655592e-06,
+      "loss": 0.9122,
+      "step": 10240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02989681954061e-06,
+      "loss": 0.8353,
+      "step": 10241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0286968614786341e-06,
+      "loss": 0.8921,
+      "step": 10242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0274975649681197e-06,
+      "loss": 0.7824,
+      "step": 10243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0262989300975013e-06,
+      "loss": 0.9261,
+      "step": 10244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251009569551673e-06,
+      "loss": 0.8312,
+      "step": 10245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0239036456294616e-06,
+      "loss": 0.7393,
+      "step": 10246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0227069962086688e-06,
+      "loss": 0.8114,
+      "step": 10247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0215110087810366e-06,
+      "loss": 0.9028,
+      "step": 10248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0203156834347515e-06,
+      "loss": 0.9113,
+      "step": 10249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0191210202579649e-06,
+      "loss": 0.8032,
+      "step": 10250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017927019338767e-06,
+      "loss": 0.2304,
+      "step": 10251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167336807652074e-06,
+      "loss": 0.7721,
+      "step": 10252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0155410046252811e-06,
+      "loss": 0.8347,
+      "step": 10253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0143489910069392e-06,
+      "loss": 0.8524,
+      "step": 10254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0131576399980825e-06,
+      "loss": 0.8148,
+      "step": 10255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0119669516865605e-06,
+      "loss": 0.9038,
+      "step": 10256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107769261601774e-06,
+      "loss": 0.8278,
+      "step": 10257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0095875635066832e-06,
+      "loss": 0.9397,
+      "step": 10258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083988638137843e-06,
+      "loss": 0.8407,
+      "step": 10259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0072108271691394e-06,
+      "loss": 0.9527,
+      "step": 10260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006023453660353e-06,
+      "loss": 0.9738,
+      "step": 10261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0048367433749805e-06,
+      "loss": 0.8042,
+      "step": 10262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0036506964005332e-06,
+      "loss": 0.2353,
+      "step": 10263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0024653128244732e-06,
+      "loss": 0.8893,
+      "step": 10264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0012805927342073e-06,
+      "loss": 0.9049,
+      "step": 10265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0000965362171033e-06,
+      "loss": 0.8183,
+      "step": 10266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.989131433604682e-07,
+      "loss": 0.8978,
+      "step": 10267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.977304142515698e-07,
+      "loss": 0.9248,
+      "step": 10268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965483489776261e-07,
+      "loss": 0.86,
+      "step": 10269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953669476257976e-07,
+      "loss": 0.856,
+      "step": 10270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94186210283208e-07,
+      "loss": 0.8255,
+      "step": 10271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.930061370369205e-07,
+      "loss": 0.9476,
+      "step": 10272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.91826727973959e-07,
+      "loss": 0.9305,
+      "step": 10273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906479831812909e-07,
+      "loss": 0.7607,
+      "step": 10274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89469902745841e-07,
+      "loss": 0.8943,
+      "step": 10275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.882924867544774e-07,
+      "loss": 0.936,
+      "step": 10276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.871157352940264e-07,
+      "loss": 0.7616,
+      "step": 10277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85939648451265e-07,
+      "loss": 0.8051,
+      "step": 10278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.847642263129153e-07,
+      "loss": 0.7737,
+      "step": 10279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.835894689656545e-07,
+      "loss": 0.7107,
+      "step": 10280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.824153764961131e-07,
+      "loss": 0.8163,
+      "step": 10281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.812419489908665e-07,
+      "loss": 0.9241,
+      "step": 10282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.800691865364441e-07,
+      "loss": 0.9545,
+      "step": 10283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.788970892193294e-07,
+      "loss": 0.8243,
+      "step": 10284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.777256571259498e-07,
+      "loss": 0.8293,
+      "step": 10285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765548903426891e-07,
+      "loss": 0.8949,
+      "step": 10286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.753847889558842e-07,
+      "loss": 0.8991,
+      "step": 10287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.742153530518139e-07,
+      "loss": 0.9995,
+      "step": 10288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.730465827167157e-07,
+      "loss": 0.8883,
+      "step": 10289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.718784780367784e-07,
+      "loss": 0.7725,
+      "step": 10290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.707110390981345e-07,
+      "loss": 0.9002,
+      "step": 10291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69544265986876e-07,
+      "loss": 0.9819,
+      "step": 10292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.683781587890407e-07,
+      "loss": 0.8611,
+      "step": 10293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.672127175906144e-07,
+      "loss": 0.8597,
+      "step": 10294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.660479424775404e-07,
+      "loss": 0.7696,
+      "step": 10295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.648838335357125e-07,
+      "loss": 0.8363,
+      "step": 10296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.637203908509696e-07,
+      "loss": 0.9218,
+      "step": 10297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.625576145091065e-07,
+      "loss": 0.7936,
+      "step": 10298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.613955045958688e-07,
+      "loss": 0.915,
+      "step": 10299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.602340611969484e-07,
+      "loss": 0.8344,
+      "step": 10300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590732843979934e-07,
+      "loss": 0.9082,
+      "step": 10301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.579131742846015e-07,
+      "loss": 0.9253,
+      "step": 10302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567537309423159e-07,
+      "loss": 0.9424,
+      "step": 10303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555949544566366e-07,
+      "loss": 0.81,
+      "step": 10304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54436844913017e-07,
+      "loss": 0.7611,
+      "step": 10305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.532794023968506e-07,
+      "loss": 0.8796,
+      "step": 10306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.521226269934914e-07,
+      "loss": 0.1952,
+      "step": 10307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.509665187882422e-07,
+      "loss": 0.8285,
+      "step": 10308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.498110778663528e-07,
+      "loss": 0.7883,
+      "step": 10309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.486563043130271e-07,
+      "loss": 0.8249,
+      "step": 10310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475021982134213e-07,
+      "loss": 0.7678,
+      "step": 10311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.463487596526389e-07,
+      "loss": 0.9091,
+      "step": 10312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.451959887157325e-07,
+      "loss": 0.8268,
+      "step": 10313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.440438854877132e-07,
+      "loss": 0.2442,
+      "step": 10314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.428924500535341e-07,
+      "loss": 0.8314,
+      "step": 10315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.417416824981052e-07,
+      "loss": 0.89,
+      "step": 10316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.40591582906285e-07,
+      "loss": 0.7828,
+      "step": 10317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.394421513628815e-07,
+      "loss": 0.8508,
+      "step": 10318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.382933879526546e-07,
+      "loss": 0.8472,
+      "step": 10319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.371452927603187e-07,
+      "loss": 0.825,
+      "step": 10320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.359978658705304e-07,
+      "loss": 0.8311,
+      "step": 10321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.348511073679046e-07,
+      "loss": 0.8453,
+      "step": 10322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337050173370066e-07,
+      "loss": 0.8373,
+      "step": 10323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.325595958623446e-07,
+      "loss": 0.8715,
+      "step": 10324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314148430283854e-07,
+      "loss": 0.8484,
+      "step": 10325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.302707589195459e-07,
+      "loss": 0.7878,
+      "step": 10326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.291273436201886e-07,
+      "loss": 0.8704,
+      "step": 10327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.279845972146318e-07,
+      "loss": 0.8987,
+      "step": 10328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.268425197871433e-07,
+      "loss": 0.889,
+      "step": 10329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.257011114219383e-07,
+      "loss": 0.8298,
+      "step": 10330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24560372203187e-07,
+      "loss": 0.8038,
+      "step": 10331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.234203022150101e-07,
+      "loss": 0.7862,
+      "step": 10332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.222809015414746e-07,
+      "loss": 0.8628,
+      "step": 10333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.211421702666002e-07,
+      "loss": 0.7899,
+      "step": 10334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.200041084743616e-07,
+      "loss": 0.9405,
+      "step": 10335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.188667162486765e-07,
+      "loss": 0.8253,
+      "step": 10336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177299936734196e-07,
+      "loss": 0.9865,
+      "step": 10337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.165939408324143e-07,
+      "loss": 0.7631,
+      "step": 10338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.1545855780943e-07,
+      "loss": 0.8646,
+      "step": 10339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.143238446881953e-07,
+      "loss": 0.7803,
+      "step": 10340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.131898015523843e-07,
+      "loss": 0.8985,
+      "step": 10341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.120564284856193e-07,
+      "loss": 0.7385,
+      "step": 10342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.109237255714809e-07,
+      "loss": 0.2264,
+      "step": 10343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.097916928934913e-07,
+      "loss": 0.7938,
+      "step": 10344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.086603305351283e-07,
+      "loss": 0.8638,
+      "step": 10345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075296385798205e-07,
+      "loss": 0.7698,
+      "step": 10346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.06399617110948e-07,
+      "loss": 0.8917,
+      "step": 10347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.052702662118351e-07,
+      "loss": 0.8574,
+      "step": 10348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.041415859657632e-07,
+      "loss": 0.8335,
+      "step": 10349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030135764559644e-07,
+      "loss": 0.9254,
+      "step": 10350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.018862377656146e-07,
+      "loss": 0.8798,
+      "step": 10351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007595699778482e-07,
+      "loss": 0.8416,
+      "step": 10352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.996335731757466e-07,
+      "loss": 0.7786,
+      "step": 10353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985082474423413e-07,
+      "loss": 0.7975,
+      "step": 10354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.973835928606112e-07,
+      "loss": 0.8181,
+      "step": 10355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.962596095134946e-07,
+      "loss": 0.8902,
+      "step": 10356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.951362974838718e-07,
+      "loss": 0.9037,
+      "step": 10357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.940136568545765e-07,
+      "loss": 0.2552,
+      "step": 10358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92891687708397e-07,
+      "loss": 0.8406,
+      "step": 10359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.917703901280627e-07,
+      "loss": 0.8725,
+      "step": 10360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90649764196263e-07,
+      "loss": 0.8717,
+      "step": 10361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.895298099956329e-07,
+      "loss": 0.2424,
+      "step": 10362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.884105276087585e-07,
+      "loss": 0.8539,
+      "step": 10363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87291917118177e-07,
+      "loss": 0.8291,
+      "step": 10364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.861739786063761e-07,
+      "loss": 0.8248,
+      "step": 10365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850567121557907e-07,
+      "loss": 0.8162,
+      "step": 10366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.839401178488116e-07,
+      "loss": 0.9188,
+      "step": 10367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.828241957677785e-07,
+      "loss": 0.8189,
+      "step": 10368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.817089459949757e-07,
+      "loss": 0.7796,
+      "step": 10369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805943686126473e-07,
+      "loss": 0.8916,
+      "step": 10370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79480463702983e-07,
+      "loss": 0.9001,
+      "step": 10371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.783672313481195e-07,
+      "loss": 0.8708,
+      "step": 10372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772546716301522e-07,
+      "loss": 0.9192,
+      "step": 10373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.761427846311188e-07,
+      "loss": 0.8899,
+      "step": 10374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750315704330104e-07,
+      "loss": 0.8344,
+      "step": 10375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.739210291177703e-07,
+      "loss": 0.8078,
+      "step": 10376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728111607672917e-07,
+      "loss": 0.9635,
+      "step": 10377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.717019654634151e-07,
+      "loss": 0.7838,
+      "step": 10378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.705934432879337e-07,
+      "loss": 0.9022,
+      "step": 10379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.694855943225943e-07,
+      "loss": 0.8497,
+      "step": 10380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683784186490851e-07,
+      "loss": 0.9256,
+      "step": 10381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672719163490562e-07,
+      "loss": 0.7975,
+      "step": 10382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.661660875040967e-07,
+      "loss": 0.8527,
+      "step": 10383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.650609321957526e-07,
+      "loss": 0.8466,
+      "step": 10384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.639564505055231e-07,
+      "loss": 0.8698,
+      "step": 10385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628526425148498e-07,
+      "loss": 0.7964,
+      "step": 10386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617495083051264e-07,
+      "loss": 0.8372,
+      "step": 10387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.606470479577023e-07,
+      "loss": 0.8548,
+      "step": 10388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.595452615538757e-07,
+      "loss": 0.9333,
+      "step": 10389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.584441491748884e-07,
+      "loss": 0.9126,
+      "step": 10390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.573437109019411e-07,
+      "loss": 0.9128,
+      "step": 10391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.562439468161776e-07,
+      "loss": 0.7792,
+      "step": 10392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.551448569986986e-07,
+      "loss": 0.8541,
+      "step": 10393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.540464415305516e-07,
+      "loss": 0.7482,
+      "step": 10394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52948700492734e-07,
+      "loss": 0.9106,
+      "step": 10395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.518516339661931e-07,
+      "loss": 0.8745,
+      "step": 10396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.507552420318265e-07,
+      "loss": 0.9119,
+      "step": 10397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496595247704886e-07,
+      "loss": 0.9109,
+      "step": 10398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.485644822629724e-07,
+      "loss": 0.8484,
+      "step": 10399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47470114590031e-07,
+      "loss": 0.8627,
+      "step": 10400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463764218323622e-07,
+      "loss": 0.8377,
+      "step": 10401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452834040706159e-07,
+      "loss": 0.8453,
+      "step": 10402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.441910613853944e-07,
+      "loss": 0.9412,
+      "step": 10403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.430993938572441e-07,
+      "loss": 0.7885,
+      "step": 10404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.420084015666686e-07,
+      "loss": 0.7566,
+      "step": 10405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409180845941168e-07,
+      "loss": 0.8987,
+      "step": 10406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.398284430199921e-07,
+      "loss": 0.795,
+      "step": 10407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.387394769246416e-07,
+      "loss": 0.7275,
+      "step": 10408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.376511863883718e-07,
+      "loss": 0.8173,
+      "step": 10409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.365635714914277e-07,
+      "loss": 0.9554,
+      "step": 10410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.354766323140162e-07,
+      "loss": 0.8582,
+      "step": 10411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.343903689362876e-07,
+      "loss": 0.8854,
+      "step": 10412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.333047814383421e-07,
+      "loss": 0.8754,
+      "step": 10413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.322198699002337e-07,
+      "loss": 0.8515,
+      "step": 10414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311356344019661e-07,
+      "loss": 0.8293,
+      "step": 10415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300520750234909e-07,
+      "loss": 0.9114,
+      "step": 10416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.289691918447062e-07,
+      "loss": 0.8676,
+      "step": 10417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.278869849454718e-07,
+      "loss": 0.9185,
+      "step": 10418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.268054544055837e-07,
+      "loss": 0.9309,
+      "step": 10419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.257246003047992e-07,
+      "loss": 1.0248,
+      "step": 10420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.246444227228223e-07,
+      "loss": 0.9332,
+      "step": 10421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.235649217393015e-07,
+      "loss": 0.8624,
+      "step": 10422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.224860974338445e-07,
+      "loss": 0.7602,
+      "step": 10423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.214079498860039e-07,
+      "loss": 0.8357,
+      "step": 10424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.203304791752797e-07,
+      "loss": 0.8459,
+      "step": 10425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.192536853811317e-07,
+      "loss": 0.8234,
+      "step": 10426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181775685829596e-07,
+      "loss": 0.8009,
+      "step": 10427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171021288601156e-07,
+      "loss": 0.876,
+      "step": 10428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.16027366291906e-07,
+      "loss": 0.8572,
+      "step": 10429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149532809575855e-07,
+      "loss": 0.8909,
+      "step": 10430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138798729363562e-07,
+      "loss": 0.944,
+      "step": 10431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128071423073724e-07,
+      "loss": 0.8529,
+      "step": 10432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1173508914974e-07,
+      "loss": 0.8269,
+      "step": 10433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1066371354251e-07,
+      "loss": 0.2726,
+      "step": 10434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.095930155646881e-07,
+      "loss": 0.8574,
+      "step": 10435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.085229952952322e-07,
+      "loss": 0.8881,
+      "step": 10436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.074536528130394e-07,
+      "loss": 0.8312,
+      "step": 10437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.063849881969665e-07,
+      "loss": 0.9299,
+      "step": 10438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053170015258216e-07,
+      "loss": 0.851,
+      "step": 10439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.042496928783539e-07,
+      "loss": 0.7733,
+      "step": 10440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031830623332693e-07,
+      "loss": 0.9443,
+      "step": 10441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.021171099692249e-07,
+      "loss": 0.8668,
+      "step": 10442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.0105183586482e-07,
+      "loss": 0.848,
+      "step": 10443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.999872400986109e-07,
+      "loss": 0.7523,
+      "step": 10444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.989233227491045e-07,
+      "loss": 0.8136,
+      "step": 10445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.978600838947515e-07,
+      "loss": 0.8674,
+      "step": 10446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96797523613958e-07,
+      "loss": 0.8832,
+      "step": 10447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95735641985077e-07,
+      "loss": 0.9609,
+      "step": 10448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.946744390864114e-07,
+      "loss": 0.8693,
+      "step": 10449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936139149962175e-07,
+      "loss": 0.7538,
+      "step": 10450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.925540697926993e-07,
+      "loss": 0.2686,
+      "step": 10451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914949035540076e-07,
+      "loss": 0.8236,
+      "step": 10452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.904364163582489e-07,
+      "loss": 0.7918,
+      "step": 10453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.893786082834776e-07,
+      "loss": 0.8246,
+      "step": 10454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.883214794076943e-07,
+      "loss": 0.8369,
+      "step": 10455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.872650298088547e-07,
+      "loss": 0.7422,
+      "step": 10456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.862092595648663e-07,
+      "loss": 0.8083,
+      "step": 10457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.851541687535735e-07,
+      "loss": 0.8105,
+      "step": 10458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.840997574527842e-07,
+      "loss": 0.7904,
+      "step": 10459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.830460257402539e-07,
+      "loss": 0.2158,
+      "step": 10460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.819929736936816e-07,
+      "loss": 0.7935,
+      "step": 10461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.809406013907217e-07,
+      "loss": 0.8827,
+      "step": 10462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.798889089089789e-07,
+      "loss": 0.2781,
+      "step": 10463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.788378963260035e-07,
+      "loss": 0.8788,
+      "step": 10464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.777875637192989e-07,
+      "loss": 0.9098,
+      "step": 10465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.767379111663187e-07,
+      "loss": 0.8247,
+      "step": 10466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.756889387444632e-07,
+      "loss": 0.8594,
+      "step": 10467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.746406465310851e-07,
+      "loss": 0.9393,
+      "step": 10468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73593034603487e-07,
+      "loss": 0.908,
+      "step": 10469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.725461030389192e-07,
+      "loss": 0.8649,
+      "step": 10470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.714998519145844e-07,
+      "loss": 0.8862,
+      "step": 10471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.704542813076354e-07,
+      "loss": 0.7329,
+      "step": 10472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.694093912951705e-07,
+      "loss": 0.8204,
+      "step": 10473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683651819542426e-07,
+      "loss": 0.8185,
+      "step": 10474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.673216533618532e-07,
+      "loss": 0.8067,
+      "step": 10475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.66278805594951e-07,
+      "loss": 0.8596,
+      "step": 10476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.652366387304399e-07,
+      "loss": 0.9064,
+      "step": 10477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641951528451663e-07,
+      "loss": 0.8456,
+      "step": 10478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631543480159309e-07,
+      "loss": 0.9036,
+      "step": 10479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.621142243194845e-07,
+      "loss": 0.8573,
+      "step": 10480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.610747818325281e-07,
+      "loss": 0.9409,
+      "step": 10481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.600360206317081e-07,
+      "loss": 0.881,
+      "step": 10482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.589979407936243e-07,
+      "loss": 0.7958,
+      "step": 10483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.579605423948288e-07,
+      "loss": 0.7796,
+      "step": 10484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569238255118161e-07,
+      "loss": 0.7616,
+      "step": 10485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.558877902210371e-07,
+      "loss": 0.8223,
+      "step": 10486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.548524365988875e-07,
+      "loss": 0.9742,
+      "step": 10487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.538177647217193e-07,
+      "loss": 0.8844,
+      "step": 10488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.527837746658251e-07,
+      "loss": 0.7914,
+      "step": 10489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.517504665074571e-07,
+      "loss": 0.8745,
+      "step": 10490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.507178403228077e-07,
+      "loss": 0.2386,
+      "step": 10491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.496858961880249e-07,
+      "loss": 0.9344,
+      "step": 10492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.486546341792089e-07,
+      "loss": 0.846,
+      "step": 10493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476240543724012e-07,
+      "loss": 0.8482,
+      "step": 10494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.465941568435997e-07,
+      "loss": 0.8066,
+      "step": 10495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.455649416687493e-07,
+      "loss": 0.8159,
+      "step": 10496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.44536408923745e-07,
+      "loss": 0.866,
+      "step": 10497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.435085586844337e-07,
+      "loss": 0.787,
+      "step": 10498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424813910266082e-07,
+      "loss": 0.8311,
+      "step": 10499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.414549060260112e-07,
+      "loss": 0.8609,
+      "step": 10500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.404291037583389e-07,
+      "loss": 0.9004,
+      "step": 10501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39403984299235e-07,
+      "loss": 0.9153,
+      "step": 10502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.383795477242906e-07,
+      "loss": 0.7549,
+      "step": 10503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.373557941090504e-07,
+      "loss": 0.8489,
+      "step": 10504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363327235290063e-07,
+      "loss": 0.7755,
+      "step": 10505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.353103360595992e-07,
+      "loss": 0.8784,
+      "step": 10506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342886317762233e-07,
+      "loss": 0.8891,
+      "step": 10507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332676107542192e-07,
+      "loss": 0.8834,
+      "step": 10508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322472730688757e-07,
+      "loss": 0.8282,
+      "step": 10509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312276187954337e-07,
+      "loss": 0.8132,
+      "step": 10510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.302086480090875e-07,
+      "loss": 0.8674,
+      "step": 10511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.291903607849715e-07,
+      "loss": 0.8341,
+      "step": 10512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2817275719818e-07,
+      "loss": 0.7982,
+      "step": 10513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.271558373237464e-07,
+      "loss": 0.2305,
+      "step": 10514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.261396012366628e-07,
+      "loss": 0.8858,
+      "step": 10515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.251240490118682e-07,
+      "loss": 0.8639,
+      "step": 10516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.241091807242484e-07,
+      "loss": 0.8715,
+      "step": 10517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2309499644864e-07,
+      "loss": 0.8347,
+      "step": 10518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.220814962598322e-07,
+      "loss": 0.8164,
+      "step": 10519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.210686802325606e-07,
+      "loss": 0.8159,
+      "step": 10520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.20056548441509e-07,
+      "loss": 0.9443,
+      "step": 10521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.190451009613153e-07,
+      "loss": 0.8128,
+      "step": 10522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.180343378665622e-07,
+      "loss": 0.9467,
+      "step": 10523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170242592317856e-07,
+      "loss": 0.8544,
+      "step": 10524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.160148651314713e-07,
+      "loss": 0.8931,
+      "step": 10525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.150061556400489e-07,
+      "loss": 0.761,
+      "step": 10526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.139981308319044e-07,
+      "loss": 0.9622,
+      "step": 10527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129907907813715e-07,
+      "loss": 0.7926,
+      "step": 10528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119841355627299e-07,
+      "loss": 0.8962,
+      "step": 10529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109781652502101e-07,
+      "loss": 0.8409,
+      "step": 10530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.099728799179972e-07,
+      "loss": 0.8789,
+      "step": 10531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.089682796402175e-07,
+      "loss": 0.7967,
+      "step": 10532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079643644909528e-07,
+      "loss": 0.8555,
+      "step": 10533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.069611345442351e-07,
+      "loss": 0.9282,
+      "step": 10534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.059585898740396e-07,
+      "loss": 0.8301,
+      "step": 10535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.049567305542959e-07,
+      "loss": 0.8025,
+      "step": 10536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03955556658884e-07,
+      "loss": 0.8052,
+      "step": 10537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029550682616293e-07,
+      "loss": 0.8161,
+      "step": 10538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019552654363094e-07,
+      "loss": 0.8998,
+      "step": 10539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009561482566518e-07,
+      "loss": 0.8297,
+      "step": 10540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.999577167963279e-07,
+      "loss": 0.8303,
+      "step": 10541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989599711289663e-07,
+      "loss": 0.895,
+      "step": 10542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.979629113281427e-07,
+      "loss": 0.8991,
+      "step": 10543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969665374673773e-07,
+      "loss": 0.7549,
+      "step": 10544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.959708496201467e-07,
+      "loss": 0.8852,
+      "step": 10545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.949758478598734e-07,
+      "loss": 0.7582,
+      "step": 10546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939815322599285e-07,
+      "loss": 0.8753,
+      "step": 10547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.929879028936338e-07,
+      "loss": 0.8705,
+      "step": 10548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.919949598342624e-07,
+      "loss": 0.9267,
+      "step": 10549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910027031550326e-07,
+      "loss": 0.8425,
+      "step": 10550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.900111329291148e-07,
+      "loss": 0.7995,
+      "step": 10551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.890202492296283e-07,
+      "loss": 0.8436,
+      "step": 10552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880300521296412e-07,
+      "loss": 0.8942,
+      "step": 10553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.870405417021709e-07,
+      "loss": 0.7868,
+      "step": 10554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860517180201875e-07,
+      "loss": 0.7679,
+      "step": 10555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.850635811566053e-07,
+      "loss": 0.8265,
+      "step": 10556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840761311842903e-07,
+      "loss": 0.8263,
+      "step": 10557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.830893681760598e-07,
+      "loss": 0.8139,
+      "step": 10558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.821032922046766e-07,
+      "loss": 0.8657,
+      "step": 10559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.811179033428561e-07,
+      "loss": 0.8989,
+      "step": 10560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.801332016632623e-07,
+      "loss": 0.9265,
+      "step": 10561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.791491872385036e-07,
+      "loss": 0.8689,
+      "step": 10562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.781658601411467e-07,
+      "loss": 0.928,
+      "step": 10563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.771832204437023e-07,
+      "loss": 0.9336,
+      "step": 10564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762012682186292e-07,
+      "loss": 0.8242,
+      "step": 10565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.752200035383371e-07,
+      "loss": 0.7932,
+      "step": 10566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.742394264751895e-07,
+      "loss": 0.8456,
+      "step": 10567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.732595371014905e-07,
+      "loss": 0.8138,
+      "step": 10568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.722803354894991e-07,
+      "loss": 0.7617,
+      "step": 10569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.713018217114264e-07,
+      "loss": 0.8787,
+      "step": 10570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703239958394215e-07,
+      "loss": 0.8355,
+      "step": 10571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.693468579455942e-07,
+      "loss": 0.2477,
+      "step": 10572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.683704081020014e-07,
+      "loss": 0.8429,
+      "step": 10573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.673946463806424e-07,
+      "loss": 0.7968,
+      "step": 10574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.66419572853475e-07,
+      "loss": 0.9091,
+      "step": 10575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.654451875924006e-07,
+      "loss": 0.9292,
+      "step": 10576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.644714906692707e-07,
+      "loss": 0.8147,
+      "step": 10577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634984821558854e-07,
+      "loss": 0.9156,
+      "step": 10578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.625261621239987e-07,
+      "loss": 0.8901,
+      "step": 10579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.615545306453075e-07,
+      "loss": 0.8421,
+      "step": 10580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.605835877914613e-07,
+      "loss": 0.7872,
+      "step": 10581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.596133336340593e-07,
+      "loss": 0.8639,
+      "step": 10582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.586437682446467e-07,
+      "loss": 0.7917,
+      "step": 10583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.576748916947207e-07,
+      "loss": 0.8272,
+      "step": 10584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.567067040557295e-07,
+      "loss": 0.8937,
+      "step": 10585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557392053990652e-07,
+      "loss": 0.8816,
+      "step": 10586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.547723957960716e-07,
+      "loss": 0.7994,
+      "step": 10587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.538062753180463e-07,
+      "loss": 0.8188,
+      "step": 10588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528408440362266e-07,
+      "loss": 0.8844,
+      "step": 10589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.51876102021809e-07,
+      "loss": 0.7332,
+      "step": 10590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.509120493459309e-07,
+      "loss": 0.8578,
+      "step": 10591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.499486860796822e-07,
+      "loss": 0.8162,
+      "step": 10592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489860122941038e-07,
+      "loss": 0.9353,
+      "step": 10593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.480240280601857e-07,
+      "loss": 0.9062,
+      "step": 10594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.47062733448861e-07,
+      "loss": 0.7564,
+      "step": 10595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461021285310188e-07,
+      "loss": 0.8961,
+      "step": 10596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.451422133774965e-07,
+      "loss": 0.8871,
+      "step": 10597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441829880590755e-07,
+      "loss": 0.8895,
+      "step": 10598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.432244526464937e-07,
+      "loss": 0.8676,
+      "step": 10599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422666072104311e-07,
+      "loss": 0.7901,
+      "step": 10600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413094518215224e-07,
+      "loss": 0.8994,
+      "step": 10601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403529865503478e-07,
+      "loss": 0.881,
+      "step": 10602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393972114674385e-07,
+      "loss": 0.8483,
+      "step": 10603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.384421266432717e-07,
+      "loss": 0.8366,
+      "step": 10604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374877321482786e-07,
+      "loss": 0.8122,
+      "step": 10605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365340280528387e-07,
+      "loss": 0.6928,
+      "step": 10606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355810144272745e-07,
+      "loss": 0.8786,
+      "step": 10607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346286913418664e-07,
+      "loss": 0.2422,
+      "step": 10608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.336770588668362e-07,
+      "loss": 0.8544,
+      "step": 10609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.327261170723586e-07,
+      "loss": 0.8429,
+      "step": 10610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3177586602856e-07,
+      "loss": 0.8968,
+      "step": 10611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.308263058055098e-07,
+      "loss": 0.2844,
+      "step": 10612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.298774364732285e-07,
+      "loss": 0.9274,
+      "step": 10613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289292581016882e-07,
+      "loss": 0.8092,
+      "step": 10614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.279817707608093e-07,
+      "loss": 0.9306,
+      "step": 10615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270349745204585e-07,
+      "loss": 0.2515,
+      "step": 10616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.260888694504541e-07,
+      "loss": 0.8304,
+      "step": 10617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251434556205626e-07,
+      "loss": 0.8353,
+      "step": 10618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.241987331004984e-07,
+      "loss": 0.918,
+      "step": 10619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232547019599289e-07,
+      "loss": 0.9124,
+      "step": 10620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223113622684651e-07,
+      "loss": 0.7868,
+      "step": 10621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213687140956726e-07,
+      "loss": 0.8769,
+      "step": 10622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.20426757511059e-07,
+      "loss": 0.8438,
+      "step": 10623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194854925840888e-07,
+      "loss": 0.7909,
+      "step": 10624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185449193841686e-07,
+      "loss": 0.8754,
+      "step": 10625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.176050379806598e-07,
+      "loss": 0.8885,
+      "step": 10626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166658484428678e-07,
+      "loss": 0.8184,
+      "step": 10627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157273508400497e-07,
+      "loss": 0.9219,
+      "step": 10628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.147895452414132e-07,
+      "loss": 0.8947,
+      "step": 10629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.138524317161099e-07,
+      "loss": 0.9064,
+      "step": 10630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.129160103332455e-07,
+      "loss": 0.8169,
+      "step": 10631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119802811618725e-07,
+      "loss": 0.8736,
+      "step": 10632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.110452442709913e-07,
+      "loss": 0.8321,
+      "step": 10633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101108997295513e-07,
+      "loss": 0.8083,
+      "step": 10634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09177247606454e-07,
+      "loss": 0.8839,
+      "step": 10635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.082442879705463e-07,
+      "loss": 0.8885,
+      "step": 10636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.073120208906258e-07,
+      "loss": 0.8512,
+      "step": 10637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.063804464354395e-07,
+      "loss": 0.9113,
+      "step": 10638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054495646736814e-07,
+      "loss": 0.8824,
+      "step": 10639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.045193756739953e-07,
+      "loss": 0.8907,
+      "step": 10640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035898795049766e-07,
+      "loss": 0.894,
+      "step": 10641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026610762351637e-07,
+      "loss": 0.7034,
+      "step": 10642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.017329659330507e-07,
+      "loss": 0.8344,
+      "step": 10643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.00805548667075e-07,
+      "loss": 0.8041,
+      "step": 10644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.998788245056242e-07,
+      "loss": 0.9225,
+      "step": 10645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.989527935170369e-07,
+      "loss": 0.8586,
+      "step": 10646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980274557696009e-07,
+      "loss": 0.7846,
+      "step": 10647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971028113315491e-07,
+      "loss": 0.88,
+      "step": 10648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961788602710661e-07,
+      "loss": 0.9155,
+      "step": 10649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952556026562872e-07,
+      "loss": 0.7082,
+      "step": 10650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.943330385552903e-07,
+      "loss": 0.9221,
+      "step": 10651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.934111680361088e-07,
+      "loss": 0.8524,
+      "step": 10652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.924899911667237e-07,
+      "loss": 0.8379,
+      "step": 10653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.915695080150586e-07,
+      "loss": 0.9333,
+      "step": 10654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906497186489924e-07,
+      "loss": 0.8434,
+      "step": 10655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.897306231363542e-07,
+      "loss": 0.8137,
+      "step": 10656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888122215449143e-07,
+      "loss": 0.8116,
+      "step": 10657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.878945139423997e-07,
+      "loss": 0.8745,
+      "step": 10658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.869775003964829e-07,
+      "loss": 0.8923,
+      "step": 10659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86061180974783e-07,
+      "loss": 0.8746,
+      "step": 10660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.851455557448716e-07,
+      "loss": 0.8278,
+      "step": 10661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842306247742691e-07,
+      "loss": 0.7685,
+      "step": 10662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833163881304415e-07,
+      "loss": 0.8931,
+      "step": 10663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.82402845880804e-07,
+      "loss": 0.8855,
+      "step": 10664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814899980927257e-07,
+      "loss": 0.801,
+      "step": 10665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.805778448335164e-07,
+      "loss": 0.9736,
+      "step": 10666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.796663861704422e-07,
+      "loss": 0.8797,
+      "step": 10667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787556221707146e-07,
+      "loss": 0.8759,
+      "step": 10668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.778455529014926e-07,
+      "loss": 0.8443,
+      "step": 10669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769361784298855e-07,
+      "loss": 0.83,
+      "step": 10670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.760274988229542e-07,
+      "loss": 0.2652,
+      "step": 10671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.751195141477028e-07,
+      "loss": 0.8028,
+      "step": 10672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742122244710868e-07,
+      "loss": 0.2271,
+      "step": 10673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.733056298600137e-07,
+      "loss": 0.8506,
+      "step": 10674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.723997303813312e-07,
+      "loss": 0.812,
+      "step": 10675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.714945261018445e-07,
+      "loss": 0.7733,
+      "step": 10676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.70590017088305e-07,
+      "loss": 0.8724,
+      "step": 10677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.696862034074091e-07,
+      "loss": 0.2699,
+      "step": 10678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.687830851258058e-07,
+      "loss": 0.9141,
+      "step": 10679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67880662310094e-07,
+      "loss": 0.9159,
+      "step": 10680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.669789350268162e-07,
+      "loss": 0.9251,
+      "step": 10681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660779033424668e-07,
+      "loss": 0.8811,
+      "step": 10682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.651775673234916e-07,
+      "loss": 0.2542,
+      "step": 10683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.642779270362797e-07,
+      "loss": 0.7571,
+      "step": 10684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.633789825471714e-07,
+      "loss": 0.8532,
+      "step": 10685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.624807339224558e-07,
+      "loss": 0.957,
+      "step": 10686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.615831812283701e-07,
+      "loss": 0.8464,
+      "step": 10687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.606863245311011e-07,
+      "loss": 0.844,
+      "step": 10688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.59790163896785e-07,
+      "loss": 0.884,
+      "step": 10689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.588946993915034e-07,
+      "loss": 0.8861,
+      "step": 10690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.57999931081289e-07,
+      "loss": 0.8638,
+      "step": 10691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571058590321244e-07,
+      "loss": 0.7831,
+      "step": 10692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562124833099369e-07,
+      "loss": 0.9052,
+      "step": 10693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553198039806073e-07,
+      "loss": 0.8633,
+      "step": 10694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.544278211099607e-07,
+      "loss": 0.7388,
+      "step": 10695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.535365347637722e-07,
+      "loss": 0.8831,
+      "step": 10696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.52645945007767e-07,
+      "loss": 0.94,
+      "step": 10697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.517560519076193e-07,
+      "loss": 0.8111,
+      "step": 10698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.508668555289475e-07,
+      "loss": 0.2473,
+      "step": 10699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.499783559373228e-07,
+      "loss": 0.8886,
+      "step": 10700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.490905531982671e-07,
+      "loss": 0.8868,
+      "step": 10701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.482034473772424e-07,
+      "loss": 0.7798,
+      "step": 10702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.473170385396698e-07,
+      "loss": 0.842,
+      "step": 10703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.464313267509091e-07,
+      "loss": 0.8626,
+      "step": 10704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455463120762772e-07,
+      "loss": 0.9388,
+      "step": 10705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.446619945810328e-07,
+      "loss": 0.8101,
+      "step": 10706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437783743303892e-07,
+      "loss": 0.7838,
+      "step": 10707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.428954513895035e-07,
+      "loss": 0.7767,
+      "step": 10708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420132258234822e-07,
+      "loss": 0.8479,
+      "step": 10709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411316976973846e-07,
+      "loss": 0.2629,
+      "step": 10710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.402508670762119e-07,
+      "loss": 0.878,
+      "step": 10711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393707340249199e-07,
+      "loss": 0.8259,
+      "step": 10712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384912986084079e-07,
+      "loss": 0.8716,
+      "step": 10713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376125608915272e-07,
+      "loss": 0.2815,
+      "step": 10714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367345209390784e-07,
+      "loss": 0.2354,
+      "step": 10715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358571788158085e-07,
+      "loss": 0.7988,
+      "step": 10716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349805345864101e-07,
+      "loss": 0.8783,
+      "step": 10717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341045883155305e-07,
+      "loss": 0.874,
+      "step": 10718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332293400677635e-07,
+      "loss": 0.8145,
+      "step": 10719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.323547899076476e-07,
+      "loss": 0.8051,
+      "step": 10720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.314809378996766e-07,
+      "loss": 0.8472,
+      "step": 10721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.306077841082858e-07,
+      "loss": 0.9181,
+      "step": 10722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297353285978635e-07,
+      "loss": 0.7843,
+      "step": 10723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288635714327461e-07,
+      "loss": 0.8383,
+      "step": 10724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279925126772179e-07,
+      "loss": 0.9115,
+      "step": 10725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271221523955094e-07,
+      "loss": 0.8088,
+      "step": 10726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262524906518029e-07,
+      "loss": 0.8597,
+      "step": 10727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.253835275102281e-07,
+      "loss": 0.7485,
+      "step": 10728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.245152630348638e-07,
+      "loss": 0.852,
+      "step": 10729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.236476972897353e-07,
+      "loss": 0.855,
+      "step": 10730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227808303388182e-07,
+      "loss": 0.8423,
+      "step": 10731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219146622460347e-07,
+      "loss": 0.9015,
+      "step": 10732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.210491930752604e-07,
+      "loss": 0.8781,
+      "step": 10733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.201844228903108e-07,
+      "loss": 0.8855,
+      "step": 10734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193203517549583e-07,
+      "loss": 0.843,
+      "step": 10735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184569797329197e-07,
+      "loss": 0.7641,
+      "step": 10736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175943068878608e-07,
+      "loss": 0.862,
+      "step": 10737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.16732333283394e-07,
+      "loss": 0.9402,
+      "step": 10738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15871058983084e-07,
+      "loss": 0.8605,
+      "step": 10739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150104840504411e-07,
+      "loss": 0.8263,
+      "step": 10740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.141506085489234e-07,
+      "loss": 0.9566,
+      "step": 10741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132914325419424e-07,
+      "loss": 0.7622,
+      "step": 10742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.12432956092852e-07,
+      "loss": 0.9222,
+      "step": 10743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115751792649559e-07,
+      "loss": 0.8396,
+      "step": 10744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.107181021215113e-07,
+      "loss": 0.8072,
+      "step": 10745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098617247257165e-07,
+      "loss": 0.783,
+      "step": 10746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09006047140721e-07,
+      "loss": 0.7734,
+      "step": 10747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.081510694296266e-07,
+      "loss": 0.8344,
+      "step": 10748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07296791655475e-07,
+      "loss": 0.7762,
+      "step": 10749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.064432138812659e-07,
+      "loss": 0.8812,
+      "step": 10750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.05590336169941e-07,
+      "loss": 0.9021,
+      "step": 10751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.047381585843924e-07,
+      "loss": 0.8636,
+      "step": 10752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.038866811874599e-07,
+      "loss": 0.8271,
+      "step": 10753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.030359040419342e-07,
+      "loss": 0.8336,
+      "step": 10754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.021858272105495e-07,
+      "loss": 0.8801,
+      "step": 10755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013364507559937e-07,
+      "loss": 0.8362,
+      "step": 10756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004877747408987e-07,
+      "loss": 0.8459,
+      "step": 10757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.996397992278468e-07,
+      "loss": 0.868,
+      "step": 10758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.98792524279369e-07,
+      "loss": 0.8365,
+      "step": 10759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.979459499579453e-07,
+      "loss": 0.8061,
+      "step": 10760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971000763259992e-07,
+      "loss": 0.7878,
+      "step": 10761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962549034459085e-07,
+      "loss": 0.9385,
+      "step": 10762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.95410431379999e-07,
+      "loss": 0.7691,
+      "step": 10763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.945666601905385e-07,
+      "loss": 0.8745,
+      "step": 10764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937235899397508e-07,
+      "loss": 0.8173,
+      "step": 10765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928812206898048e-07,
+      "loss": 0.9057,
+      "step": 10766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.920395525028132e-07,
+      "loss": 0.7997,
+      "step": 10767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.911985854408452e-07,
+      "loss": 0.86,
+      "step": 10768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.903583195659134e-07,
+      "loss": 0.8448,
+      "step": 10769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895187549399793e-07,
+      "loss": 0.8752,
+      "step": 10770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886798916249536e-07,
+      "loss": 0.8694,
+      "step": 10771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.878417296826954e-07,
+      "loss": 0.8296,
+      "step": 10772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870042691750099e-07,
+      "loss": 0.7217,
+      "step": 10773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.861675101636531e-07,
+      "loss": 0.7726,
+      "step": 10774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853314527103292e-07,
+      "loss": 0.8619,
+      "step": 10775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844960968766877e-07,
+      "loss": 0.8447,
+      "step": 10776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.836614427243303e-07,
+      "loss": 0.7745,
+      "step": 10777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.828274903148045e-07,
+      "loss": 0.8257,
+      "step": 10778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819942397096055e-07,
+      "loss": 0.7171,
+      "step": 10779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.811616909701788e-07,
+      "loss": 0.8804,
+      "step": 10780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.803298441579196e-07,
+      "loss": 0.2201,
+      "step": 10781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794986993341644e-07,
+      "loss": 0.9147,
+      "step": 10782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786682565602041e-07,
+      "loss": 0.8781,
+      "step": 10783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.778385158972799e-07,
+      "loss": 0.816,
+      "step": 10784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.770094774065725e-07,
+      "loss": 0.8119,
+      "step": 10785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7618114114921785e-07,
+      "loss": 0.8746,
+      "step": 10786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7535350718630246e-07,
+      "loss": 0.7969,
+      "step": 10787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.745265755788486e-07,
+      "loss": 0.9131,
+      "step": 10788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.737003463878398e-07,
+      "loss": 0.8668,
+      "step": 10789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.728748196742028e-07,
+      "loss": 0.8503,
+      "step": 10790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7204999549881117e-07,
+      "loss": 0.8062,
+      "step": 10791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712258739224884e-07,
+      "loss": 0.8805,
+      "step": 10792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.704024550060071e-07,
+      "loss": 0.9634,
+      "step": 10793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6957973881008643e-07,
+      "loss": 0.8306,
+      "step": 10794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.687577253953934e-07,
+      "loss": 0.9007,
+      "step": 10795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6793641482254513e-07,
+      "loss": 0.7877,
+      "step": 10796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6711580715210536e-07,
+      "loss": 0.861,
+      "step": 10797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.662959024445857e-07,
+      "loss": 0.2339,
+      "step": 10798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.654767007604477e-07,
+      "loss": 0.8159,
+      "step": 10799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.646582021600976e-07,
+      "loss": 0.902,
+      "step": 10800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6384040670389485e-07,
+      "loss": 0.8502,
+      "step": 10801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630233144521445e-07,
+      "loss": 0.7614,
+      "step": 10802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.622069254650974e-07,
+      "loss": 0.8046,
+      "step": 10803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.613912398029552e-07,
+      "loss": 0.7358,
+      "step": 10804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6057625752586986e-07,
+      "loss": 0.2747,
+      "step": 10805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5976197869393556e-07,
+      "loss": 0.8042,
+      "step": 10806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.589484033672009e-07,
+      "loss": 0.7849,
+      "step": 10807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.581355316056579e-07,
+      "loss": 0.7512,
+      "step": 10808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5732336346924645e-07,
+      "loss": 0.8634,
+      "step": 10809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5651189901785965e-07,
+      "loss": 0.8232,
+      "step": 10810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557011383113363e-07,
+      "loss": 0.8296,
+      "step": 10811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548910814094587e-07,
+      "loss": 0.9159,
+      "step": 10812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540817283719634e-07,
+      "loss": 0.9021,
+      "step": 10813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.532730792585338e-07,
+      "loss": 0.8734,
+      "step": 10814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5246513412879777e-07,
+      "loss": 0.8164,
+      "step": 10815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5165789304233655e-07,
+      "loss": 0.7013,
+      "step": 10816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.508513560586747e-07,
+      "loss": 0.7874,
+      "step": 10817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5004552323728913e-07,
+      "loss": 0.8519,
+      "step": 10818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4924039463759893e-07,
+      "loss": 0.8023,
+      "step": 10819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4843597031898e-07,
+      "loss": 0.8779,
+      "step": 10820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.47632250340746e-07,
+      "loss": 0.9334,
+      "step": 10821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.468292347621672e-07,
+      "loss": 0.7748,
+      "step": 10822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.460269236424597e-07,
+      "loss": 0.8272,
+      "step": 10823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4522531704078276e-07,
+      "loss": 0.9282,
+      "step": 10824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.444244150162513e-07,
+      "loss": 0.7783,
+      "step": 10825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.436242176279215e-07,
+      "loss": 0.8566,
+      "step": 10826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.428247249348017e-07,
+      "loss": 0.8531,
+      "step": 10827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.420259369958479e-07,
+      "loss": 0.2312,
+      "step": 10828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.412278538699644e-07,
+      "loss": 0.9098,
+      "step": 10829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4043047561599826e-07,
+      "loss": 0.9012,
+      "step": 10830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3963380229275264e-07,
+      "loss": 0.8866,
+      "step": 10831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3883783395897386e-07,
+      "loss": 0.857,
+      "step": 10832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.380425706733571e-07,
+      "loss": 0.7964,
+      "step": 10833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.372480124945466e-07,
+      "loss": 0.8538,
+      "step": 10834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3645415948113226e-07,
+      "loss": 0.9453,
+      "step": 10835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3566101169165374e-07,
+      "loss": 0.8683,
+      "step": 10836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3486856918459995e-07,
+      "loss": 0.8435,
+      "step": 10837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.340768320184052e-07,
+      "loss": 0.7831,
+      "step": 10838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3328580025145284e-07,
+      "loss": 0.7951,
+      "step": 10839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3249547394207394e-07,
+      "loss": 0.8905,
+      "step": 10840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.317058531485496e-07,
+      "loss": 0.8748,
+      "step": 10841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.309169379291034e-07,
+      "loss": 0.9312,
+      "step": 10842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.301287283419142e-07,
+      "loss": 0.8955,
+      "step": 10843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293412244451034e-07,
+      "loss": 0.8743,
+      "step": 10844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2855442629674335e-07,
+      "loss": 0.9509,
+      "step": 10845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2776833395485215e-07,
+      "loss": 0.844,
+      "step": 10846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.269829474773979e-07,
+      "loss": 0.756,
+      "step": 10847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2619826692229416e-07,
+      "loss": 0.8527,
+      "step": 10848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2541429234740585e-07,
+      "loss": 0.7975,
+      "step": 10849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.246310238105433e-07,
+      "loss": 0.7981,
+      "step": 10850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2384846136946487e-07,
+      "loss": 0.7991,
+      "step": 10851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.230666050818777e-07,
+      "loss": 0.9371,
+      "step": 10852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222854550054356e-07,
+      "loss": 0.8359,
+      "step": 10853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2150501119774146e-07,
+      "loss": 0.8411,
+      "step": 10854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.207252737163481e-07,
+      "loss": 0.7639,
+      "step": 10855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.199462426187506e-07,
+      "loss": 0.8342,
+      "step": 10856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1916791796239754e-07,
+      "loss": 0.8888,
+      "step": 10857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1839029980468517e-07,
+      "loss": 0.8108,
+      "step": 10858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1761338820295206e-07,
+      "loss": 0.272,
+      "step": 10859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168371832144891e-07,
+      "loss": 0.912,
+      "step": 10860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.160616848965349e-07,
+      "loss": 0.8073,
+      "step": 10861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.152868933062748e-07,
+      "loss": 0.9323,
+      "step": 10862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1451280850084316e-07,
+      "loss": 0.8719,
+      "step": 10863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1373943053732323e-07,
+      "loss": 0.9223,
+      "step": 10864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1296675947274155e-07,
+      "loss": 0.8841,
+      "step": 10865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.121947953640759e-07,
+      "loss": 0.8107,
+      "step": 10866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.114235382682552e-07,
+      "loss": 0.831,
+      "step": 10867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1065298824214727e-07,
+      "loss": 0.8667,
+      "step": 10868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098831453425767e-07,
+      "loss": 0.7968,
+      "step": 10869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0911400962631355e-07,
+      "loss": 0.8793,
+      "step": 10870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.083455811500703e-07,
+      "loss": 0.7964,
+      "step": 10871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.075778599705138e-07,
+      "loss": 0.7288,
+      "step": 10872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.068108461442577e-07,
+      "loss": 0.7955,
+      "step": 10873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0604453972785896e-07,
+      "loss": 0.912,
+      "step": 10874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0527894077782793e-07,
+      "loss": 0.8909,
+      "step": 10875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0451404935062054e-07,
+      "loss": 0.8243,
+      "step": 10876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037498655026395e-07,
+      "loss": 0.8599,
+      "step": 10877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029863892902375e-07,
+      "loss": 0.7298,
+      "step": 10878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.022236207697139e-07,
+      "loss": 0.8628,
+      "step": 10879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0146155999731596e-07,
+      "loss": 0.9575,
+      "step": 10880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.007002070292365e-07,
+      "loss": 0.8844,
+      "step": 10881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9993956192162173e-07,
+      "loss": 0.8552,
+      "step": 10882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9917962473055905e-07,
+      "loss": 0.7129,
+      "step": 10883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.98420395512088e-07,
+      "loss": 0.2622,
+      "step": 10884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9766187432219714e-07,
+      "loss": 0.7816,
+      "step": 10885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.969040612168162e-07,
+      "loss": 0.9301,
+      "step": 10886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9614695625183055e-07,
+      "loss": 0.8778,
+      "step": 10887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.953905594830687e-07,
+      "loss": 0.7666,
+      "step": 10888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9463487096630616e-07,
+      "loss": 1.0408,
+      "step": 10889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.938798907572705e-07,
+      "loss": 0.8035,
+      "step": 10890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9312561891163505e-07,
+      "loss": 0.7803,
+      "step": 10891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.923720554850163e-07,
+      "loss": 0.8264,
+      "step": 10892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.916192005329866e-07,
+      "loss": 0.7488,
+      "step": 10893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9086705411106154e-07,
+      "loss": 0.2383,
+      "step": 10894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.901156162747033e-07,
+      "loss": 0.8271,
+      "step": 10895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.893648870793254e-07,
+      "loss": 0.2459,
+      "step": 10896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.886148665802869e-07,
+      "loss": 0.8207,
+      "step": 10897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8786555483289354e-07,
+      "loss": 0.8683,
+      "step": 10898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.871169518924001e-07,
+      "loss": 0.827,
+      "step": 10899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8636905781401333e-07,
+      "loss": 0.8258,
+      "step": 10900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8562187265287933e-07,
+      "loss": 0.7478,
+      "step": 10901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.848753964640961e-07,
+      "loss": 0.8921,
+      "step": 10902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.84129629302713e-07,
+      "loss": 0.7655,
+      "step": 10903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.833845712237194e-07,
+      "loss": 0.8549,
+      "step": 10904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.82640222282058e-07,
+      "loss": 0.9175,
+      "step": 10905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.818965825326193e-07,
+      "loss": 0.815,
+      "step": 10906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8115365203023723e-07,
+      "loss": 0.2223,
+      "step": 10907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.80411430829698e-07,
+      "loss": 0.8987,
+      "step": 10908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7966991898573446e-07,
+      "loss": 0.851,
+      "step": 10909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.78929116553024e-07,
+      "loss": 0.8521,
+      "step": 10910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.781890235861962e-07,
+      "loss": 0.9334,
+      "step": 10911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7744964013982423e-07,
+      "loss": 0.734,
+      "step": 10912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7671096626843093e-07,
+      "loss": 0.7906,
+      "step": 10913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7597300202648734e-07,
+      "loss": 0.7464,
+      "step": 10914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.752357474684132e-07,
+      "loss": 0.8602,
+      "step": 10915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7449920264857163e-07,
+      "loss": 0.7856,
+      "step": 10916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7376336762127597e-07,
+      "loss": 0.8278,
+      "step": 10917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730282424407905e-07,
+      "loss": 0.8543,
+      "step": 10918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.722938271613208e-07,
+      "loss": 0.8358,
+      "step": 10919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.715601218370246e-07,
+      "loss": 0.8977,
+      "step": 10920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.708271265220087e-07,
+      "loss": 0.8685,
+      "step": 10921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.700948412703187e-07,
+      "loss": 0.7438,
+      "step": 10922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6936326613595695e-07,
+      "loss": 0.8724,
+      "step": 10923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6863240117287256e-07,
+      "loss": 0.8845,
+      "step": 10924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6790224643495576e-07,
+      "loss": 0.9339,
+      "step": 10925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6717280197605234e-07,
+      "loss": 0.8198,
+      "step": 10926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.664440678499515e-07,
+      "loss": 0.8117,
+      "step": 10927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6571604411038906e-07,
+      "loss": 0.7839,
+      "step": 10928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6498873081105113e-07,
+      "loss": 0.7942,
+      "step": 10929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6426212800557247e-07,
+      "loss": 0.8994,
+      "step": 10930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.635362357475303e-07,
+      "loss": 0.9254,
+      "step": 10931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.628110540904539e-07,
+      "loss": 0.9212,
+      "step": 10932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.620865830878184e-07,
+      "loss": 0.8881,
+      "step": 10933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.613628227930466e-07,
+      "loss": 0.7704,
+      "step": 10934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6063977325950903e-07,
+      "loss": 0.8736,
+      "step": 10935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.599174345405265e-07,
+      "loss": 0.8432,
+      "step": 10936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591958066893619e-07,
+      "loss": 0.92,
+      "step": 10937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.584748897592294e-07,
+      "loss": 0.9604,
+      "step": 10938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.577546838032908e-07,
+      "loss": 0.2421,
+      "step": 10939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570351888746537e-07,
+      "loss": 0.238,
+      "step": 10940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5631640502637564e-07,
+      "loss": 0.8824,
+      "step": 10941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.555983323114598e-07,
+      "loss": 0.8982,
+      "step": 10942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.54880970782856e-07,
+      "loss": 0.7894,
+      "step": 10943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.541643204934642e-07,
+      "loss": 0.802,
+      "step": 10944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.53448381496132e-07,
+      "loss": 0.8637,
+      "step": 10945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527331538436518e-07,
+      "loss": 0.9141,
+      "step": 10946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5201863758876464e-07,
+      "loss": 0.726,
+      "step": 10947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.513048327841617e-07,
+      "loss": 0.8369,
+      "step": 10948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505917394824765e-07,
+      "loss": 0.8252,
+      "step": 10949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498793577362969e-07,
+      "loss": 0.799,
+      "step": 10950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.491676875981509e-07,
+      "loss": 0.8037,
+      "step": 10951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4845672912052096e-07,
+      "loss": 0.87,
+      "step": 10952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4774648235583073e-07,
+      "loss": 0.8423,
+      "step": 10953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470369473564572e-07,
+      "loss": 0.8687,
+      "step": 10954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463281241747185e-07,
+      "loss": 0.845,
+      "step": 10955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4562001286288727e-07,
+      "loss": 0.703,
+      "step": 10956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.449126134731795e-07,
+      "loss": 0.8213,
+      "step": 10957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.442059260577568e-07,
+      "loss": 0.773,
+      "step": 10958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.434999506687353e-07,
+      "loss": 0.7874,
+      "step": 10959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4279468735817e-07,
+      "loss": 0.933,
+      "step": 10960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.420901361780704e-07,
+      "loss": 0.9012,
+      "step": 10961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.413862971803905e-07,
+      "loss": 0.9073,
+      "step": 10962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40683170417031e-07,
+      "loss": 0.8989,
+      "step": 10963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.399807559398405e-07,
+      "loss": 0.8672,
+      "step": 10964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3927905380061635e-07,
+      "loss": 0.8292,
+      "step": 10965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3857806405110385e-07,
+      "loss": 0.8402,
+      "step": 10966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3787778674299276e-07,
+      "loss": 0.8289,
+      "step": 10967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.37178221927924e-07,
+      "loss": 0.7598,
+      "step": 10968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3647936965748084e-07,
+      "loss": 0.7754,
+      "step": 10969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357812299832008e-07,
+      "loss": 0.8899,
+      "step": 10970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.35083802956564e-07,
+      "loss": 0.8939,
+      "step": 10971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.343870886289979e-07,
+      "loss": 0.8084,
+      "step": 10972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.336910870518817e-07,
+      "loss": 0.8138,
+      "step": 10973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.329957982765353e-07,
+      "loss": 0.8356,
+      "step": 10974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3230122235423433e-07,
+      "loss": 0.7572,
+      "step": 10975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3160735933619346e-07,
+      "loss": 0.8668,
+      "step": 10976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3091420927358175e-07,
+      "loss": 0.7853,
+      "step": 10977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.302217722175105e-07,
+      "loss": 0.8978,
+      "step": 10978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2953004821904e-07,
+      "loss": 0.9419,
+      "step": 10979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2883903732918276e-07,
+      "loss": 0.8193,
+      "step": 10980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2814873959888915e-07,
+      "loss": 0.8191,
+      "step": 10981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2745915507906624e-07,
+      "loss": 0.7861,
+      "step": 10982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.267702838205633e-07,
+      "loss": 0.8171,
+      "step": 10983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2608212587417866e-07,
+      "loss": 0.8593,
+      "step": 10984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253946812906561e-07,
+      "loss": 0.8391,
+      "step": 10985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247079501206907e-07,
+      "loss": 0.8704,
+      "step": 10986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2402193241491965e-07,
+      "loss": 0.8058,
+      "step": 10987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.233366282239325e-07,
+      "loss": 0.8754,
+      "step": 10988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.226520375982656e-07,
+      "loss": 0.7621,
+      "step": 10989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219681605883973e-07,
+      "loss": 0.289,
+      "step": 10990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.212849972447607e-07,
+      "loss": 0.8314,
+      "step": 10991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2060254761773216e-07,
+      "loss": 0.7787,
+      "step": 10992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1992081175763354e-07,
+      "loss": 0.8856,
+      "step": 10993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1923978971474034e-07,
+      "loss": 0.8796,
+      "step": 10994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.185594815392701e-07,
+      "loss": 0.7953,
+      "step": 10995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.178798872813882e-07,
+      "loss": 0.8519,
+      "step": 10996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.17201006991209e-07,
+      "loss": 0.8463,
+      "step": 10997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1652284071879593e-07,
+      "loss": 0.7514,
+      "step": 10998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1584538851415325e-07,
+      "loss": 0.8255,
+      "step": 10999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.151686504272411e-07,
+      "loss": 0.7972,
+      "step": 11000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1449262650796063e-07,
+      "loss": 0.9161,
+      "step": 11001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138173168061631e-07,
+      "loss": 0.2402,
+      "step": 11002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.131427213716454e-07,
+      "loss": 0.9545,
+      "step": 11003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.124688402541554e-07,
+      "loss": 0.9743,
+      "step": 11004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1179567350338226e-07,
+      "loss": 0.8958,
+      "step": 11005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.111232211689663e-07,
+      "loss": 0.8555,
+      "step": 11006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.104514833004979e-07,
+      "loss": 0.8096,
+      "step": 11007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.097804599475085e-07,
+      "loss": 0.9045,
+      "step": 11008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0911015115947964e-07,
+      "loss": 0.9206,
+      "step": 11009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0844055698584395e-07,
+      "loss": 0.8153,
+      "step": 11010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0777167747597423e-07,
+      "loss": 0.8706,
+      "step": 11011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0710351267919546e-07,
+      "loss": 0.8868,
+      "step": 11012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.064360626447793e-07,
+      "loss": 0.8211,
+      "step": 11013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.057693274219442e-07,
+      "loss": 0.8778,
+      "step": 11014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0510330705985413e-07,
+      "loss": 0.7384,
+      "step": 11015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0443800160762427e-07,
+      "loss": 0.2626,
+      "step": 11016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0377341111431203e-07,
+      "loss": 0.8537,
+      "step": 11017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0310953562892707e-07,
+      "loss": 0.9485,
+      "step": 11018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0244637520042366e-07,
+      "loss": 0.8812,
+      "step": 11019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0178392987770253e-07,
+      "loss": 0.8278,
+      "step": 11020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0112219970961474e-07,
+      "loss": 0.7825,
+      "step": 11021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.004611847449568e-07,
+      "loss": 0.8003,
+      "step": 11022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9980088503247075e-07,
+      "loss": 0.8979,
+      "step": 11023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.991413006208499e-07,
+      "loss": 0.9331,
+      "step": 11024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9848243155873204e-07,
+      "loss": 0.8312,
+      "step": 11025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.978242778947005e-07,
+      "loss": 0.8895,
+      "step": 11026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9716683967728974e-07,
+      "loss": 0.8689,
+      "step": 11027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9651011695498e-07,
+      "loss": 0.8383,
+      "step": 11028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.95854109776198e-07,
+      "loss": 0.8998,
+      "step": 11029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9519881818931837e-07,
+      "loss": 0.8054,
+      "step": 11030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9454424224266475e-07,
+      "loss": 0.8572,
+      "step": 11031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9389038198450295e-07,
+      "loss": 0.8731,
+      "step": 11032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.93237237463051e-07,
+      "loss": 0.8437,
+      "step": 11033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9258480872647264e-07,
+      "loss": 0.8014,
+      "step": 11034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.919330958228783e-07,
+      "loss": 0.902,
+      "step": 11035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.91282098800324e-07,
+      "loss": 0.9184,
+      "step": 11036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9063181770681684e-07,
+      "loss": 0.8266,
+      "step": 11037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8998225259030734e-07,
+      "loss": 0.8311,
+      "step": 11038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8933340349869723e-07,
+      "loss": 0.7634,
+      "step": 11039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8868527047983264e-07,
+      "loss": 0.8213,
+      "step": 11040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880378535815054e-07,
+      "loss": 0.8796,
+      "step": 11041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8739115285145835e-07,
+      "loss": 0.8365,
+      "step": 11042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.867451683373812e-07,
+      "loss": 0.9116,
+      "step": 11043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.860999000869058e-07,
+      "loss": 0.8652,
+      "step": 11044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8545534814761744e-07,
+      "loss": 0.8929,
+      "step": 11045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848115125670459e-07,
+      "loss": 0.8806,
+      "step": 11046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8416839339266534e-07,
+      "loss": 0.9677,
+      "step": 11047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8352599067190344e-07,
+      "loss": 0.7742,
+      "step": 11048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.828843044521301e-07,
+      "loss": 0.8737,
+      "step": 11049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.822433347806619e-07,
+      "loss": 0.8259,
+      "step": 11050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.816030817047677e-07,
+      "loss": 0.9395,
+      "step": 11051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.809635452716597e-07,
+      "loss": 0.8615,
+      "step": 11052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.803247255284958e-07,
+      "loss": 0.8635,
+      "step": 11053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7968662252238487e-07,
+      "loss": 0.7451,
+      "step": 11054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7904923630038164e-07,
+      "loss": 0.8535,
+      "step": 11055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7841256690948395e-07,
+      "loss": 0.8951,
+      "step": 11056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.777766143966443e-07,
+      "loss": 0.8542,
+      "step": 11057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7714137880875736e-07,
+      "loss": 0.8448,
+      "step": 11058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.765068601926657e-07,
+      "loss": 0.7781,
+      "step": 11059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7587305859515745e-07,
+      "loss": 0.8474,
+      "step": 11060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.752399740629741e-07,
+      "loss": 0.8004,
+      "step": 11061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.74607606642795e-07,
+      "loss": 0.784,
+      "step": 11062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7397595638125383e-07,
+      "loss": 0.9111,
+      "step": 11063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733450233249291e-07,
+      "loss": 0.2502,
+      "step": 11064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727148075203456e-07,
+      "loss": 0.825,
+      "step": 11065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7208530901397745e-07,
+      "loss": 0.8238,
+      "step": 11066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7145652785224297e-07,
+      "loss": 0.8971,
+      "step": 11067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7082846408150843e-07,
+      "loss": 0.8833,
+      "step": 11068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.702011177480879e-07,
+      "loss": 0.8712,
+      "step": 11069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6957448889824543e-07,
+      "loss": 0.678,
+      "step": 11070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.689485775781853e-07,
+      "loss": 0.8582,
+      "step": 11071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.68323383834066e-07,
+      "loss": 0.8627,
+      "step": 11072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6769890771198625e-07,
+      "loss": 0.9255,
+      "step": 11073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6707514925799815e-07,
+      "loss": 0.8188,
+      "step": 11074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664521085180982e-07,
+      "loss": 0.8794,
+      "step": 11075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.658297855382297e-07,
+      "loss": 0.8807,
+      "step": 11076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6520818036428144e-07,
+      "loss": 0.8408,
+      "step": 11077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.645872930420923e-07,
+      "loss": 0.8733,
+      "step": 11078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6396712361744794e-07,
+      "loss": 0.7229,
+      "step": 11079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633476721360784e-07,
+      "loss": 0.8579,
+      "step": 11080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6272893864366487e-07,
+      "loss": 0.881,
+      "step": 11081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.62110923185831e-07,
+      "loss": 0.8334,
+      "step": 11082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614936258081513e-07,
+      "loss": 0.8883,
+      "step": 11083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6087704655614497e-07,
+      "loss": 0.8172,
+      "step": 11084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.60261185475279e-07,
+      "loss": 0.8315,
+      "step": 11085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5964604261096927e-07,
+      "loss": 0.8892,
+      "step": 11086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590316180085739e-07,
+      "loss": 0.778,
+      "step": 11087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5841791171340445e-07,
+      "loss": 0.9218,
+      "step": 11088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5780492377071254e-07,
+      "loss": 0.8347,
+      "step": 11089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5719265422570414e-07,
+      "loss": 0.8836,
+      "step": 11090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.565811031235255e-07,
+      "loss": 0.2293,
+      "step": 11091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.559702705092737e-07,
+      "loss": 0.8008,
+      "step": 11092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5536015642799285e-07,
+      "loss": 0.8577,
+      "step": 11093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5475076092467244e-07,
+      "loss": 0.871,
+      "step": 11094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5414208404425103e-07,
+      "loss": 0.8274,
+      "step": 11095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.535341258316115e-07,
+      "loss": 0.8261,
+      "step": 11096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5292688633158704e-07,
+      "loss": 0.8813,
+      "step": 11097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.523203655889539e-07,
+      "loss": 0.7596,
+      "step": 11098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.517145636484386e-07,
+      "loss": 0.8018,
+      "step": 11099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.511094805547132e-07,
+      "loss": 0.8789,
+      "step": 11100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5050511635239636e-07,
+      "loss": 0.8699,
+      "step": 11101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4990147108605587e-07,
+      "loss": 0.8014,
+      "step": 11102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.492985448002039e-07,
+      "loss": 0.7637,
+      "step": 11103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4869633753930143e-07,
+      "loss": 0.7926,
+      "step": 11104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480948493477553e-07,
+      "loss": 0.8101,
+      "step": 11105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4749408026991994e-07,
+      "loss": 0.8339,
+      "step": 11106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4689403035009663e-07,
+      "loss": 0.7745,
+      "step": 11107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4629469963253436e-07,
+      "loss": 0.8822,
+      "step": 11108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4569608816142567e-07,
+      "loss": 0.7896,
+      "step": 11109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.450981959809151e-07,
+      "loss": 0.8598,
+      "step": 11110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4450102313509085e-07,
+      "loss": 0.8682,
+      "step": 11111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.439045696679887e-07,
+      "loss": 0.8304,
+      "step": 11112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.433088356235924e-07,
+      "loss": 0.8828,
+      "step": 11113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427138210458324e-07,
+      "loss": 0.925,
+      "step": 11114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.421195259785836e-07,
+      "loss": 0.8638,
+      "step": 11115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4152595046567086e-07,
+      "loss": 0.8417,
+      "step": 11116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.40933094550867e-07,
+      "loss": 0.8697,
+      "step": 11117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4034095827788486e-07,
+      "loss": 0.7944,
+      "step": 11118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.397495416903928e-07,
+      "loss": 0.9288,
+      "step": 11119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.391588448320026e-07,
+      "loss": 0.8646,
+      "step": 11120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.385688677462705e-07,
+      "loss": 0.92,
+      "step": 11121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3797961047670292e-07,
+      "loss": 0.8362,
+      "step": 11122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.373910730667539e-07,
+      "loss": 0.7929,
+      "step": 11123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3680325555981874e-07,
+      "loss": 0.8517,
+      "step": 11124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3621615799924725e-07,
+      "loss": 0.8681,
+      "step": 11125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.356297804283314e-07,
+      "loss": 0.8624,
+      "step": 11126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3504412289031108e-07,
+      "loss": 0.9417,
+      "step": 11127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3445918542837286e-07,
+      "loss": 0.8167,
+      "step": 11128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3387496808565114e-07,
+      "loss": 0.8724,
+      "step": 11129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3329147090522475e-07,
+      "loss": 0.844,
+      "step": 11130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3270869393012375e-07,
+      "loss": 0.2504,
+      "step": 11131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.321266372033215e-07,
+      "loss": 0.7059,
+      "step": 11132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3154530076773928e-07,
+      "loss": 0.8445,
+      "step": 11133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.30964684666245e-07,
+      "loss": 0.9155,
+      "step": 11134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.303847889416544e-07,
+      "loss": 0.8734,
+      "step": 11135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2980561363672993e-07,
+      "loss": 0.917,
+      "step": 11136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2922715879417855e-07,
+      "loss": 0.8027,
+      "step": 11137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.286494244566595e-07,
+      "loss": 0.8953,
+      "step": 11138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2807241066677199e-07,
+      "loss": 0.8464,
+      "step": 11139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2749611746706645e-07,
+      "loss": 0.897,
+      "step": 11140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2692054490004002e-07,
+      "loss": 0.8842,
+      "step": 11141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2634569300813425e-07,
+      "loss": 0.8565,
+      "step": 11142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2577156183374082e-07,
+      "loss": 0.8082,
+      "step": 11143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.25198151419197e-07,
+      "loss": 0.8397,
+      "step": 11144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2462546180678556e-07,
+      "loss": 0.8568,
+      "step": 11145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2405349303873613e-07,
+      "loss": 0.8641,
+      "step": 11146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2348224515722827e-07,
+      "loss": 0.8082,
+      "step": 11147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2291171820438607e-07,
+      "loss": 0.786,
+      "step": 11148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223419122222792e-07,
+      "loss": 0.8879,
+      "step": 11149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2177282725292625e-07,
+      "loss": 0.7842,
+      "step": 11150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.212044633382915e-07,
+      "loss": 0.8107,
+      "step": 11151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.20636820520288e-07,
+      "loss": 0.7871,
+      "step": 11152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2006989884077346e-07,
+      "loss": 0.8552,
+      "step": 11153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1950369834155216e-07,
+      "loss": 0.8232,
+      "step": 11154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.189382190643774e-07,
+      "loss": 0.8533,
+      "step": 11155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1837346105094914e-07,
+      "loss": 0.8937,
+      "step": 11156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1780942434291075e-07,
+      "loss": 0.9017,
+      "step": 11157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1724610898185672e-07,
+      "loss": 0.2692,
+      "step": 11158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1668351500932604e-07,
+      "loss": 0.8569,
+      "step": 11159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1612164246680222e-07,
+      "loss": 0.935,
+      "step": 11160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.15560491395721e-07,
+      "loss": 0.8552,
+      "step": 11161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1500006183746258e-07,
+      "loss": 0.9478,
+      "step": 11162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1444035383335171e-07,
+      "loss": 0.8129,
+      "step": 11163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1388136742466202e-07,
+      "loss": 0.9479,
+      "step": 11164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1332310265261502e-07,
+      "loss": 0.8117,
+      "step": 11165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1276555955837664e-07,
+      "loss": 0.7425,
+      "step": 11166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1220873818306065e-07,
+      "loss": 0.7278,
+      "step": 11167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1165263856772643e-07,
+      "loss": 0.8308,
+      "step": 11168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1109726075338344e-07,
+      "loss": 0.8184,
+      "step": 11169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.105426047809833e-07,
+      "loss": 0.2869,
+      "step": 11170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0998867069143004e-07,
+      "loss": 0.8139,
+      "step": 11171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0943545852556757e-07,
+      "loss": 0.239,
+      "step": 11172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.088829683241922e-07,
+      "loss": 0.8884,
+      "step": 11173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0833120012804576e-07,
+      "loss": 0.8396,
+      "step": 11174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0778015397781348e-07,
+      "loss": 0.2357,
+      "step": 11175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0722982991413288e-07,
+      "loss": 0.8858,
+      "step": 11176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0668022797758259e-07,
+      "loss": 0.9038,
+      "step": 11177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.061313482086924e-07,
+      "loss": 0.9521,
+      "step": 11178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0558319064793886e-07,
+      "loss": 0.8251,
+      "step": 11179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0503575533574072e-07,
+      "loss": 0.7748,
+      "step": 11180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0448904231246568e-07,
+      "loss": 0.8337,
+      "step": 11181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0394305161843153e-07,
+      "loss": 0.2553,
+      "step": 11182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0339778329389938e-07,
+      "loss": 0.8139,
+      "step": 11183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0285323737907592e-07,
+      "loss": 0.2521,
+      "step": 11184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0230941391411907e-07,
+      "loss": 0.8528,
+      "step": 11185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0176631293912896e-07,
+      "loss": 0.7914,
+      "step": 11186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0122393449415468e-07,
+      "loss": 0.8105,
+      "step": 11187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006822786191931e-07,
+      "loss": 0.8108,
+      "step": 11188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0014134535418339e-07,
+      "loss": 0.9177,
+      "step": 11189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9960113473901809e-07,
+      "loss": 0.9014,
+      "step": 11190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.990616468135298e-07,
+      "loss": 0.7971,
+      "step": 11191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9852288161750221e-07,
+      "loss": 0.7983,
+      "step": 11192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9798483919066358e-07,
+      "loss": 0.8656,
+      "step": 11193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9744751957269103e-07,
+      "loss": 0.8256,
+      "step": 11194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9691092280320512e-07,
+      "loss": 0.8266,
+      "step": 11195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9637504892177638e-07,
+      "loss": 0.847,
+      "step": 11196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.958398979679199e-07,
+      "loss": 0.8517,
+      "step": 11197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9530546998109856e-07,
+      "loss": 0.9065,
+      "step": 11198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.947717650007208e-07,
+      "loss": 0.8439,
+      "step": 11199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9423878306614407e-07,
+      "loss": 0.8381,
+      "step": 11200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9370652421666914e-07,
+      "loss": 0.8707,
+      "step": 11201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.931749884915457e-07,
+      "loss": 1.0019,
+      "step": 11202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.926441759299702e-07,
+      "loss": 0.8571,
+      "step": 11203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9211408657108466e-07,
+      "loss": 0.7789,
+      "step": 11204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.915847204539778e-07,
+      "loss": 0.9118,
+      "step": 11205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910560776176873e-07,
+      "loss": 0.8934,
+      "step": 11206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.905281581011942e-07,
+      "loss": 0.2151,
+      "step": 11207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.900009619434273e-07,
+      "loss": 0.8376,
+      "step": 11208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.894744891832645e-07,
+      "loss": 0.8688,
+      "step": 11209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.88948739859528e-07,
+      "loss": 0.2552,
+      "step": 11210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8842371401098458e-07,
+      "loss": 0.8666,
+      "step": 11211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8789941167635328e-07,
+      "loss": 0.8516,
+      "step": 11212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8737583289429317e-07,
+      "loss": 0.8177,
+      "step": 11213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.868529777034167e-07,
+      "loss": 0.7935,
+      "step": 11214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8633084614227747e-07,
+      "loss": 0.9333,
+      "step": 11215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8580943824937915e-07,
+      "loss": 0.8296,
+      "step": 11216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.852887540631698e-07,
+      "loss": 0.8357,
+      "step": 11217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.847687936220466e-07,
+      "loss": 0.7879,
+      "step": 11218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8424955696434987e-07,
+      "loss": 0.7597,
+      "step": 11219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8373104412836905e-07,
+      "loss": 0.9138,
+      "step": 11220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8321325515234246e-07,
+      "loss": 0.7165,
+      "step": 11221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8269619007444838e-07,
+      "loss": 0.7777,
+      "step": 11222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.821798489328175e-07,
+      "loss": 0.237,
+      "step": 11223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8166423176552595e-07,
+      "loss": 0.8442,
+      "step": 11224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8114933861059337e-07,
+      "loss": 0.9392,
+      "step": 11225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8063516950599046e-07,
+      "loss": 0.8545,
+      "step": 11226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801217244896325e-07,
+      "loss": 0.8482,
+      "step": 11227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7960900359938027e-07,
+      "loss": 0.2403,
+      "step": 11228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7909700687304244e-07,
+      "loss": 0.8112,
+      "step": 11229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7858573434837546e-07,
+      "loss": 0.7988,
+      "step": 11230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7807518606308026e-07,
+      "loss": 0.8282,
+      "step": 11231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7756536205480346e-07,
+      "loss": 0.8228,
+      "step": 11232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.770562623611427e-07,
+      "loss": 0.7782,
+      "step": 11233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7654788701963688e-07,
+      "loss": 0.8409,
+      "step": 11234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7604023606777488e-07,
+      "loss": 0.874,
+      "step": 11235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755333095429934e-07,
+      "loss": 0.8132,
+      "step": 11236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7502710748266927e-07,
+      "loss": 0.7858,
+      "step": 11237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7452162992413368e-07,
+      "loss": 0.8551,
+      "step": 11238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.740168769046613e-07,
+      "loss": 0.8778,
+      "step": 11239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.735128484614701e-07,
+      "loss": 0.2469,
+      "step": 11240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7300954463173146e-07,
+      "loss": 0.8573,
+      "step": 11241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7250696545255573e-07,
+      "loss": 0.2402,
+      "step": 11242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7200511096100548e-07,
+      "loss": 0.8604,
+      "step": 11243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7150398119408662e-07,
+      "loss": 0.8468,
+      "step": 11244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7100357618875408e-07,
+      "loss": 0.8654,
+      "step": 11245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705038959819083e-07,
+      "loss": 0.8235,
+      "step": 11246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.700049406103943e-07,
+      "loss": 0.832,
+      "step": 11247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6950671011100816e-07,
+      "loss": 0.8607,
+      "step": 11248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.690092045204872e-07,
+      "loss": 0.8322,
+      "step": 11249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.685124238755198e-07,
+      "loss": 0.9141,
+      "step": 11250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6801636821273891e-07,
+      "loss": 0.8668,
+      "step": 11251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6752103756872308e-07,
+      "loss": 0.8285,
+      "step": 11252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6702643197999857e-07,
+      "loss": 0.845,
+      "step": 11253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.665325514830385e-07,
+      "loss": 0.8435,
+      "step": 11254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6603939611426146e-07,
+      "loss": 0.868,
+      "step": 11255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6554696591003282e-07,
+      "loss": 0.9636,
+      "step": 11256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.650552609066669e-07,
+      "loss": 0.797,
+      "step": 11257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6456428114042023e-07,
+      "loss": 0.2312,
+      "step": 11258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6407402664749828e-07,
+      "loss": 0.8285,
+      "step": 11259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6358449746405435e-07,
+      "loss": 0.8432,
+      "step": 11260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6309569362618516e-07,
+      "loss": 0.8199,
+      "step": 11261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.626076151699374e-07,
+      "loss": 0.853,
+      "step": 11262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6212026213130006e-07,
+      "loss": 0.2428,
+      "step": 11263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6163363454621216e-07,
+      "loss": 0.8649,
+      "step": 11264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114773245055725e-07,
+      "loss": 0.8751,
+      "step": 11265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066255588016778e-07,
+      "loss": 0.9083,
+      "step": 11266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6017810487081954e-07,
+      "loss": 0.8907,
+      "step": 11267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.596943794582362e-07,
+      "loss": 0.8903,
+      "step": 11268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592113796780903e-07,
+      "loss": 0.9277,
+      "step": 11269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5872910556599562e-07,
+      "loss": 0.8392,
+      "step": 11270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5824755715751817e-07,
+      "loss": 0.9229,
+      "step": 11271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5776673448816727e-07,
+      "loss": 0.7289,
+      "step": 11272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.572866375933968e-07,
+      "loss": 0.9128,
+      "step": 11273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5680726650861178e-07,
+      "loss": 0.922,
+      "step": 11274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.563286212691606e-07,
+      "loss": 0.9379,
+      "step": 11275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5585070191033946e-07,
+      "loss": 1.0105,
+      "step": 11276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.553735084673902e-07,
+      "loss": 0.8861,
+      "step": 11277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.548970409755013e-07,
+      "loss": 0.8122,
+      "step": 11278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.544212994698091e-07,
+      "loss": 0.8246,
+      "step": 11279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5394628398539336e-07,
+      "loss": 0.7552,
+      "step": 11280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.534719945572827e-07,
+      "loss": 0.9443,
+      "step": 11281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5299843122045243e-07,
+      "loss": 0.8353,
+      "step": 11282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5252559400982247e-07,
+      "loss": 0.9441,
+      "step": 11283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5205348296026158e-07,
+      "loss": 0.8229,
+      "step": 11284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5158209810658186e-07,
+      "loss": 0.9181,
+      "step": 11285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5111143948354555e-07,
+      "loss": 0.8316,
+      "step": 11286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5064150712585823e-07,
+      "loss": 0.9197,
+      "step": 11287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5017230106817216e-07,
+      "loss": 0.781,
+      "step": 11288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4970382134508965e-07,
+      "loss": 0.7656,
+      "step": 11289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.492360679911531e-07,
+      "loss": 0.2726,
+      "step": 11290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.487690410408582e-07,
+      "loss": 0.8707,
+      "step": 11291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4830274052864413e-07,
+      "loss": 0.7814,
+      "step": 11292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4783716648889447e-07,
+      "loss": 0.7867,
+      "step": 11293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4737231895594063e-07,
+      "loss": 0.8906,
+      "step": 11294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4690819796406297e-07,
+      "loss": 0.9428,
+      "step": 11295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4644480354748413e-07,
+      "loss": 0.8694,
+      "step": 11296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4598213574037679e-07,
+      "loss": 0.8852,
+      "step": 11297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4552019457685806e-07,
+      "loss": 0.8437,
+      "step": 11298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4505898009099072e-07,
+      "loss": 0.8291,
+      "step": 11299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4459849231678648e-07,
+      "loss": 0.9199,
+      "step": 11300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4413873128820145e-07,
+      "loss": 0.7846,
+      "step": 11301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4367969703913852e-07,
+      "loss": 0.8549,
+      "step": 11302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4322138960344956e-07,
+      "loss": 0.243,
+      "step": 11303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4276380901492638e-07,
+      "loss": 0.8835,
+      "step": 11304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.423069553073153e-07,
+      "loss": 0.8652,
+      "step": 11305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4185082851430276e-07,
+      "loss": 0.8156,
+      "step": 11306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4139542866952517e-07,
+      "loss": 0.891,
+      "step": 11307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409407558065623e-07,
+      "loss": 0.8813,
+      "step": 11308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4048680995894403e-07,
+      "loss": 0.9067,
+      "step": 11309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.400335911601447e-07,
+      "loss": 0.8436,
+      "step": 11310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3958109944358422e-07,
+      "loss": 0.8903,
+      "step": 11311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3912933484262924e-07,
+      "loss": 0.7564,
+      "step": 11312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3867829739059425e-07,
+      "loss": 0.8282,
+      "step": 11313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3822798712073927e-07,
+      "loss": 0.8659,
+      "step": 11314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3777840406626997e-07,
+      "loss": 0.9382,
+      "step": 11315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3732954826033873e-07,
+      "loss": 0.8725,
+      "step": 11316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3688141973604575e-07,
+      "loss": 0.835,
+      "step": 11317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3643401852643455e-07,
+      "loss": 0.8248,
+      "step": 11318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3598734466449882e-07,
+      "loss": 0.8495,
+      "step": 11319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3554139818317437e-07,
+      "loss": 0.9047,
+      "step": 11320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3509617911534823e-07,
+      "loss": 0.2669,
+      "step": 11321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.346516874938497e-07,
+      "loss": 0.9191,
+      "step": 11322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3420792335145593e-07,
+      "loss": 0.8834,
+      "step": 11323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3376488672089184e-07,
+      "loss": 0.8783,
+      "step": 11324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333225776348268e-07,
+      "loss": 0.8419,
+      "step": 11325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3288099612587478e-07,
+      "loss": 0.8356,
+      "step": 11326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3244014222660085e-07,
+      "loss": 0.8799,
+      "step": 11327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.320000159695134e-07,
+      "loss": 0.8532,
+      "step": 11328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.315606173870676e-07,
+      "loss": 0.7852,
+      "step": 11329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3112194651166533e-07,
+      "loss": 0.7679,
+      "step": 11330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3068400337565402e-07,
+      "loss": 0.8685,
+      "step": 11331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3024678801132783e-07,
+      "loss": 0.9111,
+      "step": 11332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.298103004509288e-07,
+      "loss": 0.8363,
+      "step": 11333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2937454072664334e-07,
+      "loss": 0.267,
+      "step": 11334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2893950887060246e-07,
+      "loss": 0.8351,
+      "step": 11335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2850520491488828e-07,
+      "loss": 0.7797,
+      "step": 11336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2807162889152736e-07,
+      "loss": 0.7846,
+      "step": 11337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.276387808324886e-07,
+      "loss": 0.7678,
+      "step": 11338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2720666076969423e-07,
+      "loss": 0.2559,
+      "step": 11339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2677526873500655e-07,
+      "loss": 0.8592,
+      "step": 11340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.263446047602379e-07,
+      "loss": 0.7923,
+      "step": 11341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.259146688771462e-07,
+      "loss": 0.8981,
+      "step": 11342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2548546111743497e-07,
+      "loss": 0.8954,
+      "step": 11343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2505698151275448e-07,
+      "loss": 0.8115,
+      "step": 11344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.246292300946994e-07,
+      "loss": 0.9049,
+      "step": 11345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2420220689481566e-07,
+      "loss": 0.7757,
+      "step": 11346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2377591194458915e-07,
+      "loss": 0.8197,
+      "step": 11347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2335034527545698e-07,
+      "loss": 0.8229,
+      "step": 11348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2292550691880068e-07,
+      "loss": 0.8403,
+      "step": 11349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2250139690594742e-07,
+      "loss": 0.8857,
+      "step": 11350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2207801526817332e-07,
+      "loss": 0.7909,
+      "step": 11351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.216553620366967e-07,
+      "loss": 0.8885,
+      "step": 11352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2123343724268487e-07,
+      "loss": 0.8044,
+      "step": 11353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2081224091725185e-07,
+      "loss": 0.8741,
+      "step": 11354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2039177309145723e-07,
+      "loss": 0.9055,
+      "step": 11355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1997203379630507e-07,
+      "loss": 0.8412,
+      "step": 11356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.195530230627473e-07,
+      "loss": 0.8439,
+      "step": 11357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1913474092168364e-07,
+      "loss": 0.2911,
+      "step": 11358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1871718740395832e-07,
+      "loss": 0.877,
+      "step": 11359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1830036254036115e-07,
+      "loss": 0.9719,
+      "step": 11360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1788426636162976e-07,
+      "loss": 0.898,
+      "step": 11361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1746889889844738e-07,
+      "loss": 0.8995,
+      "step": 11362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1705426018144284e-07,
+      "loss": 0.9066,
+      "step": 11363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1664035024119281e-07,
+      "loss": 0.2515,
+      "step": 11364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1622716910822063e-07,
+      "loss": 0.7891,
+      "step": 11365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1581471681299084e-07,
+      "loss": 0.8185,
+      "step": 11366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1540299338592242e-07,
+      "loss": 0.937,
+      "step": 11367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1499199885737222e-07,
+      "loss": 0.9419,
+      "step": 11368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1458173325765043e-07,
+      "loss": 0.2326,
+      "step": 11369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.141721966170084e-07,
+      "loss": 0.8846,
+      "step": 11370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1376338896564643e-07,
+      "loss": 0.851,
+      "step": 11371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.133553103337115e-07,
+      "loss": 0.8045,
+      "step": 11372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1294796075129399e-07,
+      "loss": 0.8391,
+      "step": 11373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1254134024843211e-07,
+      "loss": 0.9163,
+      "step": 11374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1213544885511186e-07,
+      "loss": 0.8501,
+      "step": 11375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1173028660126373e-07,
+      "loss": 0.8326,
+      "step": 11376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1132585351676273e-07,
+      "loss": 0.915,
+      "step": 11377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1092214963143388e-07,
+      "loss": 0.8507,
+      "step": 11378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1051917497504671e-07,
+      "loss": 0.8562,
+      "step": 11379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1011692957731635e-07,
+      "loss": 0.8463,
+      "step": 11380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.097154134679046e-07,
+      "loss": 0.8699,
+      "step": 11381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0931462667642001e-07,
+      "loss": 0.811,
+      "step": 11382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0891456923241673e-07,
+      "loss": 0.7719,
+      "step": 11383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0851524116539669e-07,
+      "loss": 0.8934,
+      "step": 11384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0811664250480413e-07,
+      "loss": 0.8445,
+      "step": 11385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0771877328003444e-07,
+      "loss": 0.863,
+      "step": 11386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0732163352042413e-07,
+      "loss": 0.8967,
+      "step": 11387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0692522325526088e-07,
+      "loss": 0.8866,
+      "step": 11388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0652954251377578e-07,
+      "loss": 0.8591,
+      "step": 11389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0613459132514658e-07,
+      "loss": 0.8672,
+      "step": 11390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0574036971849777e-07,
+      "loss": 0.8591,
+      "step": 11391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0534687772289831e-07,
+      "loss": 0.9515,
+      "step": 11392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.04954115367365e-07,
+      "loss": 0.2447,
+      "step": 11393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0456208268086132e-07,
+      "loss": 0.8614,
+      "step": 11394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0417077969229528e-07,
+      "loss": 0.8421,
+      "step": 11395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0378020643052378e-07,
+      "loss": 0.893,
+      "step": 11396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0339036292434602e-07,
+      "loss": 0.8576,
+      "step": 11397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0300124920250898e-07,
+      "loss": 0.8678,
+      "step": 11398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.026128652937064e-07,
+      "loss": 0.8494,
+      "step": 11399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.022252112265798e-07,
+      "loss": 0.7495,
+      "step": 11400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.018382870297141e-07,
+      "loss": 0.897,
+      "step": 11401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0145209273164091e-07,
+      "loss": 0.8132,
+      "step": 11402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0106662836083969e-07,
+      "loss": 0.7954,
+      "step": 11403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0068189394573436e-07,
+      "loss": 0.8696,
+      "step": 11404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0029788951469443e-07,
+      "loss": 0.9145,
+      "step": 11405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99146150960395e-08,
+      "loss": 0.8484,
+      "step": 11406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.953207071802918e-08,
+      "loss": 0.8315,
+      "step": 11407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.915025640887421e-08,
+      "loss": 0.9287,
+      "step": 11408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.876917219672988e-08,
+      "loss": 0.243,
+      "step": 11409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.838881810969814e-08,
+      "loss": 0.9477,
+      "step": 11410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.800919417582543e-08,
+      "loss": 0.8862,
+      "step": 11411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.763030042310717e-08,
+      "loss": 0.8286,
+      "step": 11412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.725213687948099e-08,
+      "loss": 0.9422,
+      "step": 11413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.687470357283568e-08,
+      "loss": 0.9255,
+      "step": 11414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.649800053100233e-08,
+      "loss": 0.9005,
+      "step": 11415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.612202778175761e-08,
+      "loss": 0.9143,
+      "step": 11416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.574678535282933e-08,
+      "loss": 0.8561,
+      "step": 11417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.537227327188647e-08,
+      "loss": 0.8521,
+      "step": 11418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.499849156654473e-08,
+      "loss": 0.8789,
+      "step": 11419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.462544026436871e-08,
+      "loss": 0.9707,
+      "step": 11420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.425311939286752e-08,
+      "loss": 0.7393,
+      "step": 11421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.388152897949588e-08,
+      "loss": 0.8631,
+      "step": 11422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.351066905165518e-08,
+      "loss": 0.8558,
+      "step": 11423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.314053963669245e-08,
+      "loss": 0.8216,
+      "step": 11424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.277114076190252e-08,
+      "loss": 0.8894,
+      "step": 11425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.240247245452472e-08,
+      "loss": 0.8406,
+      "step": 11426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.203453474174506e-08,
+      "loss": 0.8743,
+      "step": 11427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.166732765069408e-08,
+      "loss": 0.781,
+      "step": 11428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.130085120845234e-08,
+      "loss": 0.8582,
+      "step": 11429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.093510544204264e-08,
+      "loss": 0.8472,
+      "step": 11430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.057009037843566e-08,
+      "loss": 0.9427,
+      "step": 11431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.020580604454765e-08,
+      "loss": 0.8745,
+      "step": 11432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.984225246724265e-08,
+      "loss": 0.8624,
+      "step": 11433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.947942967332701e-08,
+      "loss": 0.8904,
+      "step": 11434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.911733768955821e-08,
+      "loss": 0.8629,
+      "step": 11435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.87559765426349e-08,
+      "loss": 0.8945,
+      "step": 11436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.839534625920576e-08,
+      "loss": 0.7783,
+      "step": 11437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.803544686586284e-08,
+      "loss": 0.845,
+      "step": 11438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.767627838914716e-08,
+      "loss": 0.7899,
+      "step": 11439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.731784085554195e-08,
+      "loss": 0.8593,
+      "step": 11440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.696013429147943e-08,
+      "loss": 0.8241,
+      "step": 11441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.660315872333735e-08,
+      "loss": 0.795,
+      "step": 11442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.624691417744024e-08,
+      "loss": 0.822,
+      "step": 11443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.589140068005708e-08,
+      "loss": 0.8638,
+      "step": 11444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.553661825740356e-08,
+      "loss": 0.8269,
+      "step": 11445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.518256693564097e-08,
+      "loss": 0.7471,
+      "step": 11446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.482924674087955e-08,
+      "loss": 0.9064,
+      "step": 11447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.447665769917179e-08,
+      "loss": 0.8142,
+      "step": 11448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.4124799836518e-08,
+      "loss": 0.8801,
+      "step": 11449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.37736731788652e-08,
+      "loss": 0.819,
+      "step": 11450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.34232777521038e-08,
+      "loss": 0.9149,
+      "step": 11451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.307361358207422e-08,
+      "loss": 0.8312,
+      "step": 11452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.27246806945614e-08,
+      "loss": 0.8363,
+      "step": 11453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.237647911529368e-08,
+      "loss": 0.869,
+      "step": 11454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.202900886995047e-08,
+      "loss": 0.8745,
+      "step": 11455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.16822699841524e-08,
+      "loss": 0.825,
+      "step": 11456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.133626248346905e-08,
+      "loss": 0.9344,
+      "step": 11457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.099098639341552e-08,
+      "loss": 0.8585,
+      "step": 11458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.064644173945258e-08,
+      "loss": 0.931,
+      "step": 11459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.030262854698656e-08,
+      "loss": 0.8774,
+      "step": 11460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.995954684137163e-08,
+      "loss": 0.8906,
+      "step": 11461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.961719664790646e-08,
+      "loss": 0.2696,
+      "step": 11462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.927557799183527e-08,
+      "loss": 0.7973,
+      "step": 11463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.893469089835126e-08,
+      "loss": 0.92,
+      "step": 11464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.859453539259099e-08,
+      "loss": 0.8424,
+      "step": 11465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.82551114996366e-08,
+      "loss": 0.9072,
+      "step": 11466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.791641924451809e-08,
+      "loss": 0.8133,
+      "step": 11467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.757845865221325e-08,
+      "loss": 0.7855,
+      "step": 11468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.724122974763992e-08,
+      "loss": 0.8143,
+      "step": 11469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.690473255566822e-08,
+      "loss": 0.8948,
+      "step": 11470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.65689671011105e-08,
+      "loss": 0.8253,
+      "step": 11471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.623393340872586e-08,
+      "loss": 0.8723,
+      "step": 11472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.589963150322232e-08,
+      "loss": 0.8555,
+      "step": 11473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.556606140924905e-08,
+      "loss": 0.8008,
+      "step": 11474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.523322315140524e-08,
+      "loss": 0.2065,
+      "step": 11475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.490111675423461e-08,
+      "loss": 0.7933,
+      "step": 11476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.456974224222646e-08,
+      "loss": 0.8886,
+      "step": 11477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.423909963981568e-08,
+      "loss": 0.8306,
+      "step": 11478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.390918897138721e-08,
+      "loss": 0.9144,
+      "step": 11479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.358001026126604e-08,
+      "loss": 0.843,
+      "step": 11480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.325156353372609e-08,
+      "loss": 0.8918,
+      "step": 11481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.292384881298908e-08,
+      "loss": 0.7738,
+      "step": 11482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.259686612322015e-08,
+      "loss": 0.8254,
+      "step": 11483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.227061548853109e-08,
+      "loss": 0.94,
+      "step": 11484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.194509693297935e-08,
+      "loss": 0.7424,
+      "step": 11485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.162031048057128e-08,
+      "loss": 0.9205,
+      "step": 11486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.129625615525437e-08,
+      "loss": 0.8474,
+      "step": 11487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.097293398092509e-08,
+      "loss": 0.9467,
+      "step": 11488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.065034398142767e-08,
+      "loss": 0.785,
+      "step": 11489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.032848618054644e-08,
+      "loss": 0.8374,
+      "step": 11490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.000736060201796e-08,
+      "loss": 0.8726,
+      "step": 11491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.968696726952217e-08,
+      "loss": 0.8358,
+      "step": 11492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.936730620668464e-08,
+      "loss": 0.8792,
+      "step": 11493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.904837743707759e-08,
+      "loss": 0.8812,
+      "step": 11494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.873018098422002e-08,
+      "loss": 0.871,
+      "step": 11495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.841271687157425e-08,
+      "loss": 0.8062,
+      "step": 11496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.809598512255045e-08,
+      "loss": 0.8505,
+      "step": 11497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.77799857605066e-08,
+      "loss": 0.8234,
+      "step": 11498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.746471880874405e-08,
+      "loss": 0.8857,
+      "step": 11499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.715018429050868e-08,
+      "loss": 0.7683,
+      "step": 11500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.683638222899746e-08,
+      "loss": 0.9073,
+      "step": 11501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.652331264734968e-08,
+      "loss": 0.8101,
+      "step": 11502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.621097556864908e-08,
+      "loss": 0.8665,
+      "step": 11503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.589937101593169e-08,
+      "loss": 0.7754,
+      "step": 11504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.558849901217135e-08,
+      "loss": 0.8454,
+      "step": 11505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.527835958029416e-08,
+      "loss": 0.9107,
+      "step": 11506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.496895274317072e-08,
+      "loss": 0.8156,
+      "step": 11507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.466027852361611e-08,
+      "loss": 0.9064,
+      "step": 11508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.435233694439103e-08,
+      "loss": 0.7903,
+      "step": 11509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.404512802820617e-08,
+      "loss": 0.8159,
+      "step": 11510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.373865179771233e-08,
+      "loss": 0.7956,
+      "step": 11511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.343290827551141e-08,
+      "loss": 0.9082,
+      "step": 11512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.312789748414871e-08,
+      "loss": 0.8289,
+      "step": 11513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.282361944611514e-08,
+      "loss": 0.8794,
+      "step": 11514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.252007418384831e-08,
+      "loss": 0.8512,
+      "step": 11515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.221726171973364e-08,
+      "loss": 0.8272,
+      "step": 11516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.191518207609992e-08,
+      "loss": 1.0259,
+      "step": 11517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.161383527522269e-08,
+      "loss": 0.8907,
+      "step": 11518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.131322133932305e-08,
+      "loss": 0.8846,
+      "step": 11519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.101334029056882e-08,
+      "loss": 0.8406,
+      "step": 11520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.071419215107344e-08,
+      "loss": 0.9494,
+      "step": 11521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.041577694289702e-08,
+      "loss": 0.925,
+      "step": 11522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.01180946880442e-08,
+      "loss": 0.844,
+      "step": 11523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.982114540846629e-08,
+      "loss": 0.8241,
+      "step": 11524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.952492912606134e-08,
+      "loss": 0.8114,
+      "step": 11525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9229445862671875e-08,
+      "loss": 0.7831,
+      "step": 11526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.893469564008714e-08,
+      "loss": 0.9071,
+      "step": 11527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.864067848004307e-08,
+      "loss": 0.8525,
+      "step": 11528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.834739440421899e-08,
+      "loss": 0.8197,
+      "step": 11529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.805484343424317e-08,
+      "loss": 0.2248,
+      "step": 11530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.776302559168945e-08,
+      "loss": 0.9259,
+      "step": 11531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.747194089807506e-08,
+      "loss": 0.9714,
+      "step": 11532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7181589374865065e-08,
+      "loss": 0.8748,
+      "step": 11533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6891971043470106e-08,
+      "loss": 0.7777,
+      "step": 11534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6603085925248656e-08,
+      "loss": 0.8569,
+      "step": 11535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6314934041501455e-08,
+      "loss": 0.8643,
+      "step": 11536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.602751541347817e-08,
+      "loss": 0.8386,
+      "step": 11537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5740830062371855e-08,
+      "loss": 0.8968,
+      "step": 11538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.545487800932447e-08,
+      "loss": 0.279,
+      "step": 11539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.516965927542251e-08,
+      "loss": 0.8358,
+      "step": 11540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.488517388169579e-08,
+      "loss": 0.8656,
+      "step": 11541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4601421849125316e-08,
+      "loss": 0.933,
+      "step": 11542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.431840319863435e-08,
+      "loss": 0.8922,
+      "step": 11543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.403611795109176e-08,
+      "loss": 0.8666,
+      "step": 11544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.375456612731533e-08,
+      "loss": 0.7927,
+      "step": 11545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.347374774806513e-08,
+      "loss": 0.8413,
+      "step": 11546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.319366283405014e-08,
+      "loss": 0.8884,
+      "step": 11547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.291431140592385e-08,
+      "loss": 0.8858,
+      "step": 11548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.263569348428643e-08,
+      "loss": 0.7933,
+      "step": 11549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2357809089681466e-08,
+      "loss": 0.8972,
+      "step": 11550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2080658242602556e-08,
+      "loss": 0.9213,
+      "step": 11551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1804240963486685e-08,
+      "loss": 0.7862,
+      "step": 11552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.152855727271533e-08,
+      "loss": 0.8929,
+      "step": 11553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1253607190618896e-08,
+      "loss": 0.9341,
+      "step": 11554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.097939073747338e-08,
+      "loss": 0.8732,
+      "step": 11555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.070590793349817e-08,
+      "loss": 0.2284,
+      "step": 11556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.043315879886046e-08,
+      "loss": 0.8472,
+      "step": 11557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.016114335367306e-08,
+      "loss": 0.8854,
+      "step": 11558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9889861617995474e-08,
+      "loss": 0.8269,
+      "step": 11559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.96193136118317e-08,
+      "loss": 0.9218,
+      "step": 11560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934949935513134e-08,
+      "loss": 0.7467,
+      "step": 11561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.90804188677918e-08,
+      "loss": 0.8364,
+      "step": 11562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8812072169655e-08,
+      "loss": 0.9023,
+      "step": 11563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.854445928050955e-08,
+      "loss": 0.7979,
+      "step": 11564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.827758022008744e-08,
+      "loss": 0.7923,
+      "step": 11565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8011435008071816e-08,
+      "loss": 0.8711,
+      "step": 11566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.774602366408476e-08,
+      "loss": 0.8815,
+      "step": 11567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7481346207700616e-08,
+      "loss": 0.76,
+      "step": 11568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7217402658437105e-08,
+      "loss": 0.8397,
+      "step": 11569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.695419303575532e-08,
+      "loss": 0.9333,
+      "step": 11570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6691717359066416e-08,
+      "loss": 0.958,
+      "step": 11571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.642997564772489e-08,
+      "loss": 0.8002,
+      "step": 11572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.616896792103198e-08,
+      "loss": 0.8633,
+      "step": 11573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5908694198234515e-08,
+      "loss": 0.7061,
+      "step": 11574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.564915449852603e-08,
+      "loss": 0.8568,
+      "step": 11575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.539034884104454e-08,
+      "loss": 0.7656,
+      "step": 11576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.513227724487368e-08,
+      "loss": 0.8365,
+      "step": 11577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4874939729044884e-08,
+      "loss": 0.7897,
+      "step": 11578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4618336312535203e-08,
+      "loss": 0.8191,
+      "step": 11579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.436246701426505e-08,
+      "loss": 0.8766,
+      "step": 11580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.410733185310379e-08,
+      "loss": 0.8276,
+      "step": 11581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3852930847865236e-08,
+      "loss": 0.8232,
+      "step": 11582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.359926401730774e-08,
+      "loss": 0.9581,
+      "step": 11583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.334633138013855e-08,
+      "loss": 0.2231,
+      "step": 11584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3094132955008304e-08,
+      "loss": 0.7992,
+      "step": 11585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.284266876051435e-08,
+      "loss": 0.8743,
+      "step": 11586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2591938815200736e-08,
+      "loss": 0.9271,
+      "step": 11587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.234194313755491e-08,
+      "loss": 0.8639,
+      "step": 11588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2092681746013217e-08,
+      "loss": 0.8518,
+      "step": 11589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.184415465895542e-08,
+      "loss": 0.8866,
+      "step": 11590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.159636189470906e-08,
+      "loss": 0.8018,
+      "step": 11591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1349303471546196e-08,
+      "loss": 0.9061,
+      "step": 11592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1102979407684486e-08,
+      "loss": 0.7476,
+      "step": 11593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0857389721288274e-08,
+      "loss": 0.8096,
+      "step": 11594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.061253443046864e-08,
+      "loss": 0.8526,
+      "step": 11595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0368413553280026e-08,
+      "loss": 0.8896,
+      "step": 11596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.01250271077247e-08,
+      "loss": 0.8458,
+      "step": 11597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.988237511175053e-08,
+      "loss": 0.2109,
+      "step": 11598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.964045758324986e-08,
+      "loss": 0.9035,
+      "step": 11599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.939927454006287e-08,
+      "loss": 0.8737,
+      "step": 11600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.915882599997423e-08,
+      "loss": 0.2743,
+      "step": 11601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8919111980716404e-08,
+      "loss": 0.9001,
+      "step": 11602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.868013249996305e-08,
+      "loss": 0.84,
+      "step": 11603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.844188757533895e-08,
+      "loss": 0.8989,
+      "step": 11604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.820437722441117e-08,
+      "loss": 0.8547,
+      "step": 11605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.796760146469458e-08,
+      "loss": 0.8015,
+      "step": 11606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7731560313649665e-08,
+      "loss": 0.7906,
+      "step": 11607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.749625378868138e-08,
+      "loss": 0.8211,
+      "step": 11608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.726168190714252e-08,
+      "loss": 0.8251,
+      "step": 11609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.702784468632925e-08,
+      "loss": 0.8189,
+      "step": 11610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.679474214348555e-08,
+      "loss": 0.8038,
+      "step": 11611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6562374295801005e-08,
+      "loss": 0.8498,
+      "step": 11612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.63307411604108e-08,
+      "loss": 0.8407,
+      "step": 11613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.609984275439571e-08,
+      "loss": 0.9515,
+      "step": 11614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.58696790947799e-08,
+      "loss": 0.9176,
+      "step": 11615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5640250198539785e-08,
+      "loss": 0.835,
+      "step": 11616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5411556082590726e-08,
+      "loss": 0.7886,
+      "step": 11617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.518359676379812e-08,
+      "loss": 0.9719,
+      "step": 11618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.495637225897075e-08,
+      "loss": 0.8674,
+      "step": 11619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4729882584866313e-08,
+      "loss": 0.8242,
+      "step": 11620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.450412775818479e-08,
+      "loss": 0.9042,
+      "step": 11621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427910779557397e-08,
+      "loss": 0.9629,
+      "step": 11622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4054822713627256e-08,
+      "loss": 0.8718,
+      "step": 11623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383127252888363e-08,
+      "loss": 0.8171,
+      "step": 11624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3608457257826574e-08,
+      "loss": 0.8173,
+      "step": 11625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.338637691688851e-08,
+      "loss": 0.8438,
+      "step": 11626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3165031522444105e-08,
+      "loss": 0.8425,
+      "step": 11627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.294442109081697e-08,
+      "loss": 0.8403,
+      "step": 11628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2724545638275206e-08,
+      "loss": 0.8806,
+      "step": 11629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.25054051810314e-08,
+      "loss": 0.8816,
+      "step": 11630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.228699973524596e-08,
+      "loss": 0.8105,
+      "step": 11631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2069329317023776e-08,
+      "loss": 0.9257,
+      "step": 11632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1852393942416456e-08,
+      "loss": 0.8332,
+      "step": 11633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.163619362742121e-08,
+      "loss": 0.8191,
+      "step": 11634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.142072838798083e-08,
+      "loss": 0.8792,
+      "step": 11635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1205998239982606e-08,
+      "loss": 0.8478,
+      "step": 11636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.099200319926277e-08,
+      "loss": 0.8764,
+      "step": 11637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.07787432815998e-08,
+      "loss": 0.8591,
+      "step": 11638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.056621850272112e-08,
+      "loss": 0.8785,
+      "step": 11639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0354428878297535e-08,
+      "loss": 0.8467,
+      "step": 11640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.014337442394766e-08,
+      "loss": 0.8112,
+      "step": 11641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.993305515523348e-08,
+      "loss": 0.9473,
+      "step": 11642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9723471087664825e-08,
+      "loss": 0.8086,
+      "step": 11643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.951462223669599e-08,
+      "loss": 0.7948,
+      "step": 11644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9306508617729102e-08,
+      "loss": 0.8347,
+      "step": 11645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9099130246108553e-08,
+      "loss": 0.9514,
+      "step": 11646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.889248713712878e-08,
+      "loss": 0.9392,
+      "step": 11647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8686579306026475e-08,
+      "loss": 0.8903,
+      "step": 11648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8481406767986164e-08,
+      "loss": 0.8901,
+      "step": 11649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8276969538136855e-08,
+      "loss": 0.82,
+      "step": 11650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8073267631553157e-08,
+      "loss": 0.8704,
+      "step": 11651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7870301063258608e-08,
+      "loss": 0.9003,
+      "step": 11652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7668069848217905e-08,
+      "loss": 0.8508,
+      "step": 11653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7466574001344672e-08,
+      "loss": 0.8477,
+      "step": 11654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7265813537497023e-08,
+      "loss": 0.8983,
+      "step": 11655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7065788471480893e-08,
+      "loss": 0.8454,
+      "step": 11656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6866498818043374e-08,
+      "loss": 0.839,
+      "step": 11657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.66679445918816e-08,
+      "loss": 0.8103,
+      "step": 11658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6470125807638303e-08,
+      "loss": 0.9061,
+      "step": 11659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.627304247989848e-08,
+      "loss": 0.8747,
+      "step": 11660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6076694623197173e-08,
+      "loss": 0.825,
+      "step": 11661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.58810822520128e-08,
+      "loss": 0.8307,
+      "step": 11662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.568620538076938e-08,
+      "loss": 0.8621,
+      "step": 11663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5492064023838748e-08,
+      "loss": 0.8586,
+      "step": 11664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.529865819553501e-08,
+      "loss": 0.821,
+      "step": 11665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5105987910121198e-08,
+      "loss": 0.8684,
+      "step": 11666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4914053181804843e-08,
+      "loss": 0.8741,
+      "step": 11667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4722854024739062e-08,
+      "loss": 0.9041,
+      "step": 11668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.453239045302369e-08,
+      "loss": 0.8718,
+      "step": 11669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.434266248070305e-08,
+      "loss": 0.7492,
+      "step": 11670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.415367012176817e-08,
+      "loss": 0.8714,
+      "step": 11671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.396541339015568e-08,
+      "loss": 0.8172,
+      "step": 11672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3777892299746697e-08,
+      "loss": 0.7959,
+      "step": 11673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.359110686437127e-08,
+      "loss": 0.8505,
+      "step": 11674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.340505709780172e-08,
+      "loss": 0.9395,
+      "step": 11675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.321974301375707e-08,
+      "loss": 0.9208,
+      "step": 11676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.303516462590305e-08,
+      "loss": 0.7762,
+      "step": 11677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2851321947851e-08,
+      "loss": 0.8051,
+      "step": 11678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2668214993157857e-08,
+      "loss": 0.791,
+      "step": 11679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2485843775325034e-08,
+      "loss": 0.8067,
+      "step": 11680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2304208307801777e-08,
+      "loss": 0.9018,
+      "step": 11681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.212330860398071e-08,
+      "loss": 0.7451,
+      "step": 11682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1943144677202265e-08,
+      "loss": 0.9179,
+      "step": 11683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.176371654075249e-08,
+      "loss": 0.7757,
+      "step": 11684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1585024207863016e-08,
+      "loss": 0.8256,
+      "step": 11685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.140706769170886e-08,
+      "loss": 0.8711,
+      "step": 11686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1229847005413973e-08,
+      "loss": 0.8739,
+      "step": 11687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1053362162045678e-08,
+      "loss": 0.9122,
+      "step": 11688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0877613174619117e-08,
+      "loss": 0.9304,
+      "step": 11689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0702600056093925e-08,
+      "loss": 0.9036,
+      "step": 11690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0528322819376445e-08,
+      "loss": 0.8699,
+      "step": 11691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.03547814773164e-08,
+      "loss": 0.2661,
+      "step": 11692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0181976042712438e-08,
+      "loss": 0.2146,
+      "step": 11693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0009906528306587e-08,
+      "loss": 0.7968,
+      "step": 11694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9838572946787593e-08,
+      "loss": 0.8433,
+      "step": 11695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9667975310788677e-08,
+      "loss": 0.8599,
+      "step": 11696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9498113632892003e-08,
+      "loss": 0.9069,
+      "step": 11697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.932898792562199e-08,
+      "loss": 0.7745,
+      "step": 11698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9160598201449776e-08,
+      "loss": 0.814,
+      "step": 11699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8992944472793206e-08,
+      "loss": 0.8343,
+      "step": 11700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8826026752014615e-08,
+      "loss": 0.7658,
+      "step": 11701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8659845051423043e-08,
+      "loss": 0.8274,
+      "step": 11702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.849439938327424e-08,
+      "loss": 0.8196,
+      "step": 11703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8329689759766233e-08,
+      "loss": 0.8872,
+      "step": 11704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8165716193044858e-08,
+      "loss": 0.8179,
+      "step": 11705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8002478695203772e-08,
+      "loss": 0.791,
+      "step": 11706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7839977278277798e-08,
+      "loss": 0.8191,
+      "step": 11707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.767821195425179e-08,
+      "loss": 0.8838,
+      "step": 11708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7517182735052875e-08,
+      "loss": 0.8769,
+      "step": 11709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7356889632557107e-08,
+      "loss": 0.8037,
+      "step": 11710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.719733265858392e-08,
+      "loss": 0.2655,
+      "step": 11711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7038511824898353e-08,
+      "loss": 0.8971,
+      "step": 11712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.688042714321325e-08,
+      "loss": 0.8216,
+      "step": 11713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6723078625184853e-08,
+      "loss": 0.8161,
+      "step": 11714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6566466282417203e-08,
+      "loss": 0.9188,
+      "step": 11715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6410590126457736e-08,
+      "loss": 0.9205,
+      "step": 11716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.625545016880281e-08,
+      "loss": 0.8071,
+      "step": 11717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6101046420889945e-08,
+      "loss": 0.8804,
+      "step": 11718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.59473788941078e-08,
+      "loss": 0.2115,
+      "step": 11719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.579444759978621e-08,
+      "loss": 0.8273,
+      "step": 11720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.564225254920282e-08,
+      "loss": 0.9816,
+      "step": 11721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5490793753580867e-08,
+      "loss": 0.8259,
+      "step": 11722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5340071224088095e-08,
+      "loss": 0.8407,
+      "step": 11723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5190084971841157e-08,
+      "loss": 0.8394,
+      "step": 11724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.504083500789788e-08,
+      "loss": 0.2704,
+      "step": 11725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4892321343265015e-08,
+      "loss": 0.8101,
+      "step": 11726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4744543988893801e-08,
+      "loss": 0.9254,
+      "step": 11727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4597502955682185e-08,
+      "loss": 0.8593,
+      "step": 11728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.44511982544715e-08,
+      "loss": 0.2298,
+      "step": 11729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4305629896051998e-08,
+      "loss": 0.8885,
+      "step": 11730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4160797891157319e-08,
+      "loss": 0.8419,
+      "step": 11731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4016702250467806e-08,
+      "loss": 0.8934,
+      "step": 11732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3873342984609406e-08,
+      "loss": 0.7925,
+      "step": 11733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.373072010415255e-08,
+      "loss": 0.8681,
+      "step": 11734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3588833619615493e-08,
+      "loss": 0.8831,
+      "step": 11735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3447683541459866e-08,
+      "loss": 0.8182,
+      "step": 11736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.330726988009623e-08,
+      "loss": 0.8416,
+      "step": 11737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3167592645876304e-08,
+      "loss": 0.9525,
+      "step": 11738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3028651849100737e-08,
+      "loss": 0.8664,
+      "step": 11739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.289044750001578e-08,
+      "loss": 0.8288,
+      "step": 11740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2752979608813275e-08,
+      "loss": 0.9085,
+      "step": 11741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2616248185629565e-08,
+      "loss": 0.8911,
+      "step": 11742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.248025324054658e-08,
+      "loss": 0.8816,
+      "step": 11743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2344994783592968e-08,
+      "loss": 0.7511,
+      "step": 11744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2210472824744079e-08,
+      "loss": 0.8348,
+      "step": 11745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.207668737391865e-08,
+      "loss": 0.8725,
+      "step": 11746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1943638440982119e-08,
+      "loss": 0.8843,
+      "step": 11747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.181132603574553e-08,
+      "loss": 0.822,
+      "step": 11748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.167975016796552e-08,
+      "loss": 0.8973,
+      "step": 11749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.154891084734655e-08,
+      "loss": 0.9293,
+      "step": 11750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1418808083534238e-08,
+      "loss": 0.7839,
+      "step": 11751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1289441886123132e-08,
+      "loss": 0.787,
+      "step": 11752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1160812264654486e-08,
+      "loss": 0.9114,
+      "step": 11753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1032919228610718e-08,
+      "loss": 0.8151,
+      "step": 11754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.090576278742539e-08,
+      "loss": 0.8119,
+      "step": 11755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0779342950474336e-08,
+      "loss": 0.8663,
+      "step": 11756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0653659727078991e-08,
+      "loss": 0.9354,
+      "step": 11757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0528713126508605e-08,
+      "loss": 0.8354,
+      "step": 11758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0404503157975809e-08,
+      "loss": 0.9164,
+      "step": 11759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0281029830639944e-08,
+      "loss": 0.8224,
+      "step": 11760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0158293153607057e-08,
+      "loss": 0.2318,
+      "step": 11761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0036293135926579e-08,
+      "loss": 0.8893,
+      "step": 11762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.915029786595754e-09,
+      "loss": 0.7761,
+      "step": 11763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.794503114556319e-09,
+      "loss": 0.8785,
+      "step": 11764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.674713128696722e-09,
+      "loss": 0.7575,
+      "step": 11765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.555659837849896e-09,
+      "loss": 0.7864,
+      "step": 11766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.437343250794373e-09,
+      "loss": 0.8445,
+      "step": 11767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.319763376255397e-09,
+      "loss": 0.8319,
+      "step": 11768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.2029202229027e-09,
+      "loss": 0.8923,
+      "step": 11769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.086813799353834e-09,
+      "loss": 0.8122,
+      "step": 11770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.971444114169725e-09,
+      "loss": 0.8184,
+      "step": 11771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.856811175858015e-09,
+      "loss": 0.7781,
+      "step": 11772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.742914992871942e-09,
+      "loss": 0.8839,
+      "step": 11773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.629755573609233e-09,
+      "loss": 0.7869,
+      "step": 11774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.517332926415434e-09,
+      "loss": 0.8832,
+      "step": 11775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.40564705958058e-09,
+      "loss": 0.814,
+      "step": 11776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.294697981341415e-09,
+      "loss": 0.8007,
+      "step": 11777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.184485699876954e-09,
+      "loss": 0.7635,
+      "step": 11778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.075010223317359e-09,
+      "loss": 0.8654,
+      "step": 11779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.966271559732842e-09,
+      "loss": 0.2286,
+      "step": 11780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.858269717142542e-09,
+      "loss": 0.8105,
+      "step": 11781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.7510047035112e-09,
+      "loss": 0.8616,
+      "step": 11782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.644476526749157e-09,
+      "loss": 0.8317,
+      "step": 11783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.538685194710126e-09,
+      "loss": 0.8287,
+      "step": 11784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.43363071519676e-09,
+      "loss": 0.8222,
+      "step": 11785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.329313095955081e-09,
+      "loss": 0.7348,
+      "step": 11786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.225732344677827e-09,
+      "loss": 0.8584,
+      "step": 11787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.122888469003331e-09,
+      "loss": 0.8728,
+      "step": 11788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0207814765155255e-09,
+      "loss": 0.7755,
+      "step": 11789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.919411374742835e-09,
+      "loss": 0.8941,
+      "step": 11790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8187781711614995e-09,
+      "loss": 0.9139,
+      "step": 11791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.718881873192251e-09,
+      "loss": 0.8266,
+      "step": 11792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.619722488200308e-09,
+      "loss": 0.7548,
+      "step": 11793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.521300023498711e-09,
+      "loss": 0.8348,
+      "step": 11794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4236144863460966e-09,
+      "loss": 0.903,
+      "step": 11795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.326665883944483e-09,
+      "loss": 0.8181,
+      "step": 11796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.230454223442595e-09,
+      "loss": 0.2825,
+      "step": 11797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.134979511936978e-09,
+      "loss": 0.8914,
+      "step": 11798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.040241756466447e-09,
+      "loss": 0.8652,
+      "step": 11799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.946240964016526e-09,
+      "loss": 0.839,
+      "step": 11800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.852977141521665e-09,
+      "loss": 0.8461,
+      "step": 11801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.760450295856368e-09,
+      "loss": 0.7241,
+      "step": 11802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6686604338451745e-09,
+      "loss": 0.8475,
+      "step": 11803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5776075622560044e-09,
+      "loss": 0.9473,
+      "step": 11804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.487291687804597e-09,
+      "loss": 0.915,
+      "step": 11805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.397712817148959e-09,
+      "loss": 0.8228,
+      "step": 11806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3088709568949185e-09,
+      "loss": 0.7809,
+      "step": 11807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.22076611359501e-09,
+      "loss": 0.8531,
+      "step": 11808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133398293746261e-09,
+      "loss": 0.7658,
+      "step": 11809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0467675037901845e-09,
+      "loss": 0.7982,
+      "step": 11810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.960873750115003e-09,
+      "loss": 0.8645,
+      "step": 11811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8757170390556495e-09,
+      "loss": 0.8956,
+      "step": 11812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.791297376890436e-09,
+      "loss": 0.8035,
+      "step": 11813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.707614769844382e-09,
+      "loss": 0.9475,
+      "step": 11814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.624669224090328e-09,
+      "loss": 0.7804,
+      "step": 11815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.542460745742272e-09,
+      "loss": 0.8134,
+      "step": 11816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.460989340864253e-09,
+      "loss": 0.2605,
+      "step": 11817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380255015463686e-09,
+      "loss": 0.8209,
+      "step": 11818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.300257775492478e-09,
+      "loss": 0.8858,
+      "step": 11819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.220997626851464e-09,
+      "loss": 0.8419,
+      "step": 11820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1424745753848585e-09,
+      "loss": 0.2388,
+      "step": 11821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0646886268824736e-09,
+      "loss": 0.7574,
+      "step": 11822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.987639787080833e-09,
+      "loss": 0.7381,
+      "step": 11823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.911328061660946e-09,
+      "loss": 0.8249,
+      "step": 11824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.835753456251645e-09,
+      "loss": 0.881,
+      "step": 11825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7609159764240285e-09,
+      "loss": 0.7823,
+      "step": 11826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.686815627697016e-09,
+      "loss": 0.8148,
+      "step": 11827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6134524155362338e-09,
+      "loss": 0.8389,
+      "step": 11828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5408263453495796e-09,
+      "loss": 0.7937,
+      "step": 11829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.468937422494989e-09,
+      "loss": 0.8857,
+      "step": 11830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3977856522704466e-09,
+      "loss": 0.8859,
+      "step": 11831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3273710399261972e-09,
+      "loss": 0.2531,
+      "step": 11832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2576935906514227e-09,
+      "loss": 0.8345,
+      "step": 11833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1887533095864563e-09,
+      "loss": 0.8368,
+      "step": 11834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1205502018150092e-09,
+      "loss": 0.7793,
+      "step": 11835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.053084272364171e-09,
+      "loss": 0.8353,
+      "step": 11836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9863555262121813e-09,
+      "loss": 0.8865,
+      "step": 11837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9203639682773287e-09,
+      "loss": 0.8316,
+      "step": 11838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.855109603425721e-09,
+      "loss": 0.9771,
+      "step": 11839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7905924364712844e-09,
+      "loss": 0.9299,
+      "step": 11840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7268124721691046e-09,
+      "loss": 0.872,
+      "step": 11841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.663769715224307e-09,
+      "loss": 0.8164,
+      "step": 11842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.601464170285395e-09,
+      "loss": 0.8397,
+      "step": 11843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5398958419464713e-09,
+      "loss": 0.8716,
+      "step": 11844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4790647347472384e-09,
+      "loss": 0.8694,
+      "step": 11845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4189708531741073e-09,
+      "loss": 0.8981,
+      "step": 11846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.359614201657978e-09,
+      "loss": 0.8903,
+      "step": 11847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.30099478457535e-09,
+      "loss": 0.7446,
+      "step": 11848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2431126062505416e-09,
+      "loss": 0.8094,
+      "step": 11849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.18596767095125e-09,
+      "loss": 0.8753,
+      "step": 11850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1295599828896617e-09,
+      "loss": 0.8494,
+      "step": 11851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.073889546228003e-09,
+      "loss": 0.8332,
+      "step": 11852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0189563650696574e-09,
+      "loss": 0.8177,
+      "step": 11853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9647604434658297e-09,
+      "loss": 0.8519,
+      "step": 11854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.911301785413322e-09,
+      "loss": 0.2636,
+      "step": 11855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8585803948534265e-09,
+      "loss": 0.8182,
+      "step": 11856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8065962756752542e-09,
+      "loss": 0.244,
+      "step": 11857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7553494317101848e-09,
+      "loss": 0.9223,
+      "step": 11858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7048398667385279e-09,
+      "loss": 0.9032,
+      "step": 11859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6550675844850816e-09,
+      "loss": 0.8901,
+      "step": 11860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6060325886202433e-09,
+      "loss": 0.8684,
+      "step": 11861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.557734882758899e-09,
+      "loss": 0.8447,
+      "step": 11862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5101744704626442e-09,
+      "loss": 0.9851,
+      "step": 11863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4633513552397838e-09,
+      "loss": 0.867,
+      "step": 11864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.417265540542001e-09,
+      "loss": 0.9491,
+      "step": 11865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3719170297676888e-09,
+      "loss": 0.8417,
+      "step": 11866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3273058262619488e-09,
+      "loss": 0.9547,
+      "step": 11867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2834319333132617e-09,
+      "loss": 0.2803,
+      "step": 11868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2402953541568174e-09,
+      "loss": 0.8944,
+      "step": 11869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1978960919745152e-09,
+      "loss": 0.2607,
+      "step": 11870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1562341498927431e-09,
+      "loss": 0.8408,
+      "step": 11871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1153095309834883e-09,
+      "loss": 0.8489,
+      "step": 11872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0751222382643368e-09,
+      "loss": 0.8326,
+      "step": 11873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.035672274699584e-09,
+      "loss": 0.9181,
+      "step": 11874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.969596431969043e-10,
+      "loss": 0.8683,
+      "step": 11875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.58984346611791e-10,
+      "loss": 0.8348,
+      "step": 11876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.217463877442268e-10,
+      "loss": 0.7022,
+      "step": 11877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.852457693409033e-10,
+      "loss": 0.8678,
+      "step": 11878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.494824940930013e-10,
+      "loss": 0.8167,
+      "step": 11879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.144565646373004e-10,
+      "loss": 0.8989,
+      "step": 11880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.801679835572895e-10,
+      "loss": 0.9446,
+      "step": 11881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.466167533820567e-10,
+      "loss": 0.8301,
+      "step": 11882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.138028765840688e-10,
+      "loss": 0.9035,
+      "step": 11883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.817263555836118e-10,
+      "loss": 0.8494,
+      "step": 11884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.50387192746571e-10,
+      "loss": 0.8131,
+      "step": 11885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.197853903833206e-10,
+      "loss": 0.2879,
+      "step": 11886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.89920950750944e-10,
+      "loss": 0.8426,
+      "step": 11887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.607938760510134e-10,
+      "loss": 0.8419,
+      "step": 11888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.324041684329206e-10,
+      "loss": 0.8169,
+      "step": 11889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.047518299883258e-10,
+      "loss": 0.9203,
+      "step": 11890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.778368627567087e-10,
+      "loss": 0.8416,
+      "step": 11891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.51659268723148e-10,
+      "loss": 0.7966,
+      "step": 11892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2621904981832163e-10,
+      "loss": 0.8785,
+      "step": 11893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0151620791850643e-10,
+      "loss": 0.9223,
+      "step": 11894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.775507448444682e-10,
+      "loss": 0.8742,
+      "step": 11895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5432266236257174e-10,
+      "loss": 0.8978,
+      "step": 11896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3183196218811164e-10,
+      "loss": 0.8336,
+      "step": 11897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.100786459775407e-10,
+      "loss": 0.8756,
+      "step": 11898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.890627153362413e-10,
+      "loss": 0.8649,
+      "step": 11899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6878417181297465e-10,
+      "loss": 0.7971,
+      "step": 11900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4924301690321117e-10,
+      "loss": 0.8588,
+      "step": 11901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3043925204913054e-10,
+      "loss": 0.743,
+      "step": 11902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1237287863518086e-10,
+      "loss": 0.7524,
+      "step": 11903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.950438979958502e-10,
+      "loss": 0.8322,
+      "step": 11904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.784523114078951e-10,
+      "loss": 0.8419,
+      "step": 11905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6259812009367103e-10,
+      "loss": 0.8903,
+      "step": 11906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4748132522446334e-10,
+      "loss": 0.9752,
+      "step": 11907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.331019279138257e-10,
+      "loss": 0.9207,
+      "step": 11908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1945992922313132e-10,
+      "loss": 0.8615,
+      "step": 11909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0655533015713204e-10,
+      "loss": 0.874,
+      "step": 11910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.4388131667289e-11,
+      "loss": 0.7758,
+      "step": 11911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.29583346517726e-11,
+      "loss": 0.8367,
+      "step": 11912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.226593995213193e-11,
+      "loss": 0.8238,
+      "step": 11913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.231094835884577e-11,
+      "loss": 0.8109,
+      "step": 11914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.309336060355108e-11,
+      "loss": 0.8463,
+      "step": 11915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.461317736903503e-11,
+      "loss": 0.82,
+      "step": 11916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.687039927702252e-11,
+      "loss": 0.8008,
+      "step": 11917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9865026901498836e-11,
+      "loss": 0.7851,
+      "step": 11918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3597060757607482e-11,
+      "loss": 0.9531,
+      "step": 11919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8066501308311445e-11,
+      "loss": 0.7637,
+      "step": 11920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3273348959952358e-11,
+      "loss": 0.8193,
+      "step": 11921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.217604067801589e-12,
+      "loss": 0.8132,
+      "step": 11922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.899266929398906e-12,
+      "loss": 0.8727,
+      "step": 11923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3183377901035983e-12,
+      "loss": 0.8435,
+      "step": 11924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.474816840874027e-12,
+      "loss": 0.7938,
+      "step": 11925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.687042171574007e-13,
+      "loss": 0.7975,
+      "step": 11926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8705,
+      "step": 11927
+    },
+    {
+      "epoch": 1.0,
+      "step": 11927,
+      "total_flos": 2.833373821808725e+19,
+      "train_loss": 0.8688710359752633,
+      "train_runtime": 77804.331,
+      "train_samples_per_second": 9.811,
+      "train_steps_per_second": 0.153
+    }
+  ],
+  "max_steps": 11927,
+  "num_train_epochs": 1,
+  "total_flos": 2.833373821808725e+19,
+  "trial_name": null,
+  "trial_params": null
+}