diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35646 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5936,
+  "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": 0.8372,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-07,
+      "loss": 0.9813,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6759776536312851e-07,
+      "loss": 0.8639,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 0.9027,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-07,
+      "loss": 0.2611,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 0.8814,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.910614525139665e-07,
+      "loss": 0.9604,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 0.7981,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.027932960893855e-07,
+      "loss": 0.8927,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 0.8781,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.145251396648045e-07,
+      "loss": 0.8246,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 0.8745,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.262569832402236e-07,
+      "loss": 0.9361,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 0.9154,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.379888268156425e-07,
+      "loss": 0.8861,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 0.2338,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.497206703910615e-07,
+      "loss": 0.8899,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 0.8171,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0614525139664806e-06,
+      "loss": 0.2623,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 0.861,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1731843575418997e-06,
+      "loss": 0.8181,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 0.9151,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2849162011173185e-06,
+      "loss": 0.9688,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 0.8898,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3966480446927375e-06,
+      "loss": 0.2692,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 0.8308,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5083798882681566e-06,
+      "loss": 0.8426,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 0.8928,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6201117318435754e-06,
+      "loss": 0.9258,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 0.7913,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7318435754189947e-06,
+      "loss": 0.2527,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 0.8003,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8435754189944135e-06,
+      "loss": 0.7251,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 0.7933,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9553072625698325e-06,
+      "loss": 0.8265,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 0.8305,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.067039106145252e-06,
+      "loss": 0.7757,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 0.8047,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1787709497206706e-06,
+      "loss": 0.8263,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 0.6814,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2905027932960895e-06,
+      "loss": 0.7304,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 0.8075,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4022346368715087e-06,
+      "loss": 0.7267,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 0.6903,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5139664804469276e-06,
+      "loss": 0.7463,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 0.7264,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6256983240223464e-06,
+      "loss": 0.7357,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 0.7618,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7374301675977656e-06,
+      "loss": 0.7439,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 0.7703,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8491620111731845e-06,
+      "loss": 0.7919,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 0.8032,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9608938547486037e-06,
+      "loss": 0.8093,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 0.7392,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0726256983240226e-06,
+      "loss": 0.7238,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 0.8571,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1843575418994414e-06,
+      "loss": 0.8063,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 0.7844,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2960893854748607e-06,
+      "loss": 0.7604,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 0.6913,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4078212290502795e-06,
+      "loss": 0.7229,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 0.2506,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5195530726256988e-06,
+      "loss": 0.7476,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 0.7302,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6312849162011176e-06,
+      "loss": 0.709,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 0.7985,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7430167597765364e-06,
+      "loss": 0.736,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 0.7291,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.854748603351956e-06,
+      "loss": 0.7553,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 0.6718,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9664804469273745e-06,
+      "loss": 0.6604,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 0.7317,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.078212290502794e-06,
+      "loss": 0.2337,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.763,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189944134078213e-06,
+      "loss": 0.6907,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 0.7104,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.301675977653632e-06,
+      "loss": 0.6625,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 0.7638,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.413407821229051e-06,
+      "loss": 0.7212,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 0.7165,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5251396648044695e-06,
+      "loss": 0.6953,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 0.2215,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.636871508379888e-06,
+      "loss": 0.7308,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 0.7448,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748603351955308e-06,
+      "loss": 0.7573,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 0.2485,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.860335195530727e-06,
+      "loss": 0.7119,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 0.6454,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972067039106146e-06,
+      "loss": 0.6728,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 0.7255,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0837988826815645e-06,
+      "loss": 0.6712,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 0.6793,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.195530726256983e-06,
+      "loss": 0.6614,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 0.2138,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.307262569832403e-06,
+      "loss": 0.7002,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 0.7338,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.418994413407822e-06,
+      "loss": 0.6581,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 0.7506,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.530726256983241e-06,
+      "loss": 0.6682,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 0.213,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6424581005586595e-06,
+      "loss": 0.7344,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 0.6627,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.754189944134078e-06,
+      "loss": 0.6484,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 0.7036,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.865921787709497e-06,
+      "loss": 0.6773,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 0.6762,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.977653631284917e-06,
+      "loss": 0.7198,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 0.6308,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.089385474860336e-06,
+      "loss": 0.679,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 0.7591,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2011173184357546e-06,
+      "loss": 0.6459,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 0.6221,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.312849162011173e-06,
+      "loss": 0.7172,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 0.645,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.424581005586593e-06,
+      "loss": 0.642,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 0.6999,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.536312849162012e-06,
+      "loss": 0.6529,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 0.7111,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.648044692737431e-06,
+      "loss": 0.6731,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 0.6729,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.7597765363128496e-06,
+      "loss": 0.652,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 0.6342,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.871508379888268e-06,
+      "loss": 0.6724,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 0.6229,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.983240223463687e-06,
+      "loss": 0.6755,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 0.643,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.094972067039106e-06,
+      "loss": 0.6384,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 0.745,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.206703910614526e-06,
+      "loss": 0.6727,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 0.6299,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.318435754189945e-06,
+      "loss": 0.7408,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 0.6978,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.430167597765364e-06,
+      "loss": 0.6773,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 0.6911,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.541899441340783e-06,
+      "loss": 0.6299,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 0.6957,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.653631284916202e-06,
+      "loss": 0.7081,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 0.6509,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.76536312849162e-06,
+      "loss": 0.6079,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 0.6376,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.87709497206704e-06,
+      "loss": 0.683,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 0.7167,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.988826815642458e-06,
+      "loss": 0.6471,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 0.6804,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.100558659217877e-06,
+      "loss": 0.6451,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 0.6749,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.212290502793296e-06,
+      "loss": 0.7009,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 0.6437,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.324022346368715e-06,
+      "loss": 0.5927,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 0.5734,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.435754189944135e-06,
+      "loss": 0.6633,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 0.648,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.547486033519554e-06,
+      "loss": 0.7006,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 0.6495,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.659217877094973e-06,
+      "loss": 0.6636,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 0.6639,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.770949720670392e-06,
+      "loss": 0.6256,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 0.7309,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.88268156424581e-06,
+      "loss": 0.5865,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 0.6879,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.99441340782123e-06,
+      "loss": 0.6263,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 0.6824,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.106145251396648e-06,
+      "loss": 0.758,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 0.6724,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.217877094972067e-06,
+      "loss": 0.6545,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 0.6674,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.329608938547486e-06,
+      "loss": 0.6627,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 0.6591,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.441340782122905e-06,
+      "loss": 0.6353,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 0.659,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.553072625698325e-06,
+      "loss": 0.6504,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 0.6264,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.664804469273744e-06,
+      "loss": 0.6621,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 0.6121,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.776536312849163e-06,
+      "loss": 0.6385,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 0.6658,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.888268156424582e-06,
+      "loss": 0.6652,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.2411,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.6153,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999925552986e-06,
+      "loss": 0.6719,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997022119655e-06,
+      "loss": 0.7099,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993299770054e-06,
+      "loss": 0.6735,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999988088482164e-06,
+      "loss": 0.609,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999981388257538e-06,
+      "loss": 0.7609,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999973199098173e-06,
+      "loss": 0.7434,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999963521006501e-06,
+      "loss": 0.2048,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999952353985412e-06,
+      "loss": 0.2252,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999939698038226e-06,
+      "loss": 0.7004,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999925553168714e-06,
+      "loss": 0.6321,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999909919381086e-06,
+      "loss": 0.6738,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999892796680002e-06,
+      "loss": 0.5901,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999874185070556e-06,
+      "loss": 0.69,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999854084558294e-06,
+      "loss": 0.6745,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999832495149197e-06,
+      "loss": 0.6661,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999809416849701e-06,
+      "loss": 0.6196,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999784849666674e-06,
+      "loss": 0.6943,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99975879360743e-06,
+      "loss": 0.6268,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999731248679734e-06,
+      "loss": 0.7573,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999702214891783e-06,
+      "loss": 0.5896,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999671692252226e-06,
+      "loss": 0.6929,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999639680770151e-06,
+      "loss": 0.6541,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999606180455092e-06,
+      "loss": 0.624,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999571191317023e-06,
+      "loss": 0.6547,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999534713366365e-06,
+      "loss": 0.2321,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99949674661398e-06,
+      "loss": 0.6578,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999457291071174e-06,
+      "loss": 0.6167,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999416346749695e-06,
+      "loss": 0.6601,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99937391366174e-06,
+      "loss": 0.6655,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999329991819942e-06,
+      "loss": 0.567,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999284581237378e-06,
+      "loss": 0.6628,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999237681927576e-06,
+      "loss": 0.6276,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999189293904499e-06,
+      "loss": 0.6573,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999139417182557e-06,
+      "loss": 0.6826,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999088051776603e-06,
+      "loss": 0.6779,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999035197701931e-06,
+      "loss": 0.6026,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998980854974283e-06,
+      "loss": 0.6302,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99892502360984e-06,
+      "loss": 0.74,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998867703625228e-06,
+      "loss": 0.6553,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998808895037518e-06,
+      "loss": 0.6588,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99874859786422e-06,
+      "loss": 0.722,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998686812123292e-06,
+      "loss": 0.6815,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998623537833132e-06,
+      "loss": 0.6609,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998558775012581e-06,
+      "loss": 0.6807,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998492523680927e-06,
+      "loss": 0.6455,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998424783857896e-06,
+      "loss": 0.6663,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998355555563663e-06,
+      "loss": 0.6208,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998284838818844e-06,
+      "loss": 0.6551,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998212633644494e-06,
+      "loss": 0.5499,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998138940062116e-06,
+      "loss": 0.594,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998063758093657e-06,
+      "loss": 0.6679,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997987087761503e-06,
+      "loss": 0.609,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997908929088486e-06,
+      "loss": 0.6648,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997829282097884e-06,
+      "loss": 0.6252,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99774814681341e-06,
+      "loss": 0.6009,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997665523259228e-06,
+      "loss": 0.5674,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99758141145994e-06,
+      "loss": 0.6215,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997495811440597e-06,
+      "loss": 0.6812,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997408723226687e-06,
+      "loss": 0.7049,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997320146844145e-06,
+      "loss": 0.7039,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997230082319346e-06,
+      "loss": 0.6393,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997138529679114e-06,
+      "loss": 0.7121,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99704548895071e-06,
+      "loss": 0.6341,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99695096016184e-06,
+      "loss": 0.6372,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996854943340653e-06,
+      "loss": 0.7049,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996757438515743e-06,
+      "loss": 0.6259,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996658445716145e-06,
+      "loss": 0.6426,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996557964971339e-06,
+      "loss": 0.6367,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996455996311246e-06,
+      "loss": 0.6874,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99635253976623e-06,
+      "loss": 0.6538,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996247595367102e-06,
+      "loss": 0.5922,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99614116314511e-06,
+      "loss": 0.6121,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99603324313195e-06,
+      "loss": 0.6494,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99592383535976e-06,
+      "loss": 0.6426,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99581293986112e-06,
+      "loss": 0.6121,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995700556669052e-06,
+      "loss": 0.6231,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995586685817025e-06,
+      "loss": 0.706,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995471327338944e-06,
+      "loss": 0.7068,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995354481269165e-06,
+      "loss": 0.6804,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995236147642482e-06,
+      "loss": 0.6654,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995116326494134e-06,
+      "loss": 0.5368,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994995017859801e-06,
+      "loss": 0.6496,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99487222177561e-06,
+      "loss": 0.6946,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994747938278124e-06,
+      "loss": 0.6727,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994622167404357e-06,
+      "loss": 0.6072,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994494909191761e-06,
+      "loss": 0.6718,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99436616367823e-06,
+      "loss": 0.6911,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994235930902108e-06,
+      "loss": 0.6677,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994104210902168e-06,
+      "loss": 0.6779,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993971003717645e-06,
+      "loss": 0.6777,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993836309388199e-06,
+      "loss": 0.6092,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993700127953943e-06,
+      "loss": 0.6713,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993562459455429e-06,
+      "loss": 0.6749,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993423303933654e-06,
+      "loss": 0.5899,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993282661430058e-06,
+      "loss": 0.6476,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993140531986522e-06,
+      "loss": 0.6035,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992996915645367e-06,
+      "loss": 0.6292,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992851812449366e-06,
+      "loss": 0.6462,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992705222441725e-06,
+      "loss": 0.6949,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9925571456661e-06,
+      "loss": 0.1909,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992407582166582e-06,
+      "loss": 0.6679,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992256531987712e-06,
+      "loss": 0.6121,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992103995174472e-06,
+      "loss": 0.6367,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991949971772282e-06,
+      "loss": 0.6144,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991794461827013e-06,
+      "loss": 0.1947,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991637465384969e-06,
+      "loss": 0.6277,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991478982492906e-06,
+      "loss": 0.1993,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991319013198016e-06,
+      "loss": 0.5729,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991157557547937e-06,
+      "loss": 0.6725,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990994615590747e-06,
+      "loss": 0.6558,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99083018737497e-06,
+      "loss": 0.6776,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990664272949571e-06,
+      "loss": 0.6242,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990496872363953e-06,
+      "loss": 0.6575,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990327985667972e-06,
+      "loss": 0.6093,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990157612911918e-06,
+      "loss": 0.6216,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989985754146525e-06,
+      "loss": 0.6434,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98981240942297e-06,
+      "loss": 0.5892,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989637578792874e-06,
+      "loss": 0.6537,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9894612623083e-06,
+      "loss": 0.6251,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989283460021755e-06,
+      "loss": 0.6194,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98910417198618e-06,
+      "loss": 0.5899,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988923398254971e-06,
+      "loss": 0.5883,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988741138881958e-06,
+      "loss": 0.5245,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988557393921415e-06,
+      "loss": 0.5871,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988372163428059e-06,
+      "loss": 0.7335,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988185447457053e-06,
+      "loss": 0.6585,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987997246063994e-06,
+      "loss": 0.6511,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987807559304927e-06,
+      "loss": 0.5443,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987616387236341e-06,
+      "loss": 0.1692,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987423729915163e-06,
+      "loss": 0.6,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987229587398764e-06,
+      "loss": 0.5996,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987033959744958e-06,
+      "loss": 0.6331,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986836847012001e-06,
+      "loss": 0.6168,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986638249258589e-06,
+      "loss": 0.6814,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986438166543864e-06,
+      "loss": 0.2044,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986236598927406e-06,
+      "loss": 0.606,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986033546469241e-06,
+      "loss": 0.633,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985829009229835e-06,
+      "loss": 0.6857,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985622987270099e-06,
+      "loss": 0.6218,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98541548065138e-06,
+      "loss": 0.6669,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985206489435473e-06,
+      "loss": 0.6816,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984996013684614e-06,
+      "loss": 0.6096,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98478405346148e-06,
+      "loss": 0.6187,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984570608829189e-06,
+      "loss": 0.6707,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9843556798513e-06,
+      "loss": 0.635,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984139266591824e-06,
+      "loss": 0.6829,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983921369115197e-06,
+      "loss": 0.6415,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983701987486312e-06,
+      "loss": 0.2064,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983481121770498e-06,
+      "loss": 0.6658,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983258772033523e-06,
+      "loss": 0.6789,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983034938341602e-06,
+      "loss": 0.6468,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98280962076139e-06,
+      "loss": 0.6166,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982582819359987e-06,
+      "loss": 0.7287,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982354534204925e-06,
+      "loss": 0.7034,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98212476536419e-06,
+      "loss": 0.6882,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981893512906203e-06,
+      "loss": 0.5949,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981660776899826e-06,
+      "loss": 0.5825,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981426557414368e-06,
+      "loss": 0.6624,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981190854519576e-06,
+      "loss": 0.6477,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98095366828564e-06,
+      "loss": 0.6056,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98071499878319e-06,
+      "loss": 0.6239,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980474846083297e-06,
+      "loss": 0.6746,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98023321025748e-06,
+      "loss": 0.6511,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979990091377694e-06,
+      "loss": 0.6208,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979745489516334e-06,
+      "loss": 0.6035,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979499404746243e-06,
+      "loss": 0.6501,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9792518371407e-06,
+      "loss": 0.6439,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979002786773429e-06,
+      "loss": 0.6588,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978752253718593e-06,
+      "loss": 0.7245,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9785002380508e-06,
+      "loss": 0.5713,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978246739845095e-06,
+      "loss": 0.6466,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977991759176966e-06,
+      "loss": 0.58,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977735296122344e-06,
+      "loss": 0.625,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977477350757603e-06,
+      "loss": 0.6106,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977217923159554e-06,
+      "loss": 0.6459,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976957013405452e-06,
+      "loss": 0.6505,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976694621572991e-06,
+      "loss": 0.6236,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976430747740312e-06,
+      "loss": 0.6135,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976165391985988e-06,
+      "loss": 0.6222,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975898554389044e-06,
+      "loss": 0.6057,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97563023502894e-06,
+      "loss": 0.6569,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975360433985576e-06,
+      "loss": 0.6952,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975089151339294e-06,
+      "loss": 0.6097,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974816387170885e-06,
+      "loss": 0.6752,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974542141561571e-06,
+      "loss": 0.6727,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97426641459302e-06,
+      "loss": 0.5922,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973989206347337e-06,
+      "loss": 0.607,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973710516907075e-06,
+      "loss": 0.1886,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973430346355225e-06,
+      "loss": 0.5318,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973148694775217e-06,
+      "loss": 0.6253,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972865562250923e-06,
+      "loss": 0.7282,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972580948866655e-06,
+      "loss": 0.6781,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97229485470717e-06,
+      "loss": 0.6661,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972007279857663e-06,
+      "loss": 0.6077,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971718224403769e-06,
+      "loss": 0.6405,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971427688431567e-06,
+      "loss": 0.6542,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971135672027574e-06,
+      "loss": 0.6987,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970842175278751e-06,
+      "loss": 0.6573,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970547198272496e-06,
+      "loss": 0.5499,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970250741096649e-06,
+      "loss": 0.5993,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969952803839493e-06,
+      "loss": 0.6798,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969653386589749e-06,
+      "loss": 0.6237,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969352489436579e-06,
+      "loss": 0.64,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96905011246959e-06,
+      "loss": 0.6493,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968746255778823e-06,
+      "loss": 0.6468,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968440919454764e-06,
+      "loss": 0.6806,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96813410358834e-06,
+      "loss": 0.6379,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967825808270915e-06,
+      "loss": 0.6571,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967516033594295e-06,
+      "loss": 0.647,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96720477965073e-06,
+      "loss": 0.6199,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966892046532906e-06,
+      "loss": 0.5798,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96657783433395e-06,
+      "loss": 0.1919,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966262143147434e-06,
+      "loss": 0.5974,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965944973067365e-06,
+      "loss": 0.6137,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96562632418819e-06,
+      "loss": 0.5933,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965306196604805e-06,
+      "loss": 0.652,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964984590412534e-06,
+      "loss": 0.5879,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964661505707151e-06,
+      "loss": 0.5699,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964336942584866e-06,
+      "loss": 0.6398,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96401090114233e-06,
+      "loss": 0.6734,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963683381476634e-06,
+      "loss": 0.627,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96335438368531e-06,
+      "loss": 0.6885,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963023907866329e-06,
+      "loss": 0.6429,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962691954118102e-06,
+      "loss": 0.6908,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962358522539482e-06,
+      "loss": 0.6043,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962023613229762e-06,
+      "loss": 0.6412,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961687226288671e-06,
+      "loss": 0.6194,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961349361816384e-06,
+      "loss": 0.6122,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961010019913512e-06,
+      "loss": 0.6332,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960669200681104e-06,
+      "loss": 0.6738,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960326904220656e-06,
+      "loss": 0.6314,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9599831306341e-06,
+      "loss": 0.6973,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959637880023805e-06,
+      "loss": 0.6039,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959291152492583e-06,
+      "loss": 0.6144,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958942948143686e-06,
+      "loss": 0.6493,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958593267080805e-06,
+      "loss": 0.6013,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958242109408072e-06,
+      "loss": 0.5858,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957889475230056e-06,
+      "loss": 0.6413,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957535364651766e-06,
+      "loss": 0.5928,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957179777778655e-06,
+      "loss": 0.707,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956822714716612e-06,
+      "loss": 0.6974,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956464175571965e-06,
+      "loss": 0.5492,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956104160451481e-06,
+      "loss": 0.6078,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95574266946237e-06,
+      "loss": 0.5975,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955379702712281e-06,
+      "loss": 0.6702,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9550152603093e-06,
+      "loss": 0.623,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954649342361952e-06,
+      "loss": 0.6328,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954281948979207e-06,
+      "loss": 0.6295,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953913080270465e-06,
+      "loss": 0.5564,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953542736345576e-06,
+      "loss": 0.6515,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953170917314821e-06,
+      "loss": 0.6244,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952797623288924e-06,
+      "loss": 0.6107,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952422854379045e-06,
+      "loss": 0.6168,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95204661069679e-06,
+      "loss": 0.6741,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951668892354198e-06,
+      "loss": 0.5343,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951289699463749e-06,
+      "loss": 0.5863,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950909032138362e-06,
+      "loss": 0.6317,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950526890491394e-06,
+      "loss": 0.1739,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950143274636646e-06,
+      "loss": 0.6363,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949758184688349e-06,
+      "loss": 0.6095,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949371620761181e-06,
+      "loss": 0.6213,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948983582970258e-06,
+      "loss": 0.689,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94859407143113e-06,
+      "loss": 0.7003,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94820308625979e-06,
+      "loss": 0.5761,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947810627572668e-06,
+      "loss": 0.6034,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947416695486633e-06,
+      "loss": 0.6149,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947021290118997e-06,
+      "loss": 0.6524,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946624411587502e-06,
+      "loss": 0.5935,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946226060010337e-06,
+      "loss": 0.5514,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945826235506127e-06,
+      "loss": 0.2143,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94542493819393e-06,
+      "loss": 0.6327,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945022168193254e-06,
+      "loss": 0.5424,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944617925624033e-06,
+      "loss": 0.7004,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944212210606651e-06,
+      "loss": 0.5892,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943805023261923e-06,
+      "loss": 0.6412,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943396363711104e-06,
+      "loss": 0.5793,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942986232075887e-06,
+      "loss": 0.6421,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942574628478405e-06,
+      "loss": 0.6169,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94216155304123e-06,
+      "loss": 0.5752,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94174700588737e-06,
+      "loss": 0.6483,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941330987140272e-06,
+      "loss": 0.5845,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940913496923822e-06,
+      "loss": 0.6303,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940494535362344e-06,
+      "loss": 0.6077,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940074102580596e-06,
+      "loss": 0.5693,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939652198703785e-06,
+      "loss": 0.6508,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939228823857543e-06,
+      "loss": 0.6542,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938803978167946e-06,
+      "loss": 0.585,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938377661761512e-06,
+      "loss": 0.5627,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93794987476519e-06,
+      "loss": 0.6088,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937520617306369e-06,
+      "loss": 0.6335,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937089889512881e-06,
+      "loss": 0.6097,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936657691512987e-06,
+      "loss": 0.6308,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936224023435393e-06,
+      "loss": 0.5645,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935788885409238e-06,
+      "loss": 0.6828,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935352277564103e-06,
+      "loss": 0.4998,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934914200030006e-06,
+      "loss": 0.6313,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934474652937396e-06,
+      "loss": 0.5896,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93403363641717e-06,
+      "loss": 0.6012,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933591150600655e-06,
+      "loss": 0.5495,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93314719561962e-06,
+      "loss": 0.5542,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932701771606266e-06,
+      "loss": 0.202,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932254878693237e-06,
+      "loss": 0.5667,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931806517013612e-06,
+      "loss": 0.5817,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93135668670091e-06,
+      "loss": 0.6482,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93090538788908e-06,
+      "loss": 0.618,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930452620712521e-06,
+      "loss": 0.6523,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929998385306054e-06,
+      "loss": 0.6297,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92954268180495e-06,
+      "loss": 0.6036,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92908551034491e-06,
+      "loss": 0.5965,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928626871062072e-06,
+      "loss": 0.5862,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928166764093017e-06,
+      "loss": 0.7504,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92770518957476e-06,
+      "loss": 0.5517,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927242147644746e-06,
+      "loss": 0.6505,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926777638440872e-06,
+      "loss": 0.6028,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926311662101457e-06,
+      "loss": 0.6763,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925844218765265e-06,
+      "loss": 0.6401,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925375308571497e-06,
+      "loss": 0.6453,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924904931659787e-06,
+      "loss": 0.6654,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924433088170207e-06,
+      "loss": 0.701,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923959778243267e-06,
+      "loss": 0.6404,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923485002019912e-06,
+      "loss": 0.6411,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923008759641527e-06,
+      "loss": 0.6576,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92253105124993e-06,
+      "loss": 0.6187,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922051876987377e-06,
+      "loss": 0.682,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921571236996562e-06,
+      "loss": 0.5481,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92108913142061e-06,
+      "loss": 0.6038,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92060556040309e-06,
+      "loss": 0.615,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920120524088002e-06,
+      "loss": 0.591,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919634022619783e-06,
+      "loss": 0.5717,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91914605614331e-06,
+      "loss": 0.5437,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91865662480389e-06,
+      "loss": 0.1928,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918165728747273e-06,
+      "loss": 0.5765,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917673368119642e-06,
+      "loss": 0.5817,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917179543067615e-06,
+      "loss": 0.6155,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916684253738246e-06,
+      "loss": 0.63,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916187500279027e-06,
+      "loss": 0.5545,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915689282837886e-06,
+      "loss": 0.5481,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915189601563188e-06,
+      "loss": 0.6332,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914688456603728e-06,
+      "loss": 0.5955,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914185848108743e-06,
+      "loss": 0.6186,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913681776227906e-06,
+      "loss": 0.6133,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91317624111132e-06,
+      "loss": 0.5444,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912669242909528e-06,
+      "loss": 0.5907,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91216078177351e-06,
+      "loss": 0.5914,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911650857854678e-06,
+      "loss": 0.5904,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91113947130488e-06,
+      "loss": 0.6906,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910626622276404e-06,
+      "loss": 0.588,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91011231092197e-06,
+      "loss": 0.1714,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90959653739473e-06,
+      "loss": 0.5585,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909079301848279e-06,
+      "loss": 0.658,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908560604436642e-06,
+      "loss": 0.5883,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908040445314282e-06,
+      "loss": 0.5846,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907518824636094e-06,
+      "loss": 0.1811,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906995742557413e-06,
+      "loss": 0.6042,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906471199234004e-06,
+      "loss": 0.6157,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905945194822073e-06,
+      "loss": 0.6079,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905417729478255e-06,
+      "loss": 0.5938,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904888803359625e-06,
+      "loss": 0.638,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904358416623687e-06,
+      "loss": 0.645,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90382656942839e-06,
+      "loss": 0.6393,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903293261932106e-06,
+      "loss": 0.6419,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90275849429365e-06,
+      "loss": 0.5948,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90222226667227e-06,
+      "loss": 0.6288,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901684579227647e-06,
+      "loss": 0.576,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901145432119898e-06,
+      "loss": 0.5824,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900604825509576e-06,
+      "loss": 0.6417,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900062759557666e-06,
+      "loss": 0.6368,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899519234425588e-06,
+      "loss": 0.6321,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8989742502752e-06,
+      "loss": 0.6174,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89842780726879e-06,
+      "loss": 0.6038,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897879905569078e-06,
+      "loss": 0.5385,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897330545339232e-06,
+      "loss": 0.606,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896779726742838e-06,
+      "loss": 0.6113,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896227449943923e-06,
+      "loss": 0.5316,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895673715106953e-06,
+      "loss": 0.6386,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895118522396818e-06,
+      "loss": 0.5823,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894561871978854e-06,
+      "loss": 0.1861,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89400376401882e-06,
+      "loss": 0.6096,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893444198682914e-06,
+      "loss": 0.1685,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89288317613777e-06,
+      "loss": 0.6173,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892320696550454e-06,
+      "loss": 0.64,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891756760088466e-06,
+      "loss": 0.5852,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891191366919736e-06,
+      "loss": 0.598,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890624517212635e-06,
+      "loss": 0.6704,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890056211135964e-06,
+      "loss": 0.6168,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889486448858953e-06,
+      "loss": 0.6302,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888915230551278e-06,
+      "loss": 0.6564,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888342556383032e-06,
+      "loss": 0.5927,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88776842652476e-06,
+      "loss": 0.6243,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887192841147424e-06,
+      "loss": 0.5495,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88661580042243e-06,
+      "loss": 0.6009,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886037304521612e-06,
+      "loss": 0.6367,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88545735361724e-06,
+      "loss": 0.6414,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884875947882017e-06,
+      "loss": 0.5902,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884293087489078e-06,
+      "loss": 0.6024,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883708772611992e-06,
+      "loss": 0.6682,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88312300342476e-06,
+      "loss": 0.5185,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882535780101818e-06,
+      "loss": 0.6743,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881947102818036e-06,
+      "loss": 0.5608,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881356971748713e-06,
+      "loss": 0.6372,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88076538706958e-06,
+      "loss": 0.6341,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88017234895681e-06,
+      "loss": 0.567,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879577857586998e-06,
+      "loss": 0.6216,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878981913137178e-06,
+      "loss": 0.5495,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878384515784818e-06,
+      "loss": 0.5575,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87778566570781e-06,
+      "loss": 0.5849,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877185363084488e-06,
+      "loss": 0.6338,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876583608093614e-06,
+      "loss": 0.5744,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875980400914384e-06,
+      "loss": 0.5268,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875375741726427e-06,
+      "loss": 0.5948,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874769630709802e-06,
+      "loss": 0.5741,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874162068045e-06,
+      "loss": 0.6168,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873553053912949e-06,
+      "loss": 0.6659,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872942588495003e-06,
+      "loss": 0.6685,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872330671972954e-06,
+      "loss": 0.6474,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871717304529023e-06,
+      "loss": 0.5724,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871102486345862e-06,
+      "loss": 0.1857,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870486217606557e-06,
+      "loss": 0.6197,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869868498494627e-06,
+      "loss": 0.6486,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86924932919402e-06,
+      "loss": 0.6026,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868628709889117e-06,
+      "loss": 0.5899,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868006640764732e-06,
+      "loss": 0.5971,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867383122006109e-06,
+      "loss": 0.6548,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866758153798925e-06,
+      "loss": 0.5527,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866131736329287e-06,
+      "loss": 0.5955,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865503869783734e-06,
+      "loss": 0.6227,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864874554349242e-06,
+      "loss": 0.5828,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864243790213209e-06,
+      "loss": 0.6897,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863611577563469e-06,
+      "loss": 0.6658,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86297791658829e-06,
+      "loss": 0.6121,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862342807476367e-06,
+      "loss": 0.5656,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861706250416828e-06,
+      "loss": 0.5879,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86106824559923e-06,
+      "loss": 0.6276,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860428793213568e-06,
+      "loss": 0.5649,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85978789345026e-06,
+      "loss": 0.5947,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85914554650016e-06,
+      "loss": 0.5946,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858501752554548e-06,
+      "loss": 0.5615,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857856511805143e-06,
+      "loss": 0.5762,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857209824444085e-06,
+      "loss": 0.5193,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856561690663955e-06,
+      "loss": 0.5693,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855912110657754e-06,
+      "loss": 0.2092,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855261084618924e-06,
+      "loss": 0.6163,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854608612741329e-06,
+      "loss": 0.6098,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85395469521927e-06,
+      "loss": 0.6369,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853299332247474e-06,
+      "loss": 0.5715,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852642524021101e-06,
+      "loss": 0.6475,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851984270735743e-06,
+      "loss": 0.5556,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851324572587415e-06,
+      "loss": 0.5214,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850663429772572e-06,
+      "loss": 0.5513,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85000084248809e-06,
+      "loss": 0.5648,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849336810931284e-06,
+      "loss": 0.536,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848671335299891e-06,
+      "loss": 0.6166,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848004415792086e-06,
+      "loss": 0.5927,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847336052606463e-06,
+      "loss": 0.5498,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84666624594206e-06,
+      "loss": 0.6496,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845994995998332e-06,
+      "loss": 0.6014,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845322302975171e-06,
+      "loss": 0.6769,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844648167072897e-06,
+      "loss": 0.6289,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84397258849226e-06,
+      "loss": 0.1877,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843295567434441e-06,
+      "loss": 0.6748,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842617104101045e-06,
+      "loss": 0.5357,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841937198694112e-06,
+      "loss": 0.579,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84125585141611e-06,
+      "loss": 0.6029,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840573062469937e-06,
+      "loss": 0.6653,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839888832058915e-06,
+      "loss": 0.6167,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839203160386805e-06,
+      "loss": 0.5672,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83851604765779e-06,
+      "loss": 0.6319,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837827494076485e-06,
+      "loss": 0.6553,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837137499847929e-06,
+      "loss": 0.6036,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836446065177597e-06,
+      "loss": 0.5502,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83575319027139e-06,
+      "loss": 0.5602,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835058875335639e-06,
+      "loss": 0.5882,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834363120577099e-06,
+      "loss": 0.6248,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833665926202962e-06,
+      "loss": 0.5655,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83296729242084e-06,
+      "loss": 0.6058,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832267219438781e-06,
+      "loss": 0.5652,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831565707465257e-06,
+      "loss": 0.5823,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83086275670917e-06,
+      "loss": 0.5961,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83015836737985e-06,
+      "loss": 0.4468,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829452539687056e-06,
+      "loss": 0.5284,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828745273840974e-06,
+      "loss": 0.5691,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828036570052221e-06,
+      "loss": 0.5998,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82732642853184e-06,
+      "loss": 0.6368,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826614849491305e-06,
+      "loss": 0.5756,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825901833142511e-06,
+      "loss": 0.5747,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825187379697789e-06,
+      "loss": 0.5491,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824471489369892e-06,
+      "loss": 0.5826,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823754162372006e-06,
+      "loss": 0.5491,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823035398917744e-06,
+      "loss": 0.5401,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82231519922114e-06,
+      "loss": 0.6292,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821593563496666e-06,
+      "loss": 0.5287,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820870491959213e-06,
+      "loss": 0.6455,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820145984824106e-06,
+      "loss": 0.5629,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819420042307091e-06,
+      "loss": 0.5777,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818692664624347e-06,
+      "loss": 0.6163,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81796385199248e-06,
+      "loss": 0.5275,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81723360462852e-06,
+      "loss": 0.5621,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816501922749923e-06,
+      "loss": 0.6213,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815768806574582e-06,
+      "loss": 0.5782,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815034256320804e-06,
+      "loss": 0.6017,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814298272207334e-06,
+      "loss": 0.6516,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813560854453335e-06,
+      "loss": 0.1904,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812822003278404e-06,
+      "loss": 0.569,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81208171890256e-06,
+      "loss": 0.5922,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811340001546252e-06,
+      "loss": 0.5948,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810596851430356e-06,
+      "loss": 0.612,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809852268776173e-06,
+      "loss": 0.5922,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809106253805427e-06,
+      "loss": 0.6172,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808358806740275e-06,
+      "loss": 0.6251,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807609927803299e-06,
+      "loss": 0.5634,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806859617217505e-06,
+      "loss": 0.6241,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806107875206327e-06,
+      "loss": 0.5261,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805354701993624e-06,
+      "loss": 0.5459,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804600097803682e-06,
+      "loss": 0.6139,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803844062861213e-06,
+      "loss": 0.5543,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803086597391356e-06,
+      "loss": 0.5882,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802327701619675e-06,
+      "loss": 0.576,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801567375772161e-06,
+      "loss": 0.5553,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800805620075227e-06,
+      "loss": 0.5439,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800042434755717e-06,
+      "loss": 0.5705,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7992778200409e-06,
+      "loss": 0.5822,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798511776158464e-06,
+      "loss": 0.6659,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797744303336533e-06,
+      "loss": 0.5202,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796975401803651e-06,
+      "loss": 0.608,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796205071788783e-06,
+      "loss": 0.1731,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795433313521327e-06,
+      "loss": 0.5897,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794660127231103e-06,
+      "loss": 0.5806,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793885513148358e-06,
+      "loss": 0.5682,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793109471503762e-06,
+      "loss": 0.6183,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79233200252841e-06,
+      "loss": 0.6209,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791553106453825e-06,
+      "loss": 0.5844,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79077278351195e-06,
+      "loss": 0.5616,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789991033935159e-06,
+      "loss": 0.551,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789207857956245e-06,
+      "loss": 0.5684,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78842325580843e-06,
+      "loss": 0.6081,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787637227725359e-06,
+      "loss": 0.5249,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786849773941102e-06,
+      "loss": 0.6321,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786060894690153e-06,
+      "loss": 0.5522,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78527059020743e-06,
+      "loss": 0.6099,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784478860728276e-06,
+      "loss": 0.5462,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783685706488462e-06,
+      "loss": 0.16,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782891127724177e-06,
+      "loss": 0.5994,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782095124672036e-06,
+      "loss": 0.6938,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78129769756908e-06,
+      "loss": 0.6579,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780498846652777e-06,
+      "loss": 0.601,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77969857216101e-06,
+      "loss": 0.6136,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778896874332094e-06,
+      "loss": 0.5822,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778093753404764e-06,
+      "loss": 0.6347,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77728920961818e-06,
+      "loss": 0.5314,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776483243211927e-06,
+      "loss": 0.5374,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77567585442601e-06,
+      "loss": 0.6295,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774867043500861e-06,
+      "loss": 0.6588,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774056810677334e-06,
+      "loss": 0.7168,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773245156196705e-06,
+      "loss": 0.5859,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772432080300679e-06,
+      "loss": 0.6,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771617583231376e-06,
+      "loss": 0.5081,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770801665231346e-06,
+      "loss": 0.5504,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76998432654356e-06,
+      "loss": 0.5688,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76916556741141e-06,
+      "loss": 0.6297,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768345388078713e-06,
+      "loss": 0.6183,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76752378878971e-06,
+      "loss": 0.5893,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766700769789061e-06,
+      "loss": 0.5764,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765876331321855e-06,
+      "loss": 0.6458,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765050473633596e-06,
+      "loss": 0.6043,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764223196970216e-06,
+      "loss": 0.6256,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76339450157807e-06,
+      "loss": 0.5667,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76256438770393e-06,
+      "loss": 0.6484,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761732855594996e-06,
+      "loss": 0.5875,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760899905498889e-06,
+      "loss": 0.5615,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76006553766365e-06,
+      "loss": 0.6319,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759229752337746e-06,
+      "loss": 0.5521,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758392549770061e-06,
+      "loss": 0.6519,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757553930209907e-06,
+      "loss": 0.5709,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75671389390701e-06,
+      "loss": 0.5201,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75587244111153e-06,
+      "loss": 0.6017,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755029572074035e-06,
+      "loss": 0.5564,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754185287045525e-06,
+      "loss": 0.6123,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753339586277417e-06,
+      "loss": 0.5952,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75249247002155e-06,
+      "loss": 0.5472,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751643938530186e-06,
+      "loss": 0.5621,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750793992056008e-06,
+      "loss": 0.6346,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749942630852119e-06,
+      "loss": 0.5811,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749089855172044e-06,
+      "loss": 0.1893,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74823566526973e-06,
+      "loss": 0.568,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747380061399543e-06,
+      "loss": 0.5267,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746523043816274e-06,
+      "loss": 0.5367,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745664612775132e-06,
+      "loss": 0.6432,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744804768531746e-06,
+      "loss": 0.6063,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743943511342168e-06,
+      "loss": 0.6137,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74308084146287e-06,
+      "loss": 0.6511,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742216759150748e-06,
+      "loss": 0.566,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74135126466311e-06,
+      "loss": 0.5406,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740484358257693e-06,
+      "loss": 0.6293,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739616040192653e-06,
+      "loss": 0.6171,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73874631072656e-06,
+      "loss": 0.5978,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737875170118414e-06,
+      "loss": 0.6834,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737002618627627e-06,
+      "loss": 0.611,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736128656514036e-06,
+      "loss": 0.6114,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735253284037896e-06,
+      "loss": 0.5625,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734376501459882e-06,
+      "loss": 0.1795,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733498309041088e-06,
+      "loss": 0.6065,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732618707043033e-06,
+      "loss": 0.6283,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73173769572765e-06,
+      "loss": 0.5901,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730855275357291e-06,
+      "loss": 0.5513,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729971446194736e-06,
+      "loss": 0.6463,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729086208503174e-06,
+      "loss": 0.5718,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72819956254622e-06,
+      "loss": 0.5609,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727311508587907e-06,
+      "loss": 0.6231,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726422046892685e-06,
+      "loss": 0.572,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725531177725427e-06,
+      "loss": 0.6302,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724638901351424e-06,
+      "loss": 0.1542,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723745218036382e-06,
+      "loss": 0.6628,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722850128046431e-06,
+      "loss": 0.6141,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72195363164812e-06,
+      "loss": 0.5621,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72105572910841e-06,
+      "loss": 0.5974,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72015642069469e-06,
+      "loss": 0.5441,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719255706674762e-06,
+      "loss": 0.5606,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718353587316848e-06,
+      "loss": 0.6136,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717450062889589e-06,
+      "loss": 0.6151,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716545133662044e-06,
+      "loss": 0.5929,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715638799903686e-06,
+      "loss": 0.5317,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714731061884415e-06,
+      "loss": 0.6249,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713821919874543e-06,
+      "loss": 0.577,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712911374144802e-06,
+      "loss": 0.6119,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71199942496634e-06,
+      "loss": 0.6246,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711086072610726e-06,
+      "loss": 0.5456,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710171317349946e-06,
+      "loss": 0.6169,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709255159456403e-06,
+      "loss": 0.5648,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708337599202917e-06,
+      "loss": 0.6166,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707418636862727e-06,
+      "loss": 0.5439,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706498272709488e-06,
+      "loss": 0.6364,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705576507017276e-06,
+      "loss": 0.6295,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704653340060577e-06,
+      "loss": 0.5363,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703728772114304e-06,
+      "loss": 0.6192,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702802803453782e-06,
+      "loss": 0.5467,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70187543435475e-06,
+      "loss": 0.485,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700946665093372e-06,
+      "loss": 0.5916,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700016495946218e-06,
+      "loss": 0.5055,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699084927190287e-06,
+      "loss": 0.6203,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698151959102988e-06,
+      "loss": 0.6013,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697217591962144e-06,
+      "loss": 0.5384,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696281826046004e-06,
+      "loss": 0.6215,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695344661633224e-06,
+      "loss": 0.563,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694406099002883e-06,
+      "loss": 0.5404,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69346613843447e-06,
+      "loss": 0.5171,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692524780207897e-06,
+      "loss": 0.6012,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691582024603488e-06,
+      "loss": 0.548,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690637871901985e-06,
+      "loss": 0.5961,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689692322384546e-06,
+      "loss": 0.6585,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688745376332743e-06,
+      "loss": 0.6033,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687797034028567e-06,
+      "loss": 0.6006,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68684729575442e-06,
+      "loss": 0.6387,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685896161793126e-06,
+      "loss": 0.5806,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68494363242792e-06,
+      "loss": 0.6035,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683989707942452e-06,
+      "loss": 0.5721,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683034388620793e-06,
+      "loss": 0.5446,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682077674747422e-06,
+      "loss": 0.6373,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68111956660724e-06,
+      "loss": 0.6051,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680160064485558e-06,
+      "loss": 0.5634,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679199168668107e-06,
+      "loss": 0.567,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678236879441026e-06,
+      "loss": 0.5926,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677273197090878e-06,
+      "loss": 0.5549,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676308121904632e-06,
+      "loss": 0.1697,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675341654169679e-06,
+      "loss": 0.6522,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674373794173818e-06,
+      "loss": 0.5891,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673404542205272e-06,
+      "loss": 0.6352,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672433898552667e-06,
+      "loss": 0.5977,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67146186350505e-06,
+      "loss": 0.5412,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670488437351883e-06,
+      "loss": 0.6449,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669513620383041e-06,
+      "loss": 0.1528,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668537412888812e-06,
+      "loss": 0.5782,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667559815159899e-06,
+      "loss": 0.6097,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666580827487418e-06,
+      "loss": 0.5696,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665600450162901e-06,
+      "loss": 0.166,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664618683478295e-06,
+      "loss": 0.6258,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663635527725954e-06,
+      "loss": 0.5231,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66265098319865e-06,
+      "loss": 0.581,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661665050189572e-06,
+      "loss": 0.5527,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660677728992319e-06,
+      "loss": 0.6074,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6596890199009e-06,
+      "loss": 0.6346,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658698923209743e-06,
+      "loss": 0.5886,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657707439213686e-06,
+      "loss": 0.5552,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656714568207984e-06,
+      "loss": 0.5748,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655720310488298e-06,
+      "loss": 0.6225,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65472466635071e-06,
+      "loss": 0.6082,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653727636091707e-06,
+      "loss": 0.6313,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652729220008195e-06,
+      "loss": 0.5696,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651729418397491e-06,
+      "loss": 0.55,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650728231557322e-06,
+      "loss": 0.6373,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649725659785832e-06,
+      "loss": 0.5858,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648721703381572e-06,
+      "loss": 0.6034,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647716362643509e-06,
+      "loss": 0.5827,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646709637871024e-06,
+      "loss": 0.5606,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645701529363904e-06,
+      "loss": 0.5654,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644692037422353e-06,
+      "loss": 0.5756,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643681162346988e-06,
+      "loss": 0.555,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642668904438834e-06,
+      "loss": 0.5923,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641655263999327e-06,
+      "loss": 0.5356,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640640241330319e-06,
+      "loss": 0.6316,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639623836734071e-06,
+      "loss": 0.1875,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638606050513258e-06,
+      "loss": 0.5763,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637586882970964e-06,
+      "loss": 0.6095,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636566334410682e-06,
+      "loss": 0.5811,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635544405136323e-06,
+      "loss": 0.6184,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634521095452204e-06,
+      "loss": 0.5555,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633496405663051e-06,
+      "loss": 0.6128,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632470336074009e-06,
+      "loss": 0.5742,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63144288699063e-06,
+      "loss": 0.5979,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63041405871887e-06,
+      "loss": 0.5792,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629383851565109e-06,
+      "loss": 0.5827,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628352265836125e-06,
+      "loss": 0.1527,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627319301839114e-06,
+      "loss": 0.5674,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62628495988168e-06,
+      "loss": 0.6233,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625249240271839e-06,
+      "loss": 0.1645,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624212143318013e-06,
+      "loss": 0.5696,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62317366932904e-06,
+      "loss": 0.5415,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622133818614164e-06,
+      "loss": 0.5522,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621092591483041e-06,
+      "loss": 0.5467,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620049988245735e-06,
+      "loss": 0.5492,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619006009212719e-06,
+      "loss": 0.5935,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617960654694881e-06,
+      "loss": 0.5505,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616913925003514e-06,
+      "loss": 0.6424,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615865820450319e-06,
+      "loss": 0.6538,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614816341347412e-06,
+      "loss": 0.5566,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613765488007316e-06,
+      "loss": 0.642,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612713260742961e-06,
+      "loss": 0.6073,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611659659867686e-06,
+      "loss": 0.6019,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610604685695245e-06,
+      "loss": 0.5465,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609548338539792e-06,
+      "loss": 0.1511,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608490618715898e-06,
+      "loss": 0.5721,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607431526538538e-06,
+      "loss": 0.5556,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606371062323096e-06,
+      "loss": 0.6354,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605309226385367e-06,
+      "loss": 0.5982,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604246019041554e-06,
+      "loss": 0.6547,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603181440608265e-06,
+      "loss": 0.5757,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60211549140252e-06,
+      "loss": 0.577,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601048171741747e-06,
+      "loss": 0.1615,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59997948194378e-06,
+      "loss": 0.5914,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59890942232686e-06,
+      "loss": 0.5332,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597837993209642e-06,
+      "loss": 0.5522,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596765194911182e-06,
+      "loss": 0.5167,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595691027750948e-06,
+      "loss": 0.5873,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594615492048814e-06,
+      "loss": 0.5143,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59353858812506e-06,
+      "loss": 0.55,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592460316300375e-06,
+      "loss": 0.6163,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591380676895858e-06,
+      "loss": 0.6204,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590299670233011e-06,
+      "loss": 0.5311,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589217296633748e-06,
+      "loss": 0.5056,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588133556420382e-06,
+      "loss": 0.6515,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58704844991564e-06,
+      "loss": 0.5906,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585961977442654e-06,
+      "loss": 0.5912,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584874139324964e-06,
+      "loss": 0.532,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583784935886513e-06,
+      "loss": 0.6287,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582694367451653e-06,
+      "loss": 0.6852,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581602434345141e-06,
+      "loss": 0.6054,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580509136892146e-06,
+      "loss": 0.5522,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579414475418235e-06,
+      "loss": 0.6198,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578318450249386e-06,
+      "loss": 0.6146,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577221061711982e-06,
+      "loss": 0.5996,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576122310132814e-06,
+      "loss": 0.6043,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575022195839077e-06,
+      "loss": 0.5803,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57392071915837e-06,
+      "loss": 0.5461,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572817880418699e-06,
+      "loss": 0.5815,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571713679948477e-06,
+      "loss": 0.1871,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570608118076525e-06,
+      "loss": 0.5818,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56950119513206e-06,
+      "loss": 0.6698,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568392911444717e-06,
+      "loss": 0.6109,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567283267344526e-06,
+      "loss": 0.601,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566172263161925e-06,
+      "loss": 0.5569,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56505989922776e-06,
+      "loss": 0.654,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563946175873277e-06,
+      "loss": 0.5334,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562831093430134e-06,
+      "loss": 0.5808,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561714652230384e-06,
+      "loss": 0.5704,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560596852606494e-06,
+      "loss": 0.582,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559477694891328e-06,
+      "loss": 0.5692,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55835717941816e-06,
+      "loss": 0.5848,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557235306520667e-06,
+      "loss": 0.5317,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556112076532928e-06,
+      "loss": 0.5979,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554987489789426e-06,
+      "loss": 0.5385,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553861546625052e-06,
+      "loss": 0.1789,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552734247375095e-06,
+      "loss": 0.5964,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551605592375255e-06,
+      "loss": 0.5915,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550475581961633e-06,
+      "loss": 0.1568,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549344216470727e-06,
+      "loss": 0.5695,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548211496239449e-06,
+      "loss": 0.5457,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547077421605106e-06,
+      "loss": 0.6163,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545941992905416e-06,
+      "loss": 0.5431,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544805210478495e-06,
+      "loss": 0.5616,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543667074662859e-06,
+      "loss": 0.5929,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542527585797435e-06,
+      "loss": 0.6095,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541386744221548e-06,
+      "loss": 0.5373,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54024455027493e-06,
+      "loss": 0.5757,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539101004297709e-06,
+      "loss": 0.5526,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53795610663042e-06,
+      "loss": 0.6276,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536809857614e-06,
+      "loss": 0.6558,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53566225758979e-06,
+      "loss": 0.627,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53451330689953e-06,
+      "loss": 0.5164,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533363005885362e-06,
+      "loss": 0.5883,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532211354889837e-06,
+      "loss": 0.5605,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5310583542559e-06,
+      "loss": 0.5861,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5299040043269e-06,
+      "loss": 0.4778,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528748305446586e-06,
+      "loss": 0.5567,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527591257959119e-06,
+      "loss": 0.6296,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526432862209047e-06,
+      "loss": 0.5719,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52527311854133e-06,
+      "loss": 0.5987,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524112027301326e-06,
+      "loss": 0.5419,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522949588834791e-06,
+      "loss": 0.6014,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521785803487888e-06,
+      "loss": 0.5587,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520620671607177e-06,
+      "loss": 0.6034,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519454193539622e-06,
+      "loss": 0.1487,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518286369632586e-06,
+      "loss": 0.1696,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517117200233829e-06,
+      "loss": 0.5507,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515946685691522e-06,
+      "loss": 0.5734,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514774826354225e-06,
+      "loss": 0.6589,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513601622570908e-06,
+      "loss": 0.5667,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512427074690933e-06,
+      "loss": 0.5694,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511251183064068e-06,
+      "loss": 0.6076,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510073948040481e-06,
+      "loss": 0.525,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508895369970736e-06,
+      "loss": 0.6134,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507715449205802e-06,
+      "loss": 0.5702,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506534186097043e-06,
+      "loss": 0.5435,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505351580996227e-06,
+      "loss": 0.6008,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504167634255519e-06,
+      "loss": 0.5458,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502982346227483e-06,
+      "loss": 0.5631,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501795717265085e-06,
+      "loss": 0.6279,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500607747721688e-06,
+      "loss": 0.6051,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499418437951055e-06,
+      "loss": 0.5745,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498227788307352e-06,
+      "loss": 0.5816,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497035799145135e-06,
+      "loss": 0.5772,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495842470819367e-06,
+      "loss": 0.6058,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494647803685405e-06,
+      "loss": 0.5621,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493451798099008e-06,
+      "loss": 0.6159,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492254454416332e-06,
+      "loss": 0.5946,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491055772993933e-06,
+      "loss": 0.5174,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489855754188762e-06,
+      "loss": 0.5831,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48865439835817e-06,
+      "loss": 0.2121,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487451705859909e-06,
+      "loss": 0.6146,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486247677052123e-06,
+      "loss": 0.5257,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48504231229336e-06,
+      "loss": 0.5841,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483835611942562e-06,
+      "loss": 0.5773,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482627576359071e-06,
+      "loss": 0.5969,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481418205902623e-06,
+      "loss": 0.5666,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480207500933359e-06,
+      "loss": 0.546,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478995461811809e-06,
+      "loss": 0.5811,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477782088898903e-06,
+      "loss": 0.5747,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476567382555971e-06,
+      "loss": 0.5533,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475351343144736e-06,
+      "loss": 0.5074,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474133971027324e-06,
+      "loss": 0.6124,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472915266566249e-06,
+      "loss": 0.6786,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47169523012443e-06,
+      "loss": 0.6357,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470473862065176e-06,
+      "loss": 0.5918,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4692511627522e-06,
+      "loss": 0.5616,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468027132549605e-06,
+      "loss": 0.6364,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466801771821892e-06,
+      "loss": 0.5253,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465575080933959e-06,
+      "loss": 0.5746,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4643470602511e-06,
+      "loss": 0.6051,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463117710139007e-06,
+      "loss": 0.561,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461887030963763e-06,
+      "loss": 0.5545,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460655023091852e-06,
+      "loss": 0.5968,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45942168689015e-06,
+      "loss": 0.546,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458187022725928e-06,
+      "loss": 0.2188,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456951030966856e-06,
+      "loss": 0.5594,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455713711980998e-06,
+      "loss": 0.568,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454475066136813e-06,
+      "loss": 0.5165,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453235093803153e-06,
+      "loss": 0.5998,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45199379534927e-06,
+      "loss": 0.1794,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450751171144805e-06,
+      "loss": 0.5623,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449507221559798e-06,
+      "loss": 0.5702,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44826194696468e-06,
+      "loss": 0.5412,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447015347730283e-06,
+      "loss": 0.5972,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445767424227827e-06,
+      "loss": 0.5437,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444518176828929e-06,
+      "loss": 0.5628,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443267605905601e-06,
+      "loss": 0.5626,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442015711830246e-06,
+      "loss": 0.5938,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440762494975663e-06,
+      "loss": 0.5995,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439507955715048e-06,
+      "loss": 0.6049,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438252094421986e-06,
+      "loss": 0.5227,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436994911470457e-06,
+      "loss": 0.5481,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435736407234834e-06,
+      "loss": 0.5854,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434476582089888e-06,
+      "loss": 0.5869,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433215436410777e-06,
+      "loss": 0.5777,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431952970573058e-06,
+      "loss": 0.522,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430689184952674e-06,
+      "loss": 0.609,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429424079925968e-06,
+      "loss": 0.5222,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428157655869674e-06,
+      "loss": 0.6366,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426889913160916e-06,
+      "loss": 0.5599,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425620852177214e-06,
+      "loss": 0.6047,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424350473296477e-06,
+      "loss": 0.621,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42307877689701e-06,
+      "loss": 0.5977,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42180576335751e-06,
+      "loss": 0.5791,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420531433057065e-06,
+      "loss": 0.5163,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419255786375154e-06,
+      "loss": 0.6568,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417978823691652e-06,
+      "loss": 0.5614,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41670054538682e-06,
+      "loss": 0.5674,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415420951841314e-06,
+      "loss": 0.6397,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414140043436185e-06,
+      "loss": 0.533,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41285782055287e-06,
+      "loss": 0.4995,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411574283573199e-06,
+      "loss": 0.6004,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410289432879397e-06,
+      "loss": 0.6007,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409003268854074e-06,
+      "loss": 0.6001,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407715791880234e-06,
+      "loss": 0.6255,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406427002341274e-06,
+      "loss": 0.5802,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40513690062098e-06,
+      "loss": 0.4806,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403845487103531e-06,
+      "loss": 0.5368,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40255276217349e-06,
+      "loss": 0.5309,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401258726215818e-06,
+      "loss": 0.4958,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399963379615862e-06,
+      "loss": 0.6296,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398666722759362e-06,
+      "loss": 0.6054,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397368756032445e-06,
+      "loss": 0.1851,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396069479821635e-06,
+      "loss": 0.6002,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394768894513834e-06,
+      "loss": 0.558,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393467000496345e-06,
+      "loss": 0.538,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392163798156854e-06,
+      "loss": 0.6295,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390859287883444e-06,
+      "loss": 0.5408,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389553470064577e-06,
+      "loss": 0.5785,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388246345089114e-06,
+      "loss": 0.5549,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386937913346297e-06,
+      "loss": 0.1689,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385628175225765e-06,
+      "loss": 0.5896,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38431713111754e-06,
+      "loss": 0.6247,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383004781412037e-06,
+      "loss": 0.1206,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381691126500058e-06,
+      "loss": 0.5727,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380376166772793e-06,
+      "loss": 0.6489,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379059902621821e-06,
+      "loss": 0.6508,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377742334439113e-06,
+      "loss": 0.5399,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376423462617019e-06,
+      "loss": 0.6237,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375103287548288e-06,
+      "loss": 0.646,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373781809626051e-06,
+      "loss": 0.5287,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372459029243828e-06,
+      "loss": 0.6235,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371134946795526e-06,
+      "loss": 0.5242,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369809562675445e-06,
+      "loss": 0.6133,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368482877278264e-06,
+      "loss": 0.626,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367154890999058e-06,
+      "loss": 0.521,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365825604233282e-06,
+      "loss": 0.5254,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364495017376784e-06,
+      "loss": 0.5514,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363163130825795e-06,
+      "loss": 0.5818,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361829944976937e-06,
+      "loss": 0.588,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360495460227216e-06,
+      "loss": 0.6312,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359159676974025e-06,
+      "loss": 0.5214,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357822595615144e-06,
+      "loss": 0.5563,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356484216548742e-06,
+      "loss": 0.601,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355144540173371e-06,
+      "loss": 0.5724,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35380356688797e-06,
+      "loss": 0.5644,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352461297091865e-06,
+      "loss": 0.5403,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351117731184769e-06,
+      "loss": 0.5227,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349772869566778e-06,
+      "loss": 0.6241,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34842671263838e-06,
+      "loss": 0.6034,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347079260800438e-06,
+      "loss": 0.5366,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345730514454212e-06,
+      "loss": 0.6393,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34438047400134e-06,
+      "loss": 0.576,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34302913984385e-06,
+      "loss": 0.5717,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341676512384152e-06,
+      "loss": 0.5955,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340322592025042e-06,
+      "loss": 0.6056,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3389673791697e-06,
+      "loss": 0.5319,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337610874221696e-06,
+      "loss": 0.5547,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33625307758498e-06,
+      "loss": 0.5452,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334893989663885e-06,
+      "loss": 0.5894,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333533610863133e-06,
+      "loss": 0.59,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332171941587827e-06,
+      "loss": 0.5693,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330808982243457e-06,
+      "loss": 0.591,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329444733235897e-06,
+      "loss": 0.5539,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328079194971403e-06,
+      "loss": 0.6032,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326712367856617e-06,
+      "loss": 0.5423,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32534425229856e-06,
+      "loss": 0.1572,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323974848704646e-06,
+      "loss": 0.5923,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322604157482665e-06,
+      "loss": 0.5631,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32123217904079e-06,
+      "loss": 0.5752,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319858913787582e-06,
+      "loss": 0.6307,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318484362131983e-06,
+      "loss": 0.499,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317108524483319e-06,
+      "loss": 0.5855,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315731401251296e-06,
+      "loss": 0.5472,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314352992846004e-06,
+      "loss": 0.5667,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312973299677919e-06,
+      "loss": 0.5434,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311592322157896e-06,
+      "loss": 0.5788,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310210060697175e-06,
+      "loss": 0.5273,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308826515707376e-06,
+      "loss": 0.6661,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307441687600502e-06,
+      "loss": 0.6051,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306055576788938e-06,
+      "loss": 0.6161,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30466818368545e-06,
+      "loss": 0.5986,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30327950870319e-06,
+      "loss": 0.5641,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301889552255688e-06,
+      "loss": 0.1901,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300498314756857e-06,
+      "loss": 0.5208,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299105796620986e-06,
+      "loss": 0.1528,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297711998262758e-06,
+      "loss": 0.5685,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296316920097223e-06,
+      "loss": 0.5946,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294920562539821e-06,
+      "loss": 0.642,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293522926006373e-06,
+      "loss": 0.6115,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292124010913077e-06,
+      "loss": 0.5552,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29072381767651e-06,
+      "loss": 0.5641,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289322346713636e-06,
+      "loss": 0.5723,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287919598441796e-06,
+      "loss": 0.5226,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286515573278712e-06,
+      "loss": 0.5374,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285110271642483e-06,
+      "loss": 0.5145,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283703693951595e-06,
+      "loss": 0.5566,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282295840624908e-06,
+      "loss": 0.5877,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280886712081665e-06,
+      "loss": 0.6183,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279476308741486e-06,
+      "loss": 0.5337,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278064631024372e-06,
+      "loss": 0.5718,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276651679350707e-06,
+      "loss": 0.5227,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275237454141248e-06,
+      "loss": 0.6133,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273821955817135e-06,
+      "loss": 0.5644,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272405184799888e-06,
+      "loss": 0.6173,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270987141511403e-06,
+      "loss": 0.5847,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269567826373958e-06,
+      "loss": 0.5467,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268147239810203e-06,
+      "loss": 0.5926,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26672538224318e-06,
+      "loss": 0.5253,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265302254096294e-06,
+      "loss": 0.5368,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26387785579334e-06,
+      "loss": 0.6112,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262452187758483e-06,
+      "loss": 0.5371,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261025250416275e-06,
+      "loss": 0.5981,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259597044191635e-06,
+      "loss": 0.5997,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258167569509872e-06,
+      "loss": 0.5739,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25673682679666e-06,
+      "loss": 0.5547,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255304816478062e-06,
+      "loss": 0.5483,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253871538980512e-06,
+      "loss": 0.6112,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252436994730822e-06,
+      "loss": 0.6055,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251001184156184e-06,
+      "loss": 0.5733,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249564107684163e-06,
+      "loss": 0.5649,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248125765742705e-06,
+      "loss": 0.6388,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24668615876013e-06,
+      "loss": 0.5612,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245245287165137e-06,
+      "loss": 0.5326,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243803151386798e-06,
+      "loss": 0.6005,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242359751854566e-06,
+      "loss": 0.6256,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240915088998268e-06,
+      "loss": 0.5448,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239469163248106e-06,
+      "loss": 0.6153,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238021975034662e-06,
+      "loss": 0.6052,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236573524788888e-06,
+      "loss": 0.589,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235123812942118e-06,
+      "loss": 0.5934,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233672839926057e-06,
+      "loss": 0.5762,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232220606172789e-06,
+      "loss": 0.583,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23076711211477e-06,
+      "loss": 0.6557,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229312358184834e-06,
+      "loss": 0.5596,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227856344816188e-06,
+      "loss": 0.5762,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226399072442421e-06,
+      "loss": 0.5513,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224940541497485e-06,
+      "loss": 0.5555,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223480752415714e-06,
+      "loss": 0.605,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222019705631818e-06,
+      "loss": 0.513,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220557401580878e-06,
+      "loss": 0.5451,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.219093840698352e-06,
+      "loss": 0.5148,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21762902342007e-06,
+      "loss": 0.5861,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216162950182237e-06,
+      "loss": 0.6615,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214695621421432e-06,
+      "loss": 0.5647,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213227037574608e-06,
+      "loss": 0.4744,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211757199079092e-06,
+      "loss": 0.5187,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210286106372583e-06,
+      "loss": 0.182,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208813759893158e-06,
+      "loss": 0.6062,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207340160079261e-06,
+      "loss": 0.5283,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205865307369715e-06,
+      "loss": 0.5598,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204389202203711e-06,
+      "loss": 0.5762,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202911845020817e-06,
+      "loss": 0.6203,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201433236260972e-06,
+      "loss": 0.5628,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199953376364487e-06,
+      "loss": 0.5875,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198472265772049e-06,
+      "loss": 0.6104,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196989904924713e-06,
+      "loss": 0.6,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195506294263911e-06,
+      "loss": 0.6597,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19402143423144e-06,
+      "loss": 0.6084,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192535325269477e-06,
+      "loss": 0.5171,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191047967820568e-06,
+      "loss": 0.5455,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189559362327627e-06,
+      "loss": 0.5606,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188069509233948e-06,
+      "loss": 0.5844,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186578408983185e-06,
+      "loss": 0.5446,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185086062019374e-06,
+      "loss": 0.6066,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183592468786916e-06,
+      "loss": 0.6098,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18209762973059e-06,
+      "loss": 0.6032,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180601545295535e-06,
+      "loss": 0.5143,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17910421592727e-06,
+      "loss": 0.5889,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177605642071683e-06,
+      "loss": 0.6042,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17610582417503e-06,
+      "loss": 0.1893,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174604762683939e-06,
+      "loss": 0.5517,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173102458045407e-06,
+      "loss": 0.6353,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171598910706802e-06,
+      "loss": 0.5589,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170094121115866e-06,
+      "loss": 0.5928,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168588089720704e-06,
+      "loss": 0.5641,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167080816969797e-06,
+      "loss": 0.5699,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165572303311991e-06,
+      "loss": 0.53,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164062549196504e-06,
+      "loss": 0.6345,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16255155507292e-06,
+      "loss": 0.6555,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161039321391198e-06,
+      "loss": 0.4646,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159525848601665e-06,
+      "loss": 0.5728,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15801113715501e-06,
+      "loss": 0.5721,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156495187502299e-06,
+      "loss": 0.5036,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154978000094964e-06,
+      "loss": 0.5045,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153459575384805e-06,
+      "loss": 0.5906,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151939913823988e-06,
+      "loss": 0.5955,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150419015865054e-06,
+      "loss": 0.5978,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148896881960907e-06,
+      "loss": 0.5136,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14737351256482e-06,
+      "loss": 0.6275,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145848908130433e-06,
+      "loss": 0.5642,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144323069111755e-06,
+      "loss": 0.6329,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142795995963167e-06,
+      "loss": 0.5876,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141267689139407e-06,
+      "loss": 0.5417,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139738149095592e-06,
+      "loss": 0.5589,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138207376287196e-06,
+      "loss": 0.5934,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136675371170068e-06,
+      "loss": 0.6537,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135142134200419e-06,
+      "loss": 0.5631,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13360766583483e-06,
+      "loss": 0.5953,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132071966530247e-06,
+      "loss": 0.5872,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130535036743983e-06,
+      "loss": 0.594,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128996876933716e-06,
+      "loss": 0.5226,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127457487557493e-06,
+      "loss": 0.5528,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125916869073728e-06,
+      "loss": 0.5845,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124375021941194e-06,
+      "loss": 0.591,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122831946619038e-06,
+      "loss": 0.5453,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121287643566768e-06,
+      "loss": 0.5886,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119742113244258e-06,
+      "loss": 0.1657,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118195356111751e-06,
+      "loss": 0.6364,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116647372629852e-06,
+      "loss": 0.5824,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11509816325953e-06,
+      "loss": 0.5734,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113547728462125e-06,
+      "loss": 0.6423,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111996068699333e-06,
+      "loss": 0.5651,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110443184433224e-06,
+      "loss": 0.5731,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108889076126226e-06,
+      "loss": 0.4876,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107333744241135e-06,
+      "loss": 0.5436,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105777189241112e-06,
+      "loss": 0.6638,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104219411589676e-06,
+      "loss": 0.5612,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102660411750717e-06,
+      "loss": 0.5324,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101100190188487e-06,
+      "loss": 0.5296,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0995387473676e-06,
+      "loss": 0.5808,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097976083753037e-06,
+      "loss": 0.5871,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09641219981014e-06,
+      "loss": 0.5669,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094847096004613e-06,
+      "loss": 0.5301,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093280772802527e-06,
+      "loss": 0.6397,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091713230670314e-06,
+      "loss": 0.6422,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090144470074771e-06,
+      "loss": 0.5552,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088574491483053e-06,
+      "loss": 0.5355,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087003295362682e-06,
+      "loss": 0.5948,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085430882181543e-06,
+      "loss": 0.5632,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08385725240788e-06,
+      "loss": 0.6276,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0822824065103e-06,
+      "loss": 0.6034,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080706344957777e-06,
+      "loss": 0.5794,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07912906821964e-06,
+      "loss": 0.5169,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077550576765587e-06,
+      "loss": 0.559,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07597087106567e-06,
+      "loss": 0.1637,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074389951590307e-06,
+      "loss": 0.6131,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07280781881028e-06,
+      "loss": 0.5325,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071224473196727e-06,
+      "loss": 0.5947,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069639915221148e-06,
+      "loss": 0.5949,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068054145355407e-06,
+      "loss": 0.5636,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066467164071729e-06,
+      "loss": 0.6655,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064878971842695e-06,
+      "loss": 0.5245,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063289569141251e-06,
+      "loss": 0.5699,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061698956440703e-06,
+      "loss": 0.5805,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060107134214717e-06,
+      "loss": 0.5618,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058514102937315e-06,
+      "loss": 0.644,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056919863082887e-06,
+      "loss": 0.4586,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055324415126174e-06,
+      "loss": 0.5955,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053727759542286e-06,
+      "loss": 0.5625,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052129896806686e-06,
+      "loss": 0.5938,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050530827395198e-06,
+      "loss": 0.5459,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048930551784005e-06,
+      "loss": 0.5073,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047329070449653e-06,
+      "loss": 0.575,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04572638386904e-06,
+      "loss": 0.5518,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044122492519432e-06,
+      "loss": 0.6134,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042517396878443e-06,
+      "loss": 0.5895,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040911097424054e-06,
+      "loss": 0.2056,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039303594634603e-06,
+      "loss": 0.5686,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037694888988785e-06,
+      "loss": 0.4942,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.036084980965652e-06,
+      "loss": 0.5556,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034473871044613e-06,
+      "loss": 0.1544,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032861559705442e-06,
+      "loss": 0.6182,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031248047428264e-06,
+      "loss": 0.593,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029633334693564e-06,
+      "loss": 0.4947,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028017421982182e-06,
+      "loss": 0.5646,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026400309775321e-06,
+      "loss": 0.5696,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024781998554538e-06,
+      "loss": 0.6216,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023162488801741e-06,
+      "loss": 0.5722,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021541780999206e-06,
+      "loss": 0.5169,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019919875629559e-06,
+      "loss": 0.6141,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018296773175784e-06,
+      "loss": 0.5701,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016672474121222e-06,
+      "loss": 0.5026,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015046978949569e-06,
+      "loss": 0.5697,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013420288144878e-06,
+      "loss": 0.5215,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011792402191559e-06,
+      "loss": 0.4775,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010163321574377e-06,
+      "loss": 0.5278,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008533046778452e-06,
+      "loss": 0.5923,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00690157828926e-06,
+      "loss": 0.529,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005268916592634e-06,
+      "loss": 0.5845,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00363506217476e-06,
+      "loss": 0.5163,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002000015522182e-06,
+      "loss": 0.5317,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000363777121794e-06,
+      "loss": 0.5794,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998726347460854e-06,
+      "loss": 0.6289,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997087727026963e-06,
+      "loss": 0.5235,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995447916308086e-06,
+      "loss": 0.6641,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993806915792537e-06,
+      "loss": 0.5226,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99216472596899e-06,
+      "loss": 0.5457,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990521347326465e-06,
+      "loss": 0.5664,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988876780354343e-06,
+      "loss": 0.5092,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987231025542357e-06,
+      "loss": 0.6018,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985584083380592e-06,
+      "loss": 0.477,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983935954359487e-06,
+      "loss": 0.5918,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982286638969836e-06,
+      "loss": 0.615,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980636137702785e-06,
+      "loss": 0.629,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978984451049835e-06,
+      "loss": 0.58,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977331579502837e-06,
+      "loss": 0.5026,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975677523553995e-06,
+      "loss": 0.4895,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974022283695874e-06,
+      "loss": 0.5448,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972365860421375e-06,
+      "loss": 0.5442,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970708254223768e-06,
+      "loss": 0.5523,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969049465596666e-06,
+      "loss": 0.5974,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967389495034035e-06,
+      "loss": 0.5601,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965728343030198e-06,
+      "loss": 0.5384,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964066010079824e-06,
+      "loss": 0.5916,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962402496677934e-06,
+      "loss": 0.5632,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960737803319906e-06,
+      "loss": 0.5897,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959071930501464e-06,
+      "loss": 0.5492,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957404878718686e-06,
+      "loss": 0.575,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955736648467999e-06,
+      "loss": 0.5934,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954067240246182e-06,
+      "loss": 0.4829,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952396654550366e-06,
+      "loss": 0.5924,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95072489187803e-06,
+      "loss": 0.5577,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949051952727006e-06,
+      "loss": 0.6345,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947377837595474e-06,
+      "loss": 0.5622,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94570254698197e-06,
+      "loss": 0.5882,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944026081385369e-06,
+      "loss": 0.5859,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942348441304906e-06,
+      "loss": 0.532,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940669627240162e-06,
+      "loss": 0.5851,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.938989639691068e-06,
+      "loss": 0.5856,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937308479157902e-06,
+      "loss": 0.5256,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935626146141295e-06,
+      "loss": 0.6208,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933942641142229e-06,
+      "loss": 0.5868,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932257964662025e-06,
+      "loss": 0.5466,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930572117202366e-06,
+      "loss": 0.5904,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928885099265273e-06,
+      "loss": 0.5407,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927196911353121e-06,
+      "loss": 0.6138,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925507553968633e-06,
+      "loss": 0.583,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923817027614879e-06,
+      "loss": 0.5231,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922125332795275e-06,
+      "loss": 0.5563,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920432470013593e-06,
+      "loss": 0.5133,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918738439773942e-06,
+      "loss": 0.4847,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917043242580787e-06,
+      "loss": 0.505,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915346878938937e-06,
+      "loss": 0.5923,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913649349353546e-06,
+      "loss": 0.1749,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911950654330123e-06,
+      "loss": 0.6519,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910250794374517e-06,
+      "loss": 0.5793,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908549769992924e-06,
+      "loss": 0.5644,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90684758169189e-06,
+      "loss": 0.5181,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90514422997831e-06,
+      "loss": 0.5184,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903439715359414e-06,
+      "loss": 0.592,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901734038342792e-06,
+      "loss": 0.5297,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900027199436374e-06,
+      "loss": 0.1754,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898319199148435e-06,
+      "loss": 0.5517,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896610037987596e-06,
+      "loss": 0.5846,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894899716462828e-06,
+      "loss": 0.5999,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893188235083439e-06,
+      "loss": 0.6188,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891475594359093e-06,
+      "loss": 0.6448,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889761794799792e-06,
+      "loss": 0.5049,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888046836915885e-06,
+      "loss": 0.626,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886330721218065e-06,
+      "loss": 0.5638,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884613448217372e-06,
+      "loss": 0.5548,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88289501842519e-06,
+      "loss": 0.517,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881175432353244e-06,
+      "loss": 0.5543,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87945469051361e-06,
+      "loss": 0.5622,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877732793418702e-06,
+      "loss": 0.5223,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87600974158128e-06,
+      "loss": 0.1569,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87428553551445e-06,
+      "loss": 0.5793,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87256017573166e-06,
+      "loss": 0.4956,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870833662746698e-06,
+      "loss": 0.5927,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869105997073704e-06,
+      "loss": 0.5189,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867377179227152e-06,
+      "loss": 0.5602,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865647209721867e-06,
+      "loss": 0.4411,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863916089073011e-06,
+      "loss": 0.596,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86218381779609e-06,
+      "loss": 0.5437,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860450396406956e-06,
+      "loss": 0.5484,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8587158254218e-06,
+      "loss": 0.1604,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856980105357158e-06,
+      "loss": 0.5378,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855243236729905e-06,
+      "loss": 0.5676,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853505220057258e-06,
+      "loss": 0.5763,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851766055856784e-06,
+      "loss": 0.5203,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850025744646378e-06,
+      "loss": 0.5448,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848284286944289e-06,
+      "loss": 0.1616,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8465416832691e-06,
+      "loss": 0.6158,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844797934139736e-06,
+      "loss": 0.5929,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843053040075468e-06,
+      "loss": 0.5528,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841307001595904e-06,
+      "loss": 0.5753,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839559819220992e-06,
+      "loss": 0.1503,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837811493471023e-06,
+      "loss": 0.5414,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836062024866627e-06,
+      "loss": 0.6043,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834311413928775e-06,
+      "loss": 0.5802,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832559661178777e-06,
+      "loss": 0.5924,
+      "step": 1457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830806767138287e-06,
+      "loss": 0.5044,
+      "step": 1458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829052732329292e-06,
+      "loss": 0.5279,
+      "step": 1459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827297557274127e-06,
+      "loss": 0.5822,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825541242495459e-06,
+      "loss": 0.5973,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823783788516298e-06,
+      "loss": 0.4995,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822025195859994e-06,
+      "loss": 0.5431,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820265465050233e-06,
+      "loss": 0.6126,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818504596611044e-06,
+      "loss": 0.5619,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816742591066792e-06,
+      "loss": 0.5245,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814979448942179e-06,
+      "loss": 0.5665,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81321517076225e-06,
+      "loss": 0.5308,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811449757052386e-06,
+      "loss": 0.5464,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809683208338304e-06,
+      "loss": 0.5073,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807915525146065e-06,
+      "loss": 0.5169,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80614670800206e-06,
+      "loss": 0.5213,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804376757433024e-06,
+      "loss": 0.5831,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802605673966025e-06,
+      "loss": 0.551,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800833458128473e-06,
+      "loss": 0.5986,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79906011044811e-06,
+      "loss": 0.5207,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79728563145302e-06,
+      "loss": 0.6608,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795510021671622e-06,
+      "loss": 0.5044,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793733281632668e-06,
+      "loss": 0.5021,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791955411865254e-06,
+      "loss": 0.5111,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790176412898808e-06,
+      "loss": 0.5058,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78839628526309e-06,
+      "loss": 0.5398,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786615029488206e-06,
+      "loss": 0.1602,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784832646104591e-06,
+      "loss": 0.5361,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783049135643017e-06,
+      "loss": 0.5793,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781264498634591e-06,
+      "loss": 0.5114,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77947873561076e-06,
+      "loss": 0.5513,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7776918471033e-06,
+      "loss": 0.5762,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775903833644326e-06,
+      "loss": 0.5743,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774114695766286e-06,
+      "loss": 0.4884,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772324434001965e-06,
+      "loss": 0.6027,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770533048884483e-06,
+      "loss": 0.5939,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768740540947289e-06,
+      "loss": 0.5527,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766946910724175e-06,
+      "loss": 0.624,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765152158749259e-06,
+      "loss": 0.5088,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763356285557e-06,
+      "loss": 0.5595,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761559291682184e-06,
+      "loss": 0.6204,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759761177659938e-06,
+      "loss": 0.5417,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757961944025717e-06,
+      "loss": 0.5503,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75616159131531e-06,
+      "loss": 0.5573,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754360120064844e-06,
+      "loss": 0.5357,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752557530810773e-06,
+      "loss": 0.5223,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750753824089887e-06,
+      "loss": 0.5526,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748949000439306e-06,
+      "loss": 0.4531,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747143060396489e-06,
+      "loss": 0.1799,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745336004499221e-06,
+      "loss": 0.5609,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743527833285624e-06,
+      "loss": 0.5421,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741718547294148e-06,
+      "loss": 0.5333,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739908147063576e-06,
+      "loss": 0.6341,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738096633133024e-06,
+      "loss": 0.5492,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736284006041941e-06,
+      "loss": 0.5273,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734470266330103e-06,
+      "loss": 0.6265,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732655414537623e-06,
+      "loss": 0.5291,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73083945120494e-06,
+      "loss": 0.4813,
+      "step": 1514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729022376872826e-06,
+      "loss": 0.5531,
+      "step": 1515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727204192082388e-06,
+      "loss": 0.6004,
+      "step": 1516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725384897375054e-06,
+      "loss": 0.5777,
+      "step": 1517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723564493292592e-06,
+      "loss": 0.5429,
+      "step": 1518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721742980377092e-06,
+      "loss": 0.6186,
+      "step": 1519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719920359170986e-06,
+      "loss": 0.5511,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718096630217024e-06,
+      "loss": 0.5283,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71627179405829e-06,
+      "loss": 0.5851,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714445851238202e-06,
+      "loss": 0.5993,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712618802300501e-06,
+      "loss": 0.6037,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710790647789262e-06,
+      "loss": 0.4758,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708961388248887e-06,
+      "loss": 0.4954,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707131024224109e-06,
+      "loss": 0.544,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705299556259986e-06,
+      "loss": 0.5014,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703466984901907e-06,
+      "loss": 0.565,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701633310695593e-06,
+      "loss": 0.5204,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69979853418709e-06,
+      "loss": 0.5032,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697962655922768e-06,
+      "loss": 0.5058,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696125676449334e-06,
+      "loss": 0.5249,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694287596313817e-06,
+      "loss": 0.5812,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692448416063578e-06,
+      "loss": 0.534,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690608136246298e-06,
+      "loss": 0.164,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688766757409995e-06,
+      "loss": 0.5557,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686924280103005e-06,
+      "loss": 0.5794,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685080704874001e-06,
+      "loss": 0.564,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683236032271973e-06,
+      "loss": 0.6058,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681390262846245e-06,
+      "loss": 0.5492,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679543397146464e-06,
+      "loss": 0.5229,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677695435722606e-06,
+      "loss": 0.5934,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675846379124971e-06,
+      "loss": 0.5181,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673996227904185e-06,
+      "loss": 0.6087,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672144982611202e-06,
+      "loss": 0.5653,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670292643797302e-06,
+      "loss": 0.5092,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668439212014086e-06,
+      "loss": 0.5794,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666584687813486e-06,
+      "loss": 0.564,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664729071747758e-06,
+      "loss": 0.5257,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66287236436948e-06,
+      "loss": 0.5412,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66101456623156e-06,
+      "loss": 0.4819,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659155677887226e-06,
+      "loss": 0.5933,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657295699890033e-06,
+      "loss": 0.4369,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655434632793861e-06,
+      "loss": 0.5792,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653572477152914e-06,
+      "loss": 0.5339,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651709233521717e-06,
+      "loss": 0.5361,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649844902455125e-06,
+      "loss": 0.5272,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647979484508311e-06,
+      "loss": 0.5773,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646112980236776e-06,
+      "loss": 0.4916,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644245390196341e-06,
+      "loss": 0.5302,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642376714943155e-06,
+      "loss": 0.525,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640506955033683e-06,
+      "loss": 0.5967,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63863611102472e-06,
+      "loss": 0.557,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63676418347338e-06,
+      "loss": 0.566,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634891172937102e-06,
+      "loss": 0.5698,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633017079973644e-06,
+      "loss": 0.5488,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631141905141091e-06,
+      "loss": 0.4873,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629265648997847e-06,
+      "loss": 0.1782,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627388312102636e-06,
+      "loss": 0.5646,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62550989501451e-06,
+      "loss": 0.5891,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623630398292836e-06,
+      "loss": 0.5204,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621749822497308e-06,
+      "loss": 0.5816,
+      "step": 1573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619868168187939e-06,
+      "loss": 0.5018,
+      "step": 1574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617985435925061e-06,
+      "loss": 0.5129,
+      "step": 1575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616101626269332e-06,
+      "loss": 0.4946,
+      "step": 1576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614216739781727e-06,
+      "loss": 0.5384,
+      "step": 1577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61233077702354e-06,
+      "loss": 0.6008,
+      "step": 1578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610443738556393e-06,
+      "loss": 0.523,
+      "step": 1579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60855562494222e-06,
+      "loss": 0.1746,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606666436743281e-06,
+      "loss": 0.5361,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604776174522152e-06,
+      "loss": 0.6196,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602884838841732e-06,
+      "loss": 0.5457,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600992430265237e-06,
+      "loss": 0.5868,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599098949356201e-06,
+      "loss": 0.6139,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597204396678486e-06,
+      "loss": 0.579,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595308772796263e-06,
+      "loss": 0.5832,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593412078274026e-06,
+      "loss": 0.525,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591514313676589e-06,
+      "loss": 0.5316,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589615479569087e-06,
+      "loss": 0.6157,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587715576516963e-06,
+      "loss": 0.6271,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585814605085992e-06,
+      "loss": 0.5933,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583912565842258e-06,
+      "loss": 0.5538,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582009459352162e-06,
+      "loss": 0.5889,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580105286182434e-06,
+      "loss": 0.5263,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578200046900108e-06,
+      "loss": 0.554,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576293742072543e-06,
+      "loss": 0.555,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574386372267416e-06,
+      "loss": 0.5762,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572477938052716e-06,
+      "loss": 0.1532,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570568439996753e-06,
+      "loss": 0.5405,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568657878668155e-06,
+      "loss": 0.5789,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566746254635857e-06,
+      "loss": 0.5442,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564833568469127e-06,
+      "loss": 0.6335,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562919820737537e-06,
+      "loss": 0.5298,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561005012010975e-06,
+      "loss": 0.5408,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55908914285965e-06,
+      "loss": 0.5774,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557172213854086e-06,
+      "loss": 0.146,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555254225565121e-06,
+      "loss": 0.5147,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553335178563909e-06,
+      "loss": 0.461,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551415073421919e-06,
+      "loss": 0.5076,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549493910710935e-06,
+      "loss": 0.5191,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547571691003058e-06,
+      "loss": 0.5734,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545648414870701e-06,
+      "loss": 0.5094,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543724082886594e-06,
+      "loss": 0.5514,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541798695623776e-06,
+      "loss": 0.5722,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539872253655608e-06,
+      "loss": 0.5645,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537944757555762e-06,
+      "loss": 0.4715,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53601620789822e-06,
+      "loss": 0.5773,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534086605257284e-06,
+      "loss": 0.5753,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532155950207564e-06,
+      "loss": 0.5427,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530224243323988e-06,
+      "loss": 0.5638,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528291485181795e-06,
+      "loss": 0.5653,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526357676356538e-06,
+      "loss": 0.5446,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524422817424078e-06,
+      "loss": 0.5776,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5224869089606e-06,
+      "loss": 0.5353,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520549951542588e-06,
+      "loss": 0.517,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518611945746848e-06,
+      "loss": 0.5059,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516672892150495e-06,
+      "loss": 0.5599,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514732791330955e-06,
+      "loss": 0.4802,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512791643865967e-06,
+      "loss": 0.5068,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51084945033358e-06,
+      "loss": 0.5557,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.50890621131216e-06,
+      "loss": 0.546,
+      "step": 1632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506961927380377e-06,
+      "loss": 0.5733,
+      "step": 1633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505016599117216e-06,
+      "loss": 0.5535,
+      "step": 1634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503070227101974e-06,
+      "loss": 0.1645,
+      "step": 1635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501122811914256e-06,
+      "loss": 0.4988,
+      "step": 1636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49917435413398e-06,
+      "loss": 0.5833,
+      "step": 1637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497224854341372e-06,
+      "loss": 0.5218,
+      "step": 1638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495274313116971e-06,
+      "loss": 0.4593,
+      "step": 1639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493322731041625e-06,
+      "loss": 0.168,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49137010869649e-06,
+      "loss": 0.4991,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489416446663037e-06,
+      "loss": 0.5798,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48746174552304e-06,
+      "loss": 0.6002,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485506005858586e-06,
+      "loss": 0.5201,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483549228252074e-06,
+      "loss": 0.5587,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481591413286205e-06,
+      "loss": 0.5406,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479632561543996e-06,
+      "loss": 0.5203,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477672673608767e-06,
+      "loss": 0.5179,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47571175006415e-06,
+      "loss": 0.5548,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473749791494088e-06,
+      "loss": 0.5075,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471786798482822e-06,
+      "loss": 0.5417,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469822771614916e-06,
+      "loss": 0.5416,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467857711475227e-06,
+      "loss": 0.5694,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46589161864893e-06,
+      "loss": 0.5462,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463924493721503e-06,
+      "loss": 0.517,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461956337278732e-06,
+      "loss": 0.5001,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45998714990671e-06,
+      "loss": 0.5862,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458016932191839e-06,
+      "loss": 0.5353,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456045684720826e-06,
+      "loss": 0.5438,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454073408080686e-06,
+      "loss": 0.5748,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452100102858734e-06,
+      "loss": 0.4911,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450125769642604e-06,
+      "loss": 0.5518,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448150409020224e-06,
+      "loss": 0.4976,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446174021579834e-06,
+      "loss": 0.5592,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44419660790998e-06,
+      "loss": 0.5321,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442218168599512e-06,
+      "loss": 0.1684,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440238704237583e-06,
+      "loss": 0.5675,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438258215413657e-06,
+      "loss": 0.549,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436276702717497e-06,
+      "loss": 0.5763,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434294166739176e-06,
+      "loss": 0.523,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432310608069067e-06,
+      "loss": 0.6021,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430326027297855e-06,
+      "loss": 0.5106,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42834042501652e-06,
+      "loss": 0.15,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42635380181635e-06,
+      "loss": 0.5707,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424366158288944e-06,
+      "loss": 0.5618,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422377495026191e-06,
+      "loss": 0.5431,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420387812620295e-06,
+      "loss": 0.5771,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418397111663758e-06,
+      "loss": 0.5381,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41640539274939e-06,
+      "loss": 0.6368,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414412656470297e-06,
+      "loss": 0.1406,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412418903419895e-06,
+      "loss": 0.5877,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410424134191898e-06,
+      "loss": 0.5311,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408428349380329e-06,
+      "loss": 0.6187,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4064315495795e-06,
+      "loss": 0.5282,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404433735384043e-06,
+      "loss": 0.5218,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402434907388879e-06,
+      "loss": 0.5524,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400435066189236e-06,
+      "loss": 0.5799,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398434212380642e-06,
+      "loss": 0.5332,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396432346558928e-06,
+      "loss": 0.5385,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394429469320226e-06,
+      "loss": 0.58,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392425581260967e-06,
+      "loss": 0.5438,
+      "step": 1691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.390420682977888e-06,
+      "loss": 0.5446,
+      "step": 1692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388414775068023e-06,
+      "loss": 0.5614,
+      "step": 1693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386407858128707e-06,
+      "loss": 0.577,
+      "step": 1694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384399932757573e-06,
+      "loss": 0.579,
+      "step": 1695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382390999552562e-06,
+      "loss": 0.5678,
+      "step": 1696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380381059111907e-06,
+      "loss": 0.5534,
+      "step": 1697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378370112034147e-06,
+      "loss": 0.5859,
+      "step": 1698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376358158918114e-06,
+      "loss": 0.5075,
+      "step": 1699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374345200362947e-06,
+      "loss": 0.5832,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372331236968081e-06,
+      "loss": 0.502,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370316269333247e-06,
+      "loss": 0.5782,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368300298058483e-06,
+      "loss": 0.1941,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366283323744117e-06,
+      "loss": 0.6541,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364265346990784e-06,
+      "loss": 0.5823,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362246368399406e-06,
+      "loss": 0.5301,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360226388571219e-06,
+      "loss": 0.5033,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358205408107748e-06,
+      "loss": 0.5134,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35618342761081e-06,
+      "loss": 0.5528,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354160447682533e-06,
+      "loss": 0.5299,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352136468925333e-06,
+      "loss": 0.5367,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35011149194193e-06,
+      "loss": 0.5838,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348085517335335e-06,
+      "loss": 0.5236,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34605854570886e-06,
+      "loss": 0.6185,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344030577666113e-06,
+      "loss": 0.5499,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342001613810999e-06,
+      "loss": 0.5313,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339971654747719e-06,
+      "loss": 0.5381,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33794070108077e-06,
+      "loss": 0.616,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335908753414945e-06,
+      "loss": 0.5884,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333875812355334e-06,
+      "loss": 0.5094,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331841878507327e-06,
+      "loss": 0.5336,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329806952476598e-06,
+      "loss": 0.5462,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32777103486913e-06,
+      "loss": 0.4841,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325734126291191e-06,
+      "loss": 0.6347,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323696227349349e-06,
+      "loss": 0.6129,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321657338650468e-06,
+      "loss": 0.5012,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3196174608017e-06,
+      "loss": 0.5309,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317576594410503e-06,
+      "loss": 0.5368,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315534740084616e-06,
+      "loss": 0.5861,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313491898432082e-06,
+      "loss": 0.5392,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311448070061237e-06,
+      "loss": 0.5231,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309403255580703e-06,
+      "loss": 0.6103,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307357455599407e-06,
+      "loss": 0.5178,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305310670726558e-06,
+      "loss": 0.5879,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30326290157167e-06,
+      "loss": 0.576,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301214148744543e-06,
+      "loss": 0.4925,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299164412855268e-06,
+      "loss": 0.5589,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297113694514234e-06,
+      "loss": 0.5134,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295061994332119e-06,
+      "loss": 0.5917,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293009312919898e-06,
+      "loss": 0.5644,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29095565088883e-06,
+      "loss": 0.4901,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288901008850477e-06,
+      "loss": 0.5371,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28684538741668e-06,
+      "loss": 0.5524,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284788787199585e-06,
+      "loss": 0.5223,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28273120881162e-06,
+      "loss": 0.1712,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280672652865506e-06,
+      "loss": 0.5179,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27861311997426e-06,
+      "loss": 0.5772,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276552610751182e-06,
+      "loss": 0.5404,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274491125809872e-06,
+      "loss": 0.5931,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272428665764212e-06,
+      "loss": 0.6001,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270365231228377e-06,
+      "loss": 0.6137,
+      "step": 1751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268300822816838e-06,
+      "loss": 0.5748,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266235441144345e-06,
+      "loss": 0.1323,
+      "step": 1753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264169086825949e-06,
+      "loss": 0.4329,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262101760476981e-06,
+      "loss": 0.5233,
+      "step": 1755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260033462713073e-06,
+      "loss": 0.5656,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257964194150132e-06,
+      "loss": 0.5827,
+      "step": 1757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255893955404368e-06,
+      "loss": 0.4864,
+      "step": 1758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253822747092266e-06,
+      "loss": 0.551,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251750569830615e-06,
+      "loss": 0.5958,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249677424236479e-06,
+      "loss": 0.4691,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247603310927216e-06,
+      "loss": 0.5081,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245528230520477e-06,
+      "loss": 0.6072,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24345218363419e-06,
+      "loss": 0.5851,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241375170886583e-06,
+      "loss": 0.1562,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239297192896163e-06,
+      "loss": 0.5544,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237218250281724e-06,
+      "loss": 0.5649,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235138343662355e-06,
+      "loss": 0.572,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233057473657424e-06,
+      "loss": 0.5485,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230975640886591e-06,
+      "loss": 0.5698,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228892845969802e-06,
+      "loss": 0.5622,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226809089527286e-06,
+      "loss": 0.1755,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224724372179562e-06,
+      "loss": 0.5387,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222638694547434e-06,
+      "loss": 0.5101,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22055205725199e-06,
+      "loss": 0.5102,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218464460914609e-06,
+      "loss": 0.5579,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216375906156948e-06,
+      "loss": 0.1668,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21428639360096e-06,
+      "loss": 0.508,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.212195923868868e-06,
+      "loss": 0.5124,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210104497583198e-06,
+      "loss": 0.6228,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208012115366746e-06,
+      "loss": 0.6093,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205918777842602e-06,
+      "loss": 0.5837,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203824485634133e-06,
+      "loss": 0.5674,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201729239364998e-06,
+      "loss": 0.6014,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199633039659132e-06,
+      "loss": 0.5335,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197535887140763e-06,
+      "loss": 0.5265,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195437782434395e-06,
+      "loss": 0.5462,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19333872616482e-06,
+      "loss": 0.5562,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191238718957111e-06,
+      "loss": 0.5054,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189137761436624e-06,
+      "loss": 0.5587,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187035854229003e-06,
+      "loss": 0.4429,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184932997960167e-06,
+      "loss": 0.5514,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182829193256322e-06,
+      "loss": 0.5709,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180724440743957e-06,
+      "loss": 0.4847,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178618741049841e-06,
+      "loss": 0.5052,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176512094801028e-06,
+      "loss": 0.5673,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174404502624852e-06,
+      "loss": 0.5111,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172295965148925e-06,
+      "loss": 0.6008,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170186483001148e-06,
+      "loss": 0.5252,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1680760568097e-06,
+      "loss": 0.5738,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165964687203039e-06,
+      "loss": 0.5515,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163852374809906e-06,
+      "loss": 0.5296,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161739120259322e-06,
+      "loss": 0.5578,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15962492418059e-06,
+      "loss": 0.5735,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157509787203292e-06,
+      "loss": 0.5504,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155393709957291e-06,
+      "loss": 0.5635,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153276693072727e-06,
+      "loss": 0.5592,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151158737180026e-06,
+      "loss": 0.4963,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149039842909887e-06,
+      "loss": 0.5425,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146920010893294e-06,
+      "loss": 0.4823,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144799241761503e-06,
+      "loss": 0.5284,
+      "step": 1811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14267753614606e-06,
+      "loss": 0.5373,
+      "step": 1812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14055489467878e-06,
+      "loss": 0.538,
+      "step": 1813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13843131799176e-06,
+      "loss": 0.5282,
+      "step": 1814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136306806717377e-06,
+      "loss": 0.4423,
+      "step": 1815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134181361488285e-06,
+      "loss": 0.5622,
+      "step": 1816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132054982937416e-06,
+      "loss": 0.4961,
+      "step": 1817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12992767169798e-06,
+      "loss": 0.4987,
+      "step": 1818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127799428403465e-06,
+      "loss": 0.1643,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125670253687635e-06,
+      "loss": 0.5563,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123540148184535e-06,
+      "loss": 0.5187,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121409112528484e-06,
+      "loss": 0.5312,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11927714735408e-06,
+      "loss": 0.5922,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117144253296195e-06,
+      "loss": 0.5017,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115010430989981e-06,
+      "loss": 0.5995,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112875681070863e-06,
+      "loss": 0.5197,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110740004174544e-06,
+      "loss": 0.4914,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108603400937007e-06,
+      "loss": 0.5507,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106465871994502e-06,
+      "loss": 0.5523,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104327417983561e-06,
+      "loss": 0.5544,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102188039540993e-06,
+      "loss": 0.502,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100047737303877e-06,
+      "loss": 0.591,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097906511909568e-06,
+      "loss": 0.5482,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0957643639957e-06,
+      "loss": 0.5132,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093621294200176e-06,
+      "loss": 0.5414,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091477303161182e-06,
+      "loss": 0.5632,
+      "step": 1836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089332391517166e-06,
+      "loss": 0.5664,
+      "step": 1837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087186559906862e-06,
+      "loss": 0.5133,
+      "step": 1838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08503980896927e-06,
+      "loss": 0.585,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08289213934367e-06,
+      "loss": 0.5048,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08074355166961e-06,
+      "loss": 0.5354,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078594046586917e-06,
+      "loss": 0.5556,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076443624735681e-06,
+      "loss": 0.5306,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074292286756279e-06,
+      "loss": 0.5859,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07214003328935e-06,
+      "loss": 0.533,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069986864975811e-06,
+      "loss": 0.5091,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06783278245685e-06,
+      "loss": 0.5282,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065677786373926e-06,
+      "loss": 0.5833,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06352187736877e-06,
+      "loss": 0.4719,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06136505608339e-06,
+      "loss": 0.4814,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059207323160057e-06,
+      "loss": 0.1728,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05704867924132e-06,
+      "loss": 0.5816,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054889124969997e-06,
+      "loss": 0.509,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05272866098918e-06,
+      "loss": 0.5966,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050567287942225e-06,
+      "loss": 0.5573,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048405006472766e-06,
+      "loss": 0.1532,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046241817224704e-06,
+      "loss": 0.6202,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044077720842211e-06,
+      "loss": 0.5552,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04191271796973e-06,
+      "loss": 0.5605,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039746809251972e-06,
+      "loss": 0.504,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037579995333917e-06,
+      "loss": 0.5344,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035412276860819e-06,
+      "loss": 0.5181,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033243654478198e-06,
+      "loss": 0.5266,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031074128831843e-06,
+      "loss": 0.58,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028903700567812e-06,
+      "loss": 0.5973,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026732370332437e-06,
+      "loss": 0.5348,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024560138772308e-06,
+      "loss": 0.5549,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022387006534294e-06,
+      "loss": 0.6587,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020212974265527e-06,
+      "loss": 0.5841,
+      "step": 1869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.018038042613407e-06,
+      "loss": 0.4862,
+      "step": 1870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015862212225604e-06,
+      "loss": 0.5715,
+      "step": 1871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013685483750052e-06,
+      "loss": 0.5312,
+      "step": 1872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011507857834957e-06,
+      "loss": 0.5963,
+      "step": 1873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00932933512879e-06,
+      "loss": 0.5748,
+      "step": 1874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007149916280288e-06,
+      "loss": 0.5483,
+      "step": 1875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004969601938456e-06,
+      "loss": 0.5855,
+      "step": 1876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002788392752563e-06,
+      "loss": 0.4824,
+      "step": 1877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000606289372153e-06,
+      "loss": 0.5493,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998423292447025e-06,
+      "loss": 0.5241,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996239402627254e-06,
+      "loss": 0.5154,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99405462056317e-06,
+      "loss": 0.538,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991868946905382e-06,
+      "loss": 0.5281,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989682382304753e-06,
+      "loss": 0.5399,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987494927412416e-06,
+      "loss": 0.5044,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985306582879772e-06,
+      "loss": 0.5689,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983117349358481e-06,
+      "loss": 0.6229,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980927227500472e-06,
+      "loss": 0.5175,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978736217957936e-06,
+      "loss": 0.518,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97654432138333e-06,
+      "loss": 0.548,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974351538429375e-06,
+      "loss": 0.5451,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972157869749057e-06,
+      "loss": 0.6012,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96996331599562e-06,
+      "loss": 0.4647,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.967767877822581e-06,
+      "loss": 0.1457,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965571555883711e-06,
+      "loss": 0.5096,
+      "step": 1894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96337435083305e-06,
+      "loss": 0.157,
+      "step": 1895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961176263324902e-06,
+      "loss": 0.5447,
+      "step": 1896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958977294013825e-06,
+      "loss": 0.5534,
+      "step": 1897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956777443554653e-06,
+      "loss": 0.6062,
+      "step": 1898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95457671260247e-06,
+      "loss": 0.1515,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95237510181263e-06,
+      "loss": 0.606,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950172611840747e-06,
+      "loss": 0.5447,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947969243342692e-06,
+      "loss": 0.6109,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945764996974607e-06,
+      "loss": 0.5502,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943559873392888e-06,
+      "loss": 0.5697,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941353873254193e-06,
+      "loss": 0.5236,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939146997215444e-06,
+      "loss": 0.5446,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936939245933824e-06,
+      "loss": 0.5071,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93473062006677e-06,
+      "loss": 0.5342,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932521120271992e-06,
+      "loss": 0.5409,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930310747207446e-06,
+      "loss": 0.5403,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928099501531356e-06,
+      "loss": 0.5465,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925887383902207e-06,
+      "loss": 0.5724,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923674394978738e-06,
+      "loss": 0.5501,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921460535419954e-06,
+      "loss": 0.5283,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919245805885113e-06,
+      "loss": 0.5103,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917030207033737e-06,
+      "loss": 0.5079,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914813739525604e-06,
+      "loss": 0.5381,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912596404020752e-06,
+      "loss": 0.542,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910378201179476e-06,
+      "loss": 0.5522,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908159131662332e-06,
+      "loss": 0.4605,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905939196130132e-06,
+      "loss": 0.4446,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903718395243945e-06,
+      "loss": 0.6006,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901496729665099e-06,
+      "loss": 0.5128,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89927420005518e-06,
+      "loss": 0.546,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897050807076034e-06,
+      "loss": 0.1696,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894826551389757e-06,
+      "loss": 0.5812,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892601433658705e-06,
+      "loss": 0.5505,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890375454545494e-06,
+      "loss": 0.5421,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888148614712992e-06,
+      "loss": 0.5271,
+      "step": 1929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885920914824326e-06,
+      "loss": 0.6548,
+      "step": 1930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88369235554288e-06,
+      "loss": 0.5719,
+      "step": 1931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881462937532292e-06,
+      "loss": 0.5315,
+      "step": 1932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879232661456455e-06,
+      "loss": 0.5626,
+      "step": 1933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877001527979518e-06,
+      "loss": 0.6266,
+      "step": 1934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874769537765887e-06,
+      "loss": 0.5605,
+      "step": 1935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872536691480222e-06,
+      "loss": 0.5237,
+      "step": 1936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870302989787439e-06,
+      "loss": 0.5048,
+      "step": 1937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868068433352704e-06,
+      "loss": 0.4747,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865833022841445e-06,
+      "loss": 0.5112,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86359675891934e-06,
+      "loss": 0.52,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861359642252319e-06,
+      "loss": 0.5244,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85912167350657e-06,
+      "loss": 0.5282,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856882853348535e-06,
+      "loss": 0.4754,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854643182444907e-06,
+      "loss": 0.5398,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852402661462635e-06,
+      "loss": 0.5261,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850161291068915e-06,
+      "loss": 0.5881,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847919071931202e-06,
+      "loss": 0.5005,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845676004717204e-06,
+      "loss": 0.606,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843432090094879e-06,
+      "loss": 0.5493,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841187328732436e-06,
+      "loss": 0.5256,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83894172129834e-06,
+      "loss": 0.5134,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836695268461306e-06,
+      "loss": 0.5532,
+      "step": 1952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8344479708903e-06,
+      "loss": 0.6016,
+      "step": 1953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832199829254542e-06,
+      "loss": 0.5428,
+      "step": 1954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8299508442235e-06,
+      "loss": 0.5181,
+      "step": 1955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827701016466895e-06,
+      "loss": 0.5817,
+      "step": 1956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8254503466547e-06,
+      "loss": 0.5865,
+      "step": 1957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823198835457135e-06,
+      "loss": 0.5271,
+      "step": 1958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820946483544676e-06,
+      "loss": 0.5267,
+      "step": 1959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818693291588047e-06,
+      "loss": 0.6185,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816439260258219e-06,
+      "loss": 0.551,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814184390226415e-06,
+      "loss": 0.547,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811928682164114e-06,
+      "loss": 0.5213,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809672136743031e-06,
+      "loss": 0.5021,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807414754635145e-06,
+      "loss": 0.1635,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805156536512674e-06,
+      "loss": 0.5316,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802897483048088e-06,
+      "loss": 0.5369,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80063759491411e-06,
+      "loss": 0.5634,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798376872783703e-06,
+      "loss": 0.5307,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796115317330085e-06,
+      "loss": 0.5599,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793852929226721e-06,
+      "loss": 0.5263,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791589709147322e-06,
+      "loss": 0.478,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789325657765848e-06,
+      "loss": 0.582,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787060775756506e-06,
+      "loss": 0.4464,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784795063793753e-06,
+      "loss": 0.5829,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78252852255229e-06,
+      "loss": 0.522,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780261152707063e-06,
+      "loss": 0.5161,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777992954933271e-06,
+      "loss": 0.5088,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775723929906358e-06,
+      "loss": 0.5064,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773454078302008e-06,
+      "loss": 0.4951,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771183400796156e-06,
+      "loss": 0.5603,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768911898064986e-06,
+      "loss": 0.5071,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766639570784921e-06,
+      "loss": 0.1571,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764366419632636e-06,
+      "loss": 0.5319,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762092445285047e-06,
+      "loss": 0.5965,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759817648419316e-06,
+      "loss": 0.523,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75754202971285e-06,
+      "loss": 0.5531,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755265589843301e-06,
+      "loss": 0.5119,
+      "step": 1988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752988329488568e-06,
+      "loss": 0.1411,
+      "step": 1989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75071024932679e-06,
+      "loss": 0.545,
+      "step": 1990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74843135003635e-06,
+      "loss": 0.6333,
+      "step": 1991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74615163229588e-06,
+      "loss": 0.5464,
+      "step": 1992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743871096784252e-06,
+      "loss": 0.522,
+      "step": 1993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741589744180585e-06,
+      "loss": 0.1631,
+      "step": 1994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739307575164233e-06,
+      "loss": 0.5131,
+      "step": 1995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737024590414802e-06,
+      "loss": 0.1779,
+      "step": 1996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734740790612137e-06,
+      "loss": 0.6357,
+      "step": 1997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732456176436323e-06,
+      "loss": 0.4771,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730170748567699e-06,
+      "loss": 0.526,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72788450768683e-06,
+      "loss": 0.5152,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725597454474536e-06,
+      "loss": 0.189,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723309589611871e-06,
+      "loss": 0.5528,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721020913780137e-06,
+      "loss": 0.5927,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718731427660873e-06,
+      "loss": 0.5784,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71644113193586e-06,
+      "loss": 0.6002,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71415002728712e-06,
+      "loss": 0.5065,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711858114396917e-06,
+      "loss": 0.5973,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709565393947759e-06,
+      "loss": 0.1838,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707271866622386e-06,
+      "loss": 0.515,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704977533103784e-06,
+      "loss": 0.5115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702682394075181e-06,
+      "loss": 0.6157,
+      "step": 2011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70038645022004e-06,
+      "loss": 0.5567,
+      "step": 2012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698089702222062e-06,
+      "loss": 0.5609,
+      "step": 2013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695792150765198e-06,
+      "loss": 0.5023,
+      "step": 2014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693493796533627e-06,
+      "loss": 0.5479,
+      "step": 2015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691194640211771e-06,
+      "loss": 0.5845,
+      "step": 2016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688894682484294e-06,
+      "loss": 0.5503,
+      "step": 2017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686593924036097e-06,
+      "loss": 0.5815,
+      "step": 2018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684292365552313e-06,
+      "loss": 0.4696,
+      "step": 2019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681990007718323e-06,
+      "loss": 0.5423,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679686851219738e-06,
+      "loss": 0.578,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677382896742417e-06,
+      "loss": 0.5858,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675078144972444e-06,
+      "loss": 0.5036,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672772596596149e-06,
+      "loss": 0.5703,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670466252300096e-06,
+      "loss": 0.5313,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668159112771085e-06,
+      "loss": 0.561,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66585117869616e-06,
+      "loss": 0.5091,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663542450762591e-06,
+      "loss": 0.5319,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661232929657894e-06,
+      "loss": 0.596,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658922616069811e-06,
+      "loss": 0.5592,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656611510686332e-06,
+      "loss": 0.1681,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654299614195672e-06,
+      "loss": 0.5191,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651986927286287e-06,
+      "loss": 0.5669,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649673450646871e-06,
+      "loss": 0.4807,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647359184966346e-06,
+      "loss": 0.5224,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.645044130933874e-06,
+      "loss": 0.4818,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64272828923885e-06,
+      "loss": 0.5707,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640411660570905e-06,
+      "loss": 0.5064,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638094245619901e-06,
+      "loss": 0.4843,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63577604507594e-06,
+      "loss": 0.4884,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63345705962935e-06,
+      "loss": 0.5774,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631137289970703e-06,
+      "loss": 0.4986,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628816736790792e-06,
+      "loss": 0.5635,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626495400780656e-06,
+      "loss": 0.5264,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624173282631557e-06,
+      "loss": 0.5516,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621850383034997e-06,
+      "loss": 0.5616,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6195267026827045e-06,
+      "loss": 0.5254,
+      "step": 2047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.617202242266647e-06,
+      "loss": 0.5614,
+      "step": 2048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61487700247902e-06,
+      "loss": 0.5108,
+      "step": 2049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612550984012252e-06,
+      "loss": 0.6023,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610224187559001e-06,
+      "loss": 0.5564,
+      "step": 2051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607896613812165e-06,
+      "loss": 0.5674,
+      "step": 2052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6055682634648596e-06,
+      "loss": 0.532,
+      "step": 2053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603239137210447e-06,
+      "loss": 0.5372,
+      "step": 2054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600909235742511e-06,
+      "loss": 0.589,
+      "step": 2055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5985785597548675e-06,
+      "loss": 0.5063,
+      "step": 2056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596247109941566e-06,
+      "loss": 0.5888,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59391488699688e-06,
+      "loss": 0.5734,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591581891615323e-06,
+      "loss": 0.5683,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589248124491627e-06,
+      "loss": 0.6197,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586913586320768e-06,
+      "loss": 0.6029,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584578277797935e-06,
+      "loss": 0.5536,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582242199618561e-06,
+      "loss": 0.5276,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579905352478301e-06,
+      "loss": 0.6438,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577567737073039e-06,
+      "loss": 0.607,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57522935409889e-06,
+      "loss": 0.5866,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572890204252194e-06,
+      "loss": 0.1777,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5705502882295265e-06,
+      "loss": 0.564,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568209606727684e-06,
+      "loss": 0.5774,
+      "step": 2069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56586816044369e-06,
+      "loss": 0.5773,
+      "step": 2070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563525950074806e-06,
+      "loss": 0.5457,
+      "step": 2071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561182976318509e-06,
+      "loss": 0.1363,
+      "step": 2072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558839239872511e-06,
+      "loss": 0.602,
+      "step": 2073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556494741434749e-06,
+      "loss": 0.5111,
+      "step": 2074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554149481703386e-06,
+      "loss": 0.5515,
+      "step": 2075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551803461376811e-06,
+      "loss": 0.4692,
+      "step": 2076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549456681153642e-06,
+      "loss": 0.5482,
+      "step": 2077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547109141732723e-06,
+      "loss": 0.5652,
+      "step": 2078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544760843813122e-06,
+      "loss": 0.4938,
+      "step": 2079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542411788094135e-06,
+      "loss": 0.4979,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540061975275281e-06,
+      "loss": 0.5377,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53771140605631e-06,
+      "loss": 0.4959,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535360081137187e-06,
+      "loss": 0.5673,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533008001218115e-06,
+      "loss": 0.5739,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530655166999512e-06,
+      "loss": 0.5278,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528301579182024e-06,
+      "loss": 0.5217,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.525947238466522e-06,
+      "loss": 0.5685,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523592145554102e-06,
+      "loss": 0.597,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52123630114608e-06,
+      "loss": 0.5366,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5188797059439976e-06,
+      "loss": 0.4984,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516522360649625e-06,
+      "loss": 0.5226,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514164265964947e-06,
+      "loss": 0.5438,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5118054225921796e-06,
+      "loss": 0.5872,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509445831233755e-06,
+      "loss": 0.5959,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507085492592334e-06,
+      "loss": 0.5941,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504724407370796e-06,
+      "loss": 0.5694,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502362576272242e-06,
+      "loss": 0.5201,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.5516,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497636679257618e-06,
+      "loss": 0.5772,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495272614748861e-06,
+      "loss": 0.5972,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492907807177721e-06,
+      "loss": 0.5488,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490542257248411e-06,
+      "loss": 0.5847,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488175965665361e-06,
+      "loss": 0.4666,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485808933133225e-06,
+      "loss": 0.531,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483441160356879e-06,
+      "loss": 0.5616,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481072648041414e-06,
+      "loss": 0.5231,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478703396892148e-06,
+      "loss": 0.5431,
+      "step": 2107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476333407614612e-06,
+      "loss": 0.6195,
+      "step": 2108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4739626809145636e-06,
+      "loss": 0.5366,
+      "step": 2109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471591217497978e-06,
+      "loss": 0.5574,
+      "step": 2110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469219018071045e-06,
+      "loss": 0.569,
+      "step": 2111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4668460833401775e-06,
+      "loss": 0.5597,
+      "step": 2112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464472414012012e-06,
+      "loss": 0.5239,
+      "step": 2113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462098010793392e-06,
+      "loss": 0.5543,
+      "step": 2114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4597228743913916e-06,
+      "loss": 0.4899,
+      "step": 2115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457347005513297e-06,
+      "loss": 0.5224,
+      "step": 2116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454970404866612e-06,
+      "loss": 0.5966,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45259307315906e-06,
+      "loss": 0.5507,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450215011098585e-06,
+      "loss": 0.5739,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44783621939334e-06,
+      "loss": 0.5575,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445456698751707e-06,
+      "loss": 0.4908,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443076449882274e-06,
+      "loss": 0.4977,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440695473493854e-06,
+      "loss": 0.5535,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438313770295471e-06,
+      "loss": 0.5472,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43593134099637e-06,
+      "loss": 0.5594,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433548186306005e-06,
+      "loss": 0.5164,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431164306934058e-06,
+      "loss": 0.5133,
+      "step": 2127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4287797035904155e-06,
+      "loss": 0.5819,
+      "step": 2128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4263943769851845e-06,
+      "loss": 0.5804,
+      "step": 2129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424008327828688e-06,
+      "loss": 0.6601,
+      "step": 2130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4216215568314595e-06,
+      "loss": 0.616,
+      "step": 2131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419234064704256e-06,
+      "loss": 0.5242,
+      "step": 2132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41684585215804e-06,
+      "loss": 0.5295,
+      "step": 2133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414456919903996e-06,
+      "loss": 0.5844,
+      "step": 2134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412067268653516e-06,
+      "loss": 0.5034,
+      "step": 2135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409676899118213e-06,
+      "loss": 0.5831,
+      "step": 2136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407285812009908e-06,
+      "loss": 0.5589,
+      "step": 2137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404894008040638e-06,
+      "loss": 0.5361,
+      "step": 2138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402501487922657e-06,
+      "loss": 0.5457,
+      "step": 2139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400108252368424e-06,
+      "loss": 0.5177,
+      "step": 2140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397714302090621e-06,
+      "loss": 0.5865,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395319637802134e-06,
+      "loss": 0.5692,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3929242602160666e-06,
+      "loss": 0.5407,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390528170045734e-06,
+      "loss": 0.4833,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388131368004663e-06,
+      "loss": 0.5418,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385733854806593e-06,
+      "loss": 0.5828,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383335631165475e-06,
+      "loss": 0.525,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3809366977954685e-06,
+      "loss": 0.5289,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3785370554109524e-06,
+      "loss": 0.5741,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3761367047265075e-06,
+      "loss": 0.5607,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373735646456932e-06,
+      "loss": 0.5722,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371333881317229e-06,
+      "loss": 0.6144,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368931410022619e-06,
+      "loss": 0.5342,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366528233288526e-06,
+      "loss": 0.5544,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3641243518305915e-06,
+      "loss": 0.5084,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361719766364659e-06,
+      "loss": 0.5716,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359314477606787e-06,
+      "loss": 0.5497,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356908486273242e-06,
+      "loss": 0.5379,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3545017930804975e-06,
+      "loss": 0.1701,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352094398745242e-06,
+      "loss": 0.5982,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349686303984363e-06,
+      "loss": 0.5409,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347277509514969e-06,
+      "loss": 0.5415,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344868016054364e-06,
+      "loss": 0.5,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342457824320072e-06,
+      "loss": 0.4893,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340046935029813e-06,
+      "loss": 0.5222,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337635348901527e-06,
+      "loss": 0.6209,
+      "step": 2166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.335223066653353e-06,
+      "loss": 0.1769,
+      "step": 2167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332810089003639e-06,
+      "loss": 0.4983,
+      "step": 2168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3303964166709414e-06,
+      "loss": 0.5582,
+      "step": 2169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327982050374025e-06,
+      "loss": 0.5817,
+      "step": 2170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.325566990831857e-06,
+      "loss": 0.5237,
+      "step": 2171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323151238763614e-06,
+      "loss": 0.499,
+      "step": 2172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.320734794888677e-06,
+      "loss": 0.5304,
+      "step": 2173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318317659926637e-06,
+      "loss": 0.4898,
+      "step": 2174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315899834597285e-06,
+      "loss": 0.5147,
+      "step": 2175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313481319620624e-06,
+      "loss": 0.5795,
+      "step": 2176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311062115716853e-06,
+      "loss": 0.567,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308642223606388e-06,
+      "loss": 0.5212,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30622164400984e-06,
+      "loss": 0.559,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303800377648031e-06,
+      "loss": 0.4927,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301378425241984e-06,
+      "loss": 0.579,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298955787512928e-06,
+      "loss": 0.5701,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296532465182297e-06,
+      "loss": 0.5444,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294108458971724e-06,
+      "loss": 0.5126,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291683769603051e-06,
+      "loss": 0.4639,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289258397798322e-06,
+      "loss": 0.5143,
+      "step": 2186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286832344279784e-06,
+      "loss": 0.51,
+      "step": 2187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284405609769886e-06,
+      "loss": 0.5482,
+      "step": 2188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28197819499128e-06,
+      "loss": 0.559,
+      "step": 2189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279550100666824e-06,
+      "loss": 0.5412,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277121327519569e-06,
+      "loss": 0.6219,
+      "step": 2191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274691876272783e-06,
+      "loss": 0.5709,
+      "step": 2192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272261747649922e-06,
+      "loss": 0.5877,
+      "step": 2193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26983094237465e-06,
+      "loss": 0.536,
+      "step": 2194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2673994611708336e-06,
+      "loss": 0.1665,
+      "step": 2195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264967304762538e-06,
+      "loss": 0.5933,
+      "step": 2196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262534473874029e-06,
+      "loss": 0.5877,
+      "step": 2197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260100969229776e-06,
+      "loss": 0.5729,
+      "step": 2198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257666791554448e-06,
+      "loss": 0.6154,
+      "step": 2199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255231941572913e-06,
+      "loss": 0.5577,
+      "step": 2200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252796420010239e-06,
+      "loss": 0.5603,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2503602275917e-06,
+      "loss": 0.49,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2479233650427595e-06,
+      "loss": 0.545,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2454858330890885e-06,
+      "loss": 0.5542,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243047632456557e-06,
+      "loss": 0.5564,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240608763871229e-06,
+      "loss": 0.5663,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2381692280593696e-06,
+      "loss": 0.5,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235729025747447e-06,
+      "loss": 0.5383,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2332881576621206e-06,
+      "loss": 0.5372,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2308466245302545e-06,
+      "loss": 0.5168,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228404427078908e-06,
+      "loss": 0.5618,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225961566035335e-06,
+      "loss": 0.6042,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223518042126996e-06,
+      "loss": 0.4808,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221073856081538e-06,
+      "loss": 0.5093,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218629008626814e-06,
+      "loss": 0.6001,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216183500490868e-06,
+      "loss": 0.5748,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213737332401945e-06,
+      "loss": 0.5967,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2112905050884816e-06,
+      "loss": 0.1463,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208843019279117e-06,
+      "loss": 0.4932,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206394875702683e-06,
+      "loss": 0.5813,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203946075088205e-06,
+      "loss": 0.5376,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20149661816491e-06,
+      "loss": 0.4359,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1990465056622125e-06,
+      "loss": 0.5239,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196595738309733e-06,
+      "loss": 0.5724,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194144316837272e-06,
+      "loss": 0.5159,
+      "step": 2225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.191692241974844e-06,
+      "loss": 0.5588,
+      "step": 2226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18923951445264e-06,
+      "loss": 0.5601,
+      "step": 2227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186786135001055e-06,
+      "loss": 0.1898,
+      "step": 2228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184332104350676e-06,
+      "loss": 0.526,
+      "step": 2229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181877423232285e-06,
+      "loss": 0.5676,
+      "step": 2230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.179422092376856e-06,
+      "loss": 0.5601,
+      "step": 2231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176966112515557e-06,
+      "loss": 0.5512,
+      "step": 2232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174509484379752e-06,
+      "loss": 0.5056,
+      "step": 2233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172052208700989e-06,
+      "loss": 0.4865,
+      "step": 2234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169594286211024e-06,
+      "loss": 0.4906,
+      "step": 2235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16713571764179e-06,
+      "loss": 0.4766,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164676503725423e-06,
+      "loss": 0.5786,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162216645194246e-06,
+      "loss": 0.5636,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159756142780776e-06,
+      "loss": 0.5243,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157294997217721e-06,
+      "loss": 0.4236,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154833209237982e-06,
+      "loss": 0.4931,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152370779574647e-06,
+      "loss": 0.6027,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149907708961002e-06,
+      "loss": 0.5026,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147443998130517e-06,
+      "loss": 0.4995,
+      "step": 2244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144979647816856e-06,
+      "loss": 0.5296,
+      "step": 2245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1425146587538765e-06,
+      "loss": 0.5593,
+      "step": 2246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140049031675619e-06,
+      "loss": 0.525,
+      "step": 2247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137582767316319e-06,
+      "loss": 0.5934,
+      "step": 2248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1351158664104004e-06,
+      "loss": 0.4962,
+      "step": 2249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132648329692478e-06,
+      "loss": 0.5218,
+      "step": 2250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1301801578973525e-06,
+      "loss": 0.5381,
+      "step": 2251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127711351760019e-06,
+      "loss": 0.5067,
+      "step": 2252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125241912015654e-06,
+      "loss": 0.5612,
+      "step": 2253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1227718393996325e-06,
+      "loss": 0.1621,
+      "step": 2254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120301134647508e-06,
+      "loss": 0.549,
+      "step": 2255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117829798495032e-06,
+      "loss": 0.56,
+      "step": 2256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115357831678133e-06,
+      "loss": 0.5428,
+      "step": 2257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112885234932937e-06,
+      "loss": 0.4721,
+      "step": 2258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1104120089957525e-06,
+      "loss": 0.4946,
+      "step": 2259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107938154603076e-06,
+      "loss": 0.5893,
+      "step": 2260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1054636724915945e-06,
+      "loss": 0.5376,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1029885633981755e-06,
+      "loss": 0.4764,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.100512828059881e-06,
+      "loss": 0.4868,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098036467213951e-06,
+      "loss": 0.5253,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095559481597822e-06,
+      "loss": 0.5361,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093081871949104e-06,
+      "loss": 0.5512,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090603639005603e-06,
+      "loss": 0.6088,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088124783505308e-06,
+      "loss": 0.5433,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085645306186391e-06,
+      "loss": 0.4866,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083165207787211e-06,
+      "loss": 0.4681,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080684489046312e-06,
+      "loss": 0.5591,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078203150702423e-06,
+      "loss": 0.1479,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0757211934944555e-06,
+      "loss": 0.5705,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073238618161508e-06,
+      "loss": 0.4952,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0707554254428615e-06,
+      "loss": 0.5585,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06827161607798e-06,
+      "loss": 0.5952,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065787190806512e-06,
+      "loss": 0.5932,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063302150368293e-06,
+      "loss": 0.5246,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060816495503335e-06,
+      "loss": 0.5939,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058330226951838e-06,
+      "loss": 0.5555,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055843345454183e-06,
+      "loss": 0.521,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053355851750933e-06,
+      "loss": 0.4784,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0508677465828334e-06,
+      "loss": 0.5056,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048379030690814e-06,
+      "loss": 0.5171,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045889704815982e-06,
+      "loss": 0.566,
+      "step": 2285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043399769699633e-06,
+      "loss": 0.5615,
+      "step": 2286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0409092260832346e-06,
+      "loss": 0.5227,
+      "step": 2287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038418074708444e-06,
+      "loss": 0.5347,
+      "step": 2288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035926316317098e-06,
+      "loss": 0.6059,
+      "step": 2289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033433951651209e-06,
+      "loss": 0.5824,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030940981452977e-06,
+      "loss": 0.5553,
+      "step": 2291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028447406464774e-06,
+      "loss": 0.537,
+      "step": 2292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025953227429163e-06,
+      "loss": 0.5349,
+      "step": 2293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023458445088875e-06,
+      "loss": 0.6021,
+      "step": 2294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02096306018683e-06,
+      "loss": 0.625,
+      "step": 2295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01846707346612e-06,
+      "loss": 0.5305,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015970485670025e-06,
+      "loss": 0.5491,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013473297541998e-06,
+      "loss": 0.5337,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010975509825667e-06,
+      "loss": 0.5274,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008477123264849e-06,
+      "loss": 0.5114,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00597813860353e-06,
+      "loss": 0.5174,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00347855658588e-06,
+      "loss": 0.5329,
+      "step": 2302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000978377956244e-06,
+      "loss": 0.5302,
+      "step": 2303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998477603459145e-06,
+      "loss": 0.5172,
+      "step": 2304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9959762338392835e-06,
+      "loss": 0.4715,
+      "step": 2305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.993474269841539e-06,
+      "loss": 0.5629,
+      "step": 2306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990971712210966e-06,
+      "loss": 0.5293,
+      "step": 2307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988468561692794e-06,
+      "loss": 0.5312,
+      "step": 2308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985964819032434e-06,
+      "loss": 0.4701,
+      "step": 2309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983460484975471e-06,
+      "loss": 0.1542,
+      "step": 2310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980955560267663e-06,
+      "loss": 0.1688,
+      "step": 2311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978450045654947e-06,
+      "loss": 0.4961,
+      "step": 2312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97594394188344e-06,
+      "loss": 0.5734,
+      "step": 2313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9734372496994255e-06,
+      "loss": 0.4835,
+      "step": 2314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970929969849365e-06,
+      "loss": 0.5104,
+      "step": 2315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968422103079899e-06,
+      "loss": 0.5422,
+      "step": 2316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965913650137842e-06,
+      "loss": 0.1574,
+      "step": 2317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963404611770178e-06,
+      "loss": 0.5716,
+      "step": 2318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9608949887240715e-06,
+      "loss": 0.5519,
+      "step": 2319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958384781746855e-06,
+      "loss": 0.6065,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955873991586042e-06,
+      "loss": 0.5253,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953362618989312e-06,
+      "loss": 0.5096,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950850664704526e-06,
+      "loss": 0.5573,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948338129479708e-06,
+      "loss": 0.5517,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9458250140630675e-06,
+      "loss": 0.5713,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943311319202976e-06,
+      "loss": 0.5601,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940797045647984e-06,
+      "loss": 0.561,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93828219414681e-06,
+      "loss": 0.6394,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93576676544835e-06,
+      "loss": 0.5093,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933250760301665e-06,
+      "loss": 0.5468,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930734179455993e-06,
+      "loss": 0.6829,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928217023660741e-06,
+      "loss": 0.5561,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925699293665491e-06,
+      "loss": 0.5582,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923180990219989e-06,
+      "loss": 0.5443,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9206621140741544e-06,
+      "loss": 0.5384,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918142665978086e-06,
+      "loss": 0.5263,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915622646682037e-06,
+      "loss": 0.5417,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913102056936445e-06,
+      "loss": 0.1455,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910580897491908e-06,
+      "loss": 0.5514,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908059169099198e-06,
+      "loss": 0.5042,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905536872509257e-06,
+      "loss": 0.5248,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903014008473192e-06,
+      "loss": 0.5485,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.900490577742285e-06,
+      "loss": 0.1593,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897966581067981e-06,
+      "loss": 0.4981,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.895442019201898e-06,
+      "loss": 0.5509,
+      "step": 2345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892916892895818e-06,
+      "loss": 0.5101,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890391202901696e-06,
+      "loss": 0.5309,
+      "step": 2347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887864949971651e-06,
+      "loss": 0.5268,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88533813485797e-06,
+      "loss": 0.5845,
+      "step": 2349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882810758313112e-06,
+      "loss": 0.4964,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880282821089695e-06,
+      "loss": 0.5332,
+      "step": 2351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877754323940512e-06,
+      "loss": 0.5054,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875225267618518e-06,
+      "loss": 0.6093,
+      "step": 2353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872695652876834e-06,
+      "loss": 0.5157,
+      "step": 2354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870165480468752e-06,
+      "loss": 0.5643,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867634751147725e-06,
+      "loss": 0.4891,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8651034656673755e-06,
+      "loss": 0.4683,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8625716247814876e-06,
+      "loss": 0.5637,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860039229244018e-06,
+      "loss": 0.5692,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857506279809078e-06,
+      "loss": 0.1803,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854972777230954e-06,
+      "loss": 0.5719,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852438722264091e-06,
+      "loss": 0.6027,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8499041156631e-06,
+      "loss": 0.5002,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8473689581827585e-06,
+      "loss": 0.6082,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.844833250578003e-06,
+      "loss": 0.5368,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8422969936039405e-06,
+      "loss": 0.6104,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839760188015836e-06,
+      "loss": 0.503,
+      "step": 2367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8372228345691206e-06,
+      "loss": 0.5495,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834684934019387e-06,
+      "loss": 0.5294,
+      "step": 2369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832146487122392e-06,
+      "loss": 0.541,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829607494634055e-06,
+      "loss": 0.5745,
+      "step": 2371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827067957310458e-06,
+      "loss": 0.5094,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824527875907845e-06,
+      "loss": 0.5196,
+      "step": 2373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82198725118262e-06,
+      "loss": 0.4883,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819446083891352e-06,
+      "loss": 0.4977,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816904374790771e-06,
+      "loss": 0.4985,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814362124637768e-06,
+      "loss": 0.5734,
+      "step": 2377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811819334189391e-06,
+      "loss": 0.5581,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8092760042028576e-06,
+      "loss": 0.54,
+      "step": 2379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806732135435539e-06,
+      "loss": 0.5415,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804187728644967e-06,
+      "loss": 0.4874,
+      "step": 2381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801642784588837e-06,
+      "loss": 0.5391,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7990973040250055e-06,
+      "loss": 0.5644,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7965512877114826e-06,
+      "loss": 0.1635,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7940047364064444e-06,
+      "loss": 0.4888,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791457650868219e-06,
+      "loss": 0.547,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788910031855303e-06,
+      "loss": 0.5587,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786361880126344e-06,
+      "loss": 0.5788,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783813196440154e-06,
+      "loss": 0.5104,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781263981555697e-06,
+      "loss": 0.5141,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7787142362321e-06,
+      "loss": 0.583,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7761639612286485e-06,
+      "loss": 0.5469,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773613157304781e-06,
+      "loss": 0.5092,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7710618252201e-06,
+      "loss": 0.4825,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768509965734358e-06,
+      "loss": 0.5114,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7659575796074705e-06,
+      "loss": 0.4757,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763404667599507e-06,
+      "loss": 0.4799,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760851230470696e-06,
+      "loss": 0.5376,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7582972689814156e-06,
+      "loss": 0.5544,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7557427838922106e-06,
+      "loss": 0.4895,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753187775963773e-06,
+      "loss": 0.5252,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750632245956954e-06,
+      "loss": 0.5293,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748076194632762e-06,
+      "loss": 0.5307,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.745519622752355e-06,
+      "loss": 0.6078,
+      "step": 2404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742962531077053e-06,
+      "loss": 0.5615,
+      "step": 2405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7404049203683254e-06,
+      "loss": 0.5037,
+      "step": 2406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737846791387799e-06,
+      "loss": 0.5713,
+      "step": 2407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735288144897253e-06,
+      "loss": 0.4881,
+      "step": 2408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732728981658623e-06,
+      "loss": 0.534,
+      "step": 2409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730169302433996e-06,
+      "loss": 0.4634,
+      "step": 2410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727609107985615e-06,
+      "loss": 0.5029,
+      "step": 2411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725048399075875e-06,
+      "loss": 0.6163,
+      "step": 2412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722487176467325e-06,
+      "loss": 0.1475,
+      "step": 2413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719925440922665e-06,
+      "loss": 0.5174,
+      "step": 2414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7173631932047525e-06,
+      "loss": 0.5124,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71480043407659e-06,
+      "loss": 0.5109,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712237164301338e-06,
+      "loss": 0.5286,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709673384642308e-06,
+      "loss": 0.5327,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707109095862965e-06,
+      "loss": 0.5359,
+      "step": 2419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704544298726921e-06,
+      "loss": 0.5066,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701978993997942e-06,
+      "loss": 0.5279,
+      "step": 2421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699413182439946e-06,
+      "loss": 0.6033,
+      "step": 2422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696846864817e-06,
+      "loss": 0.4728,
+      "step": 2423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694280041893326e-06,
+      "loss": 0.5265,
+      "step": 2424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691712714433289e-06,
+      "loss": 0.5363,
+      "step": 2425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68914488320141e-06,
+      "loss": 0.5426,
+      "step": 2426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686576548962358e-06,
+      "loss": 0.5859,
+      "step": 2427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684007712480955e-06,
+      "loss": 0.5717,
+      "step": 2428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681438374522167e-06,
+      "loss": 0.5638,
+      "step": 2429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678868535851113e-06,
+      "loss": 0.5057,
+      "step": 2430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6762981972330595e-06,
+      "loss": 0.5318,
+      "step": 2431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673727359433424e-06,
+      "loss": 0.5347,
+      "step": 2432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6711560232177705e-06,
+      "loss": 0.5061,
+      "step": 2433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.668584189351812e-06,
+      "loss": 0.5097,
+      "step": 2434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66601185860141e-06,
+      "loss": 0.5294,
+      "step": 2435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6634390317325745e-06,
+      "loss": 0.5055,
+      "step": 2436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660865709511462e-06,
+      "loss": 0.1471,
+      "step": 2437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6582918927043774e-06,
+      "loss": 0.187,
+      "step": 2438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655717582077773e-06,
+      "loss": 0.1939,
+      "step": 2439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.653142778398247e-06,
+      "loss": 0.5854,
+      "step": 2440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650567482432546e-06,
+      "loss": 0.5102,
+      "step": 2441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647991694947562e-06,
+      "loss": 0.6175,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645415416710333e-06,
+      "loss": 0.6441,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642838648488043e-06,
+      "loss": 0.5738,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640261391048028e-06,
+      "loss": 0.4613,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637683645157757e-06,
+      "loss": 0.505,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635105411584859e-06,
+      "loss": 0.5621,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632526691097096e-06,
+      "loss": 0.5947,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629947484462382e-06,
+      "loss": 0.4712,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627367792448775e-06,
+      "loss": 0.5292,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624787615824475e-06,
+      "loss": 0.5767,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622206955357828e-06,
+      "loss": 0.5374,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619625811817325e-06,
+      "loss": 0.5443,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617044185971599e-06,
+      "loss": 0.5021,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6144620785894255e-06,
+      "loss": 0.525,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611879490439729e-06,
+      "loss": 0.5434,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6092964222915686e-06,
+      "loss": 0.5081,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6067128749141575e-06,
+      "loss": 0.5339,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6041288490768385e-06,
+      "loss": 0.5673,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601544345549109e-06,
+      "loss": 0.4732,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598959365100599e-06,
+      "loss": 0.5524,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596373908501089e-06,
+      "loss": 0.5534,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593787976520494e-06,
+      "loss": 0.5986,
+      "step": 2463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.591201569928875e-06,
+      "loss": 0.5194,
+      "step": 2464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588614689496431e-06,
+      "loss": 0.4755,
+      "step": 2465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586027335993507e-06,
+      "loss": 0.5275,
+      "step": 2466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583439510190585e-06,
+      "loss": 0.534,
+      "step": 2467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580851212858285e-06,
+      "loss": 0.5062,
+      "step": 2468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5782624447673796e-06,
+      "loss": 0.5351,
+      "step": 2469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575673206688764e-06,
+      "loss": 0.484,
+      "step": 2470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573083499393488e-06,
+      "loss": 0.5578,
+      "step": 2471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570493323652733e-06,
+      "loss": 0.5684,
+      "step": 2472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567902680237823e-06,
+      "loss": 0.5421,
+      "step": 2473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56531156992022e-06,
+      "loss": 0.5857,
+      "step": 2474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562719993471528e-06,
+      "loss": 0.5838,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560127951663484e-06,
+      "loss": 0.5788,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557535445267971e-06,
+      "loss": 0.4702,
+      "step": 2477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554942475057003e-06,
+      "loss": 0.5461,
+      "step": 2478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552349041802737e-06,
+      "loss": 0.5252,
+      "step": 2479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549755146277468e-06,
+      "loss": 0.4668,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547160789253623e-06,
+      "loss": 0.5417,
+      "step": 2481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.544565971503776e-06,
+      "loss": 0.5946,
+      "step": 2482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.541970693800628e-06,
+      "loss": 0.5136,
+      "step": 2483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539374956917024e-06,
+      "loss": 0.5568,
+      "step": 2484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536778761625942e-06,
+      "loss": 0.161,
+      "step": 2485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534182108700502e-06,
+      "loss": 0.5065,
+      "step": 2486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531584998913949e-06,
+      "loss": 0.6045,
+      "step": 2487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.528987433039679e-06,
+      "loss": 0.5074,
+      "step": 2488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52638941185121e-06,
+      "loss": 0.5607,
+      "step": 2489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.523790936122206e-06,
+      "loss": 0.5232,
+      "step": 2490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521192006626459e-06,
+      "loss": 0.4911,
+      "step": 2491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518592624137901e-06,
+      "loss": 0.5406,
+      "step": 2492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515992789430598e-06,
+      "loss": 0.5493,
+      "step": 2493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513392503278746e-06,
+      "loss": 0.5116,
+      "step": 2494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510791766456681e-06,
+      "loss": 0.502,
+      "step": 2495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.508190579738872e-06,
+      "loss": 0.6179,
+      "step": 2496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505588943899923e-06,
+      "loss": 0.5392,
+      "step": 2497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5029868597145664e-06,
+      "loss": 0.5614,
+      "step": 2498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500384327957675e-06,
+      "loss": 0.5299,
+      "step": 2499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497781349404247e-06,
+      "loss": 0.5806,
+      "step": 2500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495177924829426e-06,
+      "loss": 0.5817,
+      "step": 2501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492574055008474e-06,
+      "loss": 0.5545,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489969740716795e-06,
+      "loss": 0.5066,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4873649827299215e-06,
+      "loss": 0.4929,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484759781823521e-06,
+      "loss": 0.5162,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482154138773389e-06,
+      "loss": 0.4959,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479548054355458e-06,
+      "loss": 0.5832,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.476941529345786e-06,
+      "loss": 0.5497,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474334564520566e-06,
+      "loss": 0.2278,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4717271606561215e-06,
+      "loss": 0.5824,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469119318528906e-06,
+      "loss": 0.5193,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466511038915503e-06,
+      "loss": 0.5542,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463902322592628e-06,
+      "loss": 0.501,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461293170337124e-06,
+      "loss": 0.576,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4586835829259676e-06,
+      "loss": 0.6047,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456073561136261e-06,
+      "loss": 0.5032,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453463105745238e-06,
+      "loss": 0.5291,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450852217530263e-06,
+      "loss": 0.5328,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448240897268823e-06,
+      "loss": 0.4761,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.445629145738543e-06,
+      "loss": 0.5553,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443016963717166e-06,
+      "loss": 0.5663,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440404351982574e-06,
+      "loss": 0.5178,
+      "step": 2522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.437791311312768e-06,
+      "loss": 0.5205,
+      "step": 2523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4351778424858815e-06,
+      "loss": 0.4877,
+      "step": 2524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432563946280175e-06,
+      "loss": 0.5918,
+      "step": 2525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429949623474034e-06,
+      "loss": 0.5637,
+      "step": 2526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427334874845974e-06,
+      "loss": 0.4721,
+      "step": 2527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4247197011746335e-06,
+      "loss": 0.6189,
+      "step": 2528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4221041032387846e-06,
+      "loss": 0.5384,
+      "step": 2529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419488081817315e-06,
+      "loss": 0.5657,
+      "step": 2530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41687163768925e-06,
+      "loss": 0.5416,
+      "step": 2531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.414254771633734e-06,
+      "loss": 0.5524,
+      "step": 2532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.411637484430037e-06,
+      "loss": 0.5498,
+      "step": 2533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.409019776857557e-06,
+      "loss": 0.5097,
+      "step": 2534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406401649695814e-06,
+      "loss": 0.5274,
+      "step": 2535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4037831037244595e-06,
+      "loss": 0.1532,
+      "step": 2536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4011641397232605e-06,
+      "loss": 0.5367,
+      "step": 2537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3985447584721164e-06,
+      "loss": 0.5544,
+      "step": 2538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3959249607510454e-06,
+      "loss": 0.5781,
+      "step": 2539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393304747340194e-06,
+      "loss": 0.5504,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3906841190198275e-06,
+      "loss": 0.6206,
+      "step": 2541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388063076570341e-06,
+      "loss": 0.5508,
+      "step": 2542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385441620772248e-06,
+      "loss": 0.1835,
+      "step": 2543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382819752406188e-06,
+      "loss": 0.5438,
+      "step": 2544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38019747225292e-06,
+      "loss": 0.527,
+      "step": 2545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.377574781093328e-06,
+      "loss": 0.526,
+      "step": 2546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374951679708421e-06,
+      "loss": 0.5487,
+      "step": 2547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372328168879325e-06,
+      "loss": 0.1657,
+      "step": 2548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369704249387289e-06,
+      "loss": 0.5736,
+      "step": 2549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367079922013688e-06,
+      "loss": 0.5813,
+      "step": 2550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.364455187540011e-06,
+      "loss": 0.6201,
+      "step": 2551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3618300467478765e-06,
+      "loss": 0.1687,
+      "step": 2552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359204500419019e-06,
+      "loss": 0.5101,
+      "step": 2553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356578549335295e-06,
+      "loss": 0.5832,
+      "step": 2554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353952194278681e-06,
+      "loss": 0.6202,
+      "step": 2555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3513254360312725e-06,
+      "loss": 0.4548,
+      "step": 2556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348698275375291e-06,
+      "loss": 0.5434,
+      "step": 2557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346070713093069e-06,
+      "loss": 0.1616,
+      "step": 2558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343442749967067e-06,
+      "loss": 0.4925,
+      "step": 2559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340814386779856e-06,
+      "loss": 0.1839,
+      "step": 2560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3381856243141375e-06,
+      "loss": 0.5401,
+      "step": 2561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33555646335272e-06,
+      "loss": 0.4719,
+      "step": 2562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3329269046785405e-06,
+      "loss": 0.4705,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330296949074648e-06,
+      "loss": 0.5713,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3276665973242114e-06,
+      "loss": 0.5496,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325035850210518e-06,
+      "loss": 0.5562,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322404708516975e-06,
+      "loss": 0.5829,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3197731730271025e-06,
+      "loss": 0.574,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317141244524543e-06,
+      "loss": 0.4917,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31450892379305e-06,
+      "loss": 0.568,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311876211616501e-06,
+      "loss": 0.5734,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309243108778883e-06,
+      "loss": 0.5749,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306609616064304e-06,
+      "loss": 0.5359,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303975734256986e-06,
+      "loss": 0.5157,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30134146414127e-06,
+      "loss": 0.4989,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298706806501607e-06,
+      "loss": 0.5106,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296071762122566e-06,
+      "loss": 0.4488,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293436331788836e-06,
+      "loss": 0.57,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2908005162852135e-06,
+      "loss": 0.5409,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288164316396615e-06,
+      "loss": 0.5431,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2855277329080655e-06,
+      "loss": 0.5193,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282890766604711e-06,
+      "loss": 0.57,
+      "step": 2582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280253418271809e-06,
+      "loss": 0.5761,
+      "step": 2583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27761568869473e-06,
+      "loss": 0.4734,
+      "step": 2584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274977578658956e-06,
+      "loss": 0.5393,
+      "step": 2585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272339088950088e-06,
+      "loss": 0.5301,
+      "step": 2586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269700220353836e-06,
+      "loss": 0.5442,
+      "step": 2587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26706097365602e-06,
+      "loss": 0.5446,
+      "step": 2588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.264421349642581e-06,
+      "loss": 0.5111,
+      "step": 2589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2617813490995636e-06,
+      "loss": 0.5698,
+      "step": 2590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.259140972813133e-06,
+      "loss": 0.1813,
+      "step": 2591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256500221569556e-06,
+      "loss": 0.5125,
+      "step": 2592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2538590961552216e-06,
+      "loss": 0.5592,
+      "step": 2593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251217597356623e-06,
+      "loss": 0.1606,
+      "step": 2594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248575725960367e-06,
+      "loss": 0.5231,
+      "step": 2595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.245933482753172e-06,
+      "loss": 0.582,
+      "step": 2596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2432908685218675e-06,
+      "loss": 0.546,
+      "step": 2597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24064788405339e-06,
+      "loss": 0.544,
+      "step": 2598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238004530134789e-06,
+      "loss": 0.4936,
+      "step": 2599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.235360807553227e-06,
+      "loss": 0.5367,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.232716717095969e-06,
+      "loss": 0.5185,
+      "step": 2601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230072259550397e-06,
+      "loss": 0.5259,
+      "step": 2602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227427435703997e-06,
+      "loss": 0.548,
+      "step": 2603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224782246344365e-06,
+      "loss": 0.5562,
+      "step": 2604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222136692259209e-06,
+      "loss": 0.5392,
+      "step": 2605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219490774236341e-06,
+      "loss": 0.5022,
+      "step": 2606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2168444930636866e-06,
+      "loss": 0.5598,
+      "step": 2607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214197849529274e-06,
+      "loss": 0.5133,
+      "step": 2608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211550844421244e-06,
+      "loss": 0.6231,
+      "step": 2609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208903478527842e-06,
+      "loss": 0.5394,
+      "step": 2610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2062557526374226e-06,
+      "loss": 0.5765,
+      "step": 2611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203607667538447e-06,
+      "loss": 0.5409,
+      "step": 2612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2009592240194825e-06,
+      "loss": 0.5234,
+      "step": 2613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198310422869205e-06,
+      "loss": 0.5267,
+      "step": 2614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195661264876394e-06,
+      "loss": 0.5452,
+      "step": 2615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19301175082994e-06,
+      "loss": 0.5035,
+      "step": 2616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190361881518834e-06,
+      "loss": 0.4907,
+      "step": 2617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187711657732177e-06,
+      "loss": 0.5445,
+      "step": 2618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185061080259175e-06,
+      "loss": 0.471,
+      "step": 2619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182410149889133e-06,
+      "loss": 0.5279,
+      "step": 2620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179758867411474e-06,
+      "loss": 0.6469,
+      "step": 2621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177107233615711e-06,
+      "loss": 0.5146,
+      "step": 2622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.174455249291476e-06,
+      "loss": 0.543,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171802915228491e-06,
+      "loss": 0.5716,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1691502322165965e-06,
+      "loss": 0.5333,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166497201045725e-06,
+      "loss": 0.5399,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163843822505916e-06,
+      "loss": 0.5359,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161190097387321e-06,
+      "loss": 0.5439,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15853602648018e-06,
+      "loss": 0.5947,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15588161057485e-06,
+      "loss": 0.4958,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153226850461777e-06,
+      "loss": 0.5373,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150571746931526e-06,
+      "loss": 0.5325,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147916300774749e-06,
+      "loss": 0.5721,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145260512782206e-06,
+      "loss": 0.4758,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142604383744761e-06,
+      "loss": 0.4828,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139947914453378e-06,
+      "loss": 0.5636,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1372911056991206e-06,
+      "loss": 0.5334,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134633958273154e-06,
+      "loss": 0.4519,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131976472966745e-06,
+      "loss": 0.5834,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129318650571263e-06,
+      "loss": 0.5433,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126660491878176e-06,
+      "loss": 0.5529,
+      "step": 2641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.124001997679049e-06,
+      "loss": 0.5565,
+      "step": 2642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121343168765551e-06,
+      "loss": 0.5555,
+      "step": 2643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.118684005929451e-06,
+      "loss": 0.6151,
+      "step": 2644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.116024509962613e-06,
+      "loss": 0.485,
+      "step": 2645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.113364681657006e-06,
+      "loss": 0.6191,
+      "step": 2646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110704521804694e-06,
+      "loss": 0.5566,
+      "step": 2647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108044031197841e-06,
+      "loss": 0.5662,
+      "step": 2648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10538321062871e-06,
+      "loss": 0.499,
+      "step": 2649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.102722060889659e-06,
+      "loss": 0.168,
+      "step": 2650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1000605827731494e-06,
+      "loss": 0.5244,
+      "step": 2651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.097398777071736e-06,
+      "loss": 0.6159,
+      "step": 2652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094736644578074e-06,
+      "loss": 0.4898,
+      "step": 2653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092074186084913e-06,
+      "loss": 0.522,
+      "step": 2654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.089411402385103e-06,
+      "loss": 0.5383,
+      "step": 2655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086748294271587e-06,
+      "loss": 0.5188,
+      "step": 2656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0840848625374105e-06,
+      "loss": 0.2108,
+      "step": 2657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0814211079757095e-06,
+      "loss": 0.559,
+      "step": 2658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078757031379716e-06,
+      "loss": 0.4769,
+      "step": 2659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076092633542765e-06,
+      "loss": 0.153,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.073427915258277e-06,
+      "loss": 0.5464,
+      "step": 2661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.070762877319779e-06,
+      "loss": 0.5041,
+      "step": 2662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068097520520883e-06,
+      "loss": 0.5399,
+      "step": 2663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065431845655302e-06,
+      "loss": 0.46,
+      "step": 2664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.062765853516842e-06,
+      "loss": 0.5011,
+      "step": 2665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060099544899404e-06,
+      "loss": 0.5455,
+      "step": 2666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.057432920596983e-06,
+      "loss": 0.5086,
+      "step": 2667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0547659814036664e-06,
+      "loss": 0.5623,
+      "step": 2668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052098728113638e-06,
+      "loss": 0.5419,
+      "step": 2669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049431161521174e-06,
+      "loss": 0.5248,
+      "step": 2670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046763282420645e-06,
+      "loss": 0.5893,
+      "step": 2671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044095091606511e-06,
+      "loss": 0.5663,
+      "step": 2672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041426589873328e-06,
+      "loss": 0.5366,
+      "step": 2673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038757778015746e-06,
+      "loss": 0.6301,
+      "step": 2674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0360886568285024e-06,
+      "loss": 0.5148,
+      "step": 2675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033419227106433e-06,
+      "loss": 0.5211,
+      "step": 2676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.030749489644458e-06,
+      "loss": 0.5007,
+      "step": 2677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028079445237597e-06,
+      "loss": 0.5065,
+      "step": 2678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025409094680956e-06,
+      "loss": 0.5569,
+      "step": 2679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0227384387697326e-06,
+      "loss": 0.5634,
+      "step": 2680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020067478299217e-06,
+      "loss": 0.5404,
+      "step": 2681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.017396214064789e-06,
+      "loss": 0.4695,
+      "step": 2682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0147246468619204e-06,
+      "loss": 0.5451,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01205277748617e-06,
+      "loss": 0.4724,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.009380606733191e-06,
+      "loss": 0.5502,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.006708135398721e-06,
+      "loss": 0.4579,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.004035364278593e-06,
+      "loss": 0.5418,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001362294168725e-06,
+      "loss": 0.4968,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998688925865124e-06,
+      "loss": 0.5203,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99601526016389e-06,
+      "loss": 0.5296,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993341297861206e-06,
+      "loss": 0.5441,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990667039753348e-06,
+      "loss": 0.4948,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.987992486636677e-06,
+      "loss": 0.5278,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985317639307644e-06,
+      "loss": 0.4894,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.982642498562784e-06,
+      "loss": 0.4762,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.979967065198726e-06,
+      "loss": 0.5733,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977291340012179e-06,
+      "loss": 0.5617,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.974615323799944e-06,
+      "loss": 0.1652,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.971939017358904e-06,
+      "loss": 0.5055,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.969262421486034e-06,
+      "loss": 0.4512,
+      "step": 2700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.966585536978391e-06,
+      "loss": 0.5547,
+      "step": 2701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.963908364633117e-06,
+      "loss": 0.458,
+      "step": 2702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.961230905247447e-06,
+      "loss": 0.4834,
+      "step": 2703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958553159618693e-06,
+      "loss": 0.562,
+      "step": 2704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955875128544256e-06,
+      "loss": 0.5192,
+      "step": 2705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.953196812821622e-06,
+      "loss": 0.5367,
+      "step": 2706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950518213248362e-06,
+      "loss": 0.5807,
+      "step": 2707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947839330622129e-06,
+      "loss": 0.5439,
+      "step": 2708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945160165740665e-06,
+      "loss": 0.5379,
+      "step": 2709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942480719401791e-06,
+      "loss": 0.1479,
+      "step": 2710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939800992403415e-06,
+      "loss": 0.5204,
+      "step": 2711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937120985543529e-06,
+      "loss": 0.4957,
+      "step": 2712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.934440699620204e-06,
+      "loss": 0.5833,
+      "step": 2713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9317601354316e-06,
+      "loss": 0.1692,
+      "step": 2714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929079293775957e-06,
+      "loss": 0.5491,
+      "step": 2715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.926398175451594e-06,
+      "loss": 0.6118,
+      "step": 2716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.923716781256921e-06,
+      "loss": 0.5514,
+      "step": 2717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921035111990421e-06,
+      "loss": 0.548,
+      "step": 2718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918353168450666e-06,
+      "loss": 0.6083,
+      "step": 2719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.915670951436305e-06,
+      "loss": 0.5065,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.912988461746071e-06,
+      "loss": 0.5785,
+      "step": 2721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910305700178778e-06,
+      "loss": 0.5225,
+      "step": 2722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.907622667533317e-06,
+      "loss": 0.5221,
+      "step": 2723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904939364608669e-06,
+      "loss": 0.5742,
+      "step": 2724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.902255792203882e-06,
+      "loss": 0.1838,
+      "step": 2725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899571951118099e-06,
+      "loss": 0.5272,
+      "step": 2726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896887842150531e-06,
+      "loss": 0.5739,
+      "step": 2727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8942034661004775e-06,
+      "loss": 0.5755,
+      "step": 2728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.891518823767311e-06,
+      "loss": 0.517,
+      "step": 2729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.888833915950483e-06,
+      "loss": 0.5614,
+      "step": 2730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8861487434495345e-06,
+      "loss": 0.5356,
+      "step": 2731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.883463307064071e-06,
+      "loss": 0.5642,
+      "step": 2732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.88077760759379e-06,
+      "loss": 0.5169,
+      "step": 2733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878091645838451e-06,
+      "loss": 0.4631,
+      "step": 2734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.875405422597913e-06,
+      "loss": 0.5305,
+      "step": 2735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.872718938672093e-06,
+      "loss": 0.5266,
+      "step": 2736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.870032194860995e-06,
+      "loss": 0.5803,
+      "step": 2737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867345191964702e-06,
+      "loss": 0.5962,
+      "step": 2738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864657930783369e-06,
+      "loss": 0.5224,
+      "step": 2739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861970412117231e-06,
+      "loss": 0.5476,
+      "step": 2740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.859282636766599e-06,
+      "loss": 0.562,
+      "step": 2741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85659460553186e-06,
+      "loss": 0.609,
+      "step": 2742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853906319213478e-06,
+      "loss": 0.5319,
+      "step": 2743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.851217778611994e-06,
+      "loss": 0.6087,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84852898452802e-06,
+      "loss": 0.506,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8458399377622495e-06,
+      "loss": 0.5302,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.843150639115448e-06,
+      "loss": 0.139,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.840461089388456e-06,
+      "loss": 0.5409,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8377712893821895e-06,
+      "loss": 0.1745,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.835081239897639e-06,
+      "loss": 0.5759,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832390941735867e-06,
+      "loss": 0.528,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.829700395698017e-06,
+      "loss": 0.5503,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827009602585295e-06,
+      "loss": 0.5257,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824318563198994e-06,
+      "loss": 0.4861,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.821627278340467e-06,
+      "loss": 0.5092,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818935748811152e-06,
+      "loss": 0.5873,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8162439754125485e-06,
+      "loss": 0.5564,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81355195894624e-06,
+      "loss": 0.536,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8108597002138755e-06,
+      "loss": 0.5706,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808167200017174e-06,
+      "loss": 0.5933,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.805474459157935e-06,
+      "loss": 0.5564,
+      "step": 2761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80278147843802e-06,
+      "loss": 0.1984,
+      "step": 2762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800088258659371e-06,
+      "loss": 0.5449,
+      "step": 2763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797394800623991e-06,
+      "loss": 0.5716,
+      "step": 2764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794701105133965e-06,
+      "loss": 0.552,
+      "step": 2765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.792007172991441e-06,
+      "loss": 0.471,
+      "step": 2766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78931300499864e-06,
+      "loss": 0.5831,
+      "step": 2767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.786618601957853e-06,
+      "loss": 0.1668,
+      "step": 2768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78392396467144e-06,
+      "loss": 0.5587,
+      "step": 2769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781229093941835e-06,
+      "loss": 0.1547,
+      "step": 2770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.778533990571535e-06,
+      "loss": 0.4843,
+      "step": 2771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775838655363111e-06,
+      "loss": 0.5286,
+      "step": 2772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.773143089119202e-06,
+      "loss": 0.5771,
+      "step": 2773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770447292642513e-06,
+      "loss": 0.5422,
+      "step": 2774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7677512667358236e-06,
+      "loss": 0.4894,
+      "step": 2775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765055012201977e-06,
+      "loss": 0.4602,
+      "step": 2776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7623585298438836e-06,
+      "loss": 0.1985,
+      "step": 2777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.759661820464526e-06,
+      "loss": 0.5522,
+      "step": 2778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75696488486695e-06,
+      "loss": 0.6144,
+      "step": 2779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.754267723854273e-06,
+      "loss": 0.5235,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751570338229675e-06,
+      "loss": 0.1569,
+      "step": 2781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748872728796409e-06,
+      "loss": 0.609,
+      "step": 2782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746174896357786e-06,
+      "loss": 0.559,
+      "step": 2783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743476841717192e-06,
+      "loss": 0.5298,
+      "step": 2784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740778565678072e-06,
+      "loss": 0.5717,
+      "step": 2785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738080069043944e-06,
+      "loss": 0.5319,
+      "step": 2786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.735381352618387e-06,
+      "loss": 0.5002,
+      "step": 2787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.732682417205044e-06,
+      "loss": 0.5083,
+      "step": 2788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.729983263607631e-06,
+      "loss": 0.5402,
+      "step": 2789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727283892629918e-06,
+      "loss": 0.5485,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.724584305075749e-06,
+      "loss": 0.4861,
+      "step": 2791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721884501749027e-06,
+      "loss": 0.5429,
+      "step": 2792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7191844834537225e-06,
+      "loss": 0.5752,
+      "step": 2793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7164842509938674e-06,
+      "loss": 0.5242,
+      "step": 2794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713783805173562e-06,
+      "loss": 0.4526,
+      "step": 2795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7110831467969605e-06,
+      "loss": 0.5582,
+      "step": 2796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708382276668294e-06,
+      "loss": 0.5458,
+      "step": 2797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705681195591845e-06,
+      "loss": 0.5004,
+      "step": 2798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702979904371964e-06,
+      "loss": 0.5393,
+      "step": 2799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700278403813062e-06,
+      "loss": 0.5352,
+      "step": 2800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697576694719616e-06,
+      "loss": 0.5456,
+      "step": 2801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694874777896162e-06,
+      "loss": 0.5706,
+      "step": 2802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.692172654147298e-06,
+      "loss": 0.5326,
+      "step": 2803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6894703242776835e-06,
+      "loss": 0.5673,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686767789092041e-06,
+      "loss": 0.5487,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6840650493951545e-06,
+      "loss": 0.4812,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.681362105991866e-06,
+      "loss": 0.4926,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.678658959687079e-06,
+      "loss": 0.5364,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67595561128576e-06,
+      "loss": 0.511,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673252061592933e-06,
+      "loss": 0.5888,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670548311413683e-06,
+      "loss": 0.1545,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.667844361553153e-06,
+      "loss": 0.4668,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66514021281655e-06,
+      "loss": 0.5726,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662435866009135e-06,
+      "loss": 0.464,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65973132193623e-06,
+      "loss": 0.5372,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.657026581403217e-06,
+      "loss": 0.535,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654321645215537e-06,
+      "loss": 0.5697,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651616514178684e-06,
+      "loss": 0.5685,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648911189098216e-06,
+      "loss": 0.5553,
+      "step": 2819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.646205670779745e-06,
+      "loss": 0.5288,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643499960028945e-06,
+      "loss": 0.4721,
+      "step": 2821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6407940576515405e-06,
+      "loss": 0.4873,
+      "step": 2822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63808796445332e-06,
+      "loss": 0.5393,
+      "step": 2823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.635381681240124e-06,
+      "loss": 0.5203,
+      "step": 2824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632675208817853e-06,
+      "loss": 0.5696,
+      "step": 2825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.629968547992459e-06,
+      "loss": 0.5591,
+      "step": 2826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.627261699569957e-06,
+      "loss": 0.1852,
+      "step": 2827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.624554664356412e-06,
+      "loss": 0.5552,
+      "step": 2828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6218474431579475e-06,
+      "loss": 0.4608,
+      "step": 2829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61914003678074e-06,
+      "loss": 0.5384,
+      "step": 2830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.616432446031025e-06,
+      "loss": 0.1894,
+      "step": 2831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613724671715089e-06,
+      "loss": 0.5353,
+      "step": 2832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.611016714639276e-06,
+      "loss": 0.4937,
+      "step": 2833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6083085756099845e-06,
+      "loss": 0.6056,
+      "step": 2834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6056002554336605e-06,
+      "loss": 0.4988,
+      "step": 2835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602891754916817e-06,
+      "loss": 0.5177,
+      "step": 2836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600183074866008e-06,
+      "loss": 0.5406,
+      "step": 2837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597474216087848e-06,
+      "loss": 0.4729,
+      "step": 2838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594765179389003e-06,
+      "loss": 0.5891,
+      "step": 2839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.592055965576189e-06,
+      "loss": 0.561,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.589346575456181e-06,
+      "loss": 0.5236,
+      "step": 2841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5866370098358e-06,
+      "loss": 0.1418,
+      "step": 2842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5839272695219245e-06,
+      "loss": 0.5607,
+      "step": 2843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5812173553214824e-06,
+      "loss": 0.1692,
+      "step": 2844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5785072680414535e-06,
+      "loss": 0.5787,
+      "step": 2845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.575797008488869e-06,
+      "loss": 0.5652,
+      "step": 2846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573086577470813e-06,
+      "loss": 0.5961,
+      "step": 2847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5703759757944185e-06,
+      "loss": 0.1906,
+      "step": 2848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56766520426687e-06,
+      "loss": 0.5967,
+      "step": 2849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.564954263695405e-06,
+      "loss": 0.563,
+      "step": 2850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5622431548873055e-06,
+      "loss": 0.5144,
+      "step": 2851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559531878649911e-06,
+      "loss": 0.5516,
+      "step": 2852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556820435790607e-06,
+      "loss": 0.532,
+      "step": 2853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.554108827116824e-06,
+      "loss": 0.5636,
+      "step": 2854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551397053436053e-06,
+      "loss": 0.5724,
+      "step": 2855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548685115555823e-06,
+      "loss": 0.5629,
+      "step": 2856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.545973014283721e-06,
+      "loss": 0.1616,
+      "step": 2857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543260750427373e-06,
+      "loss": 0.5469,
+      "step": 2858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5405483247944645e-06,
+      "loss": 0.1719,
+      "step": 2859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.53783573819272e-06,
+      "loss": 0.531,
+      "step": 2860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535122991429914e-06,
+      "loss": 0.5186,
+      "step": 2861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5324100853138755e-06,
+      "loss": 0.4856,
+      "step": 2862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52969702065247e-06,
+      "loss": 0.5166,
+      "step": 2863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52698379825362e-06,
+      "loss": 0.5193,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524270418925286e-06,
+      "loss": 0.5402,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.521556883475486e-06,
+      "loss": 0.5402,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518843192712275e-06,
+      "loss": 0.6201,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.516129347443755e-06,
+      "loss": 0.5475,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513415348478081e-06,
+      "loss": 0.5276,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510701196623449e-06,
+      "loss": 0.5446,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5079868926881e-06,
+      "loss": 0.4575,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5052724374803205e-06,
+      "loss": 0.5938,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5025578318084435e-06,
+      "loss": 0.1375,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4998430764808465e-06,
+      "loss": 0.5303,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497128172305951e-06,
+      "loss": 0.5754,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494413120092224e-06,
+      "loss": 0.5623,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4916979206481745e-06,
+      "loss": 0.5321,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488982574782357e-06,
+      "loss": 0.4849,
+      "step": 2878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.486267083303367e-06,
+      "loss": 0.5284,
+      "step": 2879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4835514470198494e-06,
+      "loss": 0.5324,
+      "step": 2880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.480835666740485e-06,
+      "loss": 0.5076,
+      "step": 2881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478119743274002e-06,
+      "loss": 0.645,
+      "step": 2882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.475403677429169e-06,
+      "loss": 0.529,
+      "step": 2883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472687470014799e-06,
+      "loss": 0.5191,
+      "step": 2884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4699711218397475e-06,
+      "loss": 0.5084,
+      "step": 2885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467254633712907e-06,
+      "loss": 0.512,
+      "step": 2886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464538006443218e-06,
+      "loss": 0.5171,
+      "step": 2887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.461821240839658e-06,
+      "loss": 0.5323,
+      "step": 2888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459104337711248e-06,
+      "loss": 0.4968,
+      "step": 2889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.456387297867051e-06,
+      "loss": 0.5621,
+      "step": 2890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453670122116166e-06,
+      "loss": 0.5085,
+      "step": 2891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450952811267739e-06,
+      "loss": 0.561,
+      "step": 2892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448235366130948e-06,
+      "loss": 0.5679,
+      "step": 2893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44551778751502e-06,
+      "loss": 0.5134,
+      "step": 2894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442800076229214e-06,
+      "loss": 0.5037,
+      "step": 2895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440082233082837e-06,
+      "loss": 0.1664,
+      "step": 2896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4373642588852235e-06,
+      "loss": 0.5717,
+      "step": 2897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.43464615444576e-06,
+      "loss": 0.1599,
+      "step": 2898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4319279205738615e-06,
+      "loss": 0.4919,
+      "step": 2899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429209558078987e-06,
+      "loss": 0.475,
+      "step": 2900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426491067770633e-06,
+      "loss": 0.4911,
+      "step": 2901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.423772450458332e-06,
+      "loss": 0.5115,
+      "step": 2902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.421053706951657e-06,
+      "loss": 0.4571,
+      "step": 2903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418334838060215e-06,
+      "loss": 0.5735,
+      "step": 2904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4156158445936555e-06,
+      "loss": 0.5426,
+      "step": 2905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.412896727361663e-06,
+      "loss": 0.5471,
+      "step": 2906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.410177487173954e-06,
+      "loss": 0.5245,
+      "step": 2907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407458124840289e-06,
+      "loss": 0.587,
+      "step": 2908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404738641170462e-06,
+      "loss": 0.5512,
+      "step": 2909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4020190369743e-06,
+      "loss": 0.5056,
+      "step": 2910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.399299313061671e-06,
+      "loss": 0.5044,
+      "step": 2911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3965794702424755e-06,
+      "loss": 0.191,
+      "step": 2912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39385950932665e-06,
+      "loss": 0.5482,
+      "step": 2913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.391139431124166e-06,
+      "loss": 0.553,
+      "step": 2914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388419236445033e-06,
+      "loss": 0.58,
+      "step": 2915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385698926099287e-06,
+      "loss": 0.563,
+      "step": 2916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.382978500897009e-06,
+      "loss": 0.5533,
+      "step": 2917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380257961648308e-06,
+      "loss": 0.5323,
+      "step": 2918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.377537309163328e-06,
+      "loss": 0.6063,
+      "step": 2919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374816544252245e-06,
+      "loss": 0.4741,
+      "step": 2920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372095667725272e-06,
+      "loss": 0.5006,
+      "step": 2921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369374680392654e-06,
+      "loss": 0.4524,
+      "step": 2922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.366653583064666e-06,
+      "loss": 0.5402,
+      "step": 2923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.363932376551623e-06,
+      "loss": 0.5193,
+      "step": 2924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361211061663861e-06,
+      "loss": 0.5306,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3584896392117614e-06,
+      "loss": 0.5814,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355768110005728e-06,
+      "loss": 0.4905,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353046474856201e-06,
+      "loss": 0.5258,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35032473457365e-06,
+      "loss": 0.5177,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.347602889968577e-06,
+      "loss": 0.5469,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344880941851514e-06,
+      "loss": 0.5616,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.342158891033027e-06,
+      "loss": 0.5853,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339436738323708e-06,
+      "loss": 0.1483,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336714484534183e-06,
+      "loss": 0.6806,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.333992130475105e-06,
+      "loss": 0.555,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.331269676957158e-06,
+      "loss": 0.5254,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.328547124791061e-06,
+      "loss": 0.5505,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.325824474787553e-06,
+      "loss": 0.5179,
+      "step": 2938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3231017277574075e-06,
+      "loss": 0.5348,
+      "step": 2939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320378884511427e-06,
+      "loss": 0.6205,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3176559458604395e-06,
+      "loss": 0.5583,
+      "step": 2941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.314932912615306e-06,
+      "loss": 0.5513,
+      "step": 2942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3122097855869125e-06,
+      "loss": 0.1613,
+      "step": 2943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309486565586174e-06,
+      "loss": 0.5227,
+      "step": 2944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3067632534240325e-06,
+      "loss": 0.4863,
+      "step": 2945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.304039849911458e-06,
+      "loss": 0.6072,
+      "step": 2946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301316355859446e-06,
+      "loss": 0.5064,
+      "step": 2947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298592772079023e-06,
+      "loss": 0.5582,
+      "step": 2948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.295869099381239e-06,
+      "loss": 0.5589,
+      "step": 2949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.29314533857717e-06,
+      "loss": 0.5073,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2904214904779206e-06,
+      "loss": 0.5126,
+      "step": 2951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287697555894621e-06,
+      "loss": 0.503,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.284973535638424e-06,
+      "loss": 0.5795,
+      "step": 2953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.282249430520513e-06,
+      "loss": 0.6099,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27952524135209e-06,
+      "loss": 0.5366,
+      "step": 2955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.276800968944391e-06,
+      "loss": 0.5189,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274076614108668e-06,
+      "loss": 0.5471,
+      "step": 2957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.271352177656204e-06,
+      "loss": 0.5625,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2686276603983e-06,
+      "loss": 0.5346,
+      "step": 2959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.265903063146291e-06,
+      "loss": 0.5434,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.263178386711521e-06,
+      "loss": 0.5056,
+      "step": 2961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.260453631905373e-06,
+      "loss": 0.5163,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2577287995392445e-06,
+      "loss": 0.5359,
+      "step": 2963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255003890424556e-06,
+      "loss": 0.5761,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252278905372756e-06,
+      "loss": 0.4759,
+      "step": 2965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249553845195309e-06,
+      "loss": 0.5555,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2468287107037095e-06,
+      "loss": 0.468,
+      "step": 2967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.244103502709468e-06,
+      "loss": 0.533,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241378222024118e-06,
+      "loss": 0.548,
+      "step": 2969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238652869459217e-06,
+      "loss": 0.5704,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2359274458263396e-06,
+      "loss": 0.5659,
+      "step": 2971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233201951937088e-06,
+      "loss": 0.6446,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.23047638860308e-06,
+      "loss": 0.5663,
+      "step": 2973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227750756635957e-06,
+      "loss": 0.5166,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.225025056847378e-06,
+      "loss": 0.557,
+      "step": 2975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222299290049025e-06,
+      "loss": 0.4978,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219573457052598e-06,
+      "loss": 0.5839,
+      "step": 2977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216847558669819e-06,
+      "loss": 0.6296,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214121595712426e-06,
+      "loss": 0.5728,
+      "step": 2979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2113955689921795e-06,
+      "loss": 0.5689,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208669479320856e-06,
+      "loss": 0.486,
+      "step": 2981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205943327510254e-06,
+      "loss": 0.6137,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20321711437219e-06,
+      "loss": 0.5034,
+      "step": 2983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200490840718498e-06,
+      "loss": 0.577,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197764507361024e-06,
+      "loss": 0.5217,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1950381151116446e-06,
+      "loss": 0.4825,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192311664782241e-06,
+      "loss": 0.5605,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.189585157184725e-06,
+      "loss": 0.5799,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186858593131009e-06,
+      "loss": 0.4895,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.184131973433039e-06,
+      "loss": 0.5681,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181405298902763e-06,
+      "loss": 0.5705,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178678570352158e-06,
+      "loss": 0.549,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.175951788593209e-06,
+      "loss": 0.5358,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1732249544379164e-06,
+      "loss": 0.5009,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170498068698304e-06,
+      "loss": 0.5284,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167771132186401e-06,
+      "loss": 0.5201,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1650441457142606e-06,
+      "loss": 0.5147,
+      "step": 2997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.162317110093943e-06,
+      "loss": 0.5068,
+      "step": 2998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15959002613753e-06,
+      "loss": 0.5252,
+      "step": 2999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.156862894657113e-06,
+      "loss": 0.5717,
+      "step": 3000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.154135716464799e-06,
+      "loss": 0.5144,
+      "step": 3001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.151408492372709e-06,
+      "loss": 0.5683,
+      "step": 3002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148681223192978e-06,
+      "loss": 0.6017,
+      "step": 3003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145953909737756e-06,
+      "loss": 0.5569,
+      "step": 3004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143226552819202e-06,
+      "loss": 0.5873,
+      "step": 3005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140499153249492e-06,
+      "loss": 0.5103,
+      "step": 3006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137771711840811e-06,
+      "loss": 0.1865,
+      "step": 3007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.13504422940536e-06,
+      "loss": 0.5245,
+      "step": 3008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.13231670675535e-06,
+      "loss": 0.4391,
+      "step": 3009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1295891447030056e-06,
+      "loss": 0.5253,
+      "step": 3010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126861544060561e-06,
+      "loss": 0.5433,
+      "step": 3011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.124133905640265e-06,
+      "loss": 0.5449,
+      "step": 3012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121406230254373e-06,
+      "loss": 0.5211,
+      "step": 3013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118678518715157e-06,
+      "loss": 0.5209,
+      "step": 3014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.115950771834894e-06,
+      "loss": 0.5008,
+      "step": 3015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.113222990425878e-06,
+      "loss": 0.4844,
+      "step": 3016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110495175300405e-06,
+      "loss": 0.5275,
+      "step": 3017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1077673272707916e-06,
+      "loss": 0.4881,
+      "step": 3018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105039447149352e-06,
+      "loss": 0.5009,
+      "step": 3019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102311535748421e-06,
+      "loss": 0.5293,
+      "step": 3020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099583593880337e-06,
+      "loss": 0.446,
+      "step": 3021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096855622357448e-06,
+      "loss": 0.5049,
+      "step": 3022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094127621992112e-06,
+      "loss": 0.4935,
+      "step": 3023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091399593596693e-06,
+      "loss": 0.5126,
+      "step": 3024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088671537983568e-06,
+      "loss": 0.4981,
+      "step": 3025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085943455965117e-06,
+      "loss": 0.506,
+      "step": 3026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083215348353732e-06,
+      "loss": 0.4587,
+      "step": 3027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080487215961808e-06,
+      "loss": 0.5397,
+      "step": 3028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077759059601756e-06,
+      "loss": 0.4336,
+      "step": 3029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.075030880085981e-06,
+      "loss": 0.5437,
+      "step": 3030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0723026782269094e-06,
+      "loss": 0.5101,
+      "step": 3031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069574454836962e-06,
+      "loss": 0.1669,
+      "step": 3032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.066846210728573e-06,
+      "loss": 0.6261,
+      "step": 3033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.064117946714181e-06,
+      "loss": 0.5756,
+      "step": 3034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06138966360623e-06,
+      "loss": 0.5173,
+      "step": 3035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05866136221717e-06,
+      "loss": 0.5814,
+      "step": 3036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.055933043359458e-06,
+      "loss": 0.5401,
+      "step": 3037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053204707845553e-06,
+      "loss": 0.5502,
+      "step": 3038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050476356487921e-06,
+      "loss": 0.5047,
+      "step": 3039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.047747990099033e-06,
+      "loss": 0.5068,
+      "step": 3040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045019609491365e-06,
+      "loss": 0.5081,
+      "step": 3041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.042291215477394e-06,
+      "loss": 0.5712,
+      "step": 3042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0395628088696044e-06,
+      "loss": 0.614,
+      "step": 3043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.036834390480483e-06,
+      "loss": 0.5081,
+      "step": 3044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03410596112252e-06,
+      "loss": 0.5175,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031377521608209e-06,
+      "loss": 0.5073,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028649072750046e-06,
+      "loss": 0.5066,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025920615360532e-06,
+      "loss": 0.561,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023192150252169e-06,
+      "loss": 0.5136,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020463678237459e-06,
+      "loss": 0.5176,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.017735200128912e-06,
+      "loss": 0.523,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0150067167390325e-06,
+      "loss": 0.5381,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.012278228880333e-06,
+      "loss": 0.1576,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009549737365323e-06,
+      "loss": 0.5118,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.006821243006516e-06,
+      "loss": 0.5132,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.004092746616424e-06,
+      "loss": 0.4908,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.001364249007561e-06,
+      "loss": 0.4872,
+      "step": 3057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99863575099244e-06,
+      "loss": 0.544,
+      "step": 3058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.995907253383577e-06,
+      "loss": 0.5693,
+      "step": 3059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993178756993486e-06,
+      "loss": 0.5414,
+      "step": 3060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9904502626346785e-06,
+      "loss": 0.4974,
+      "step": 3061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987721771119667e-06,
+      "loss": 0.5631,
+      "step": 3062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.984993283260969e-06,
+      "loss": 0.5473,
+      "step": 3063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98226479987109e-06,
+      "loss": 0.6361,
+      "step": 3064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979536321762542e-06,
+      "loss": 0.4832,
+      "step": 3065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976807849747832e-06,
+      "loss": 0.5404,
+      "step": 3066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974079384639469e-06,
+      "loss": 0.4955,
+      "step": 3067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.971350927249955e-06,
+      "loss": 0.5725,
+      "step": 3068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968622478391792e-06,
+      "loss": 0.5259,
+      "step": 3069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965894038877484e-06,
+      "loss": 0.4784,
+      "step": 3070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9631656095195195e-06,
+      "loss": 0.4959,
+      "step": 3071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960437191130397e-06,
+      "loss": 0.4771,
+      "step": 3072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.957708784522606e-06,
+      "loss": 0.4998,
+      "step": 3073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9549803905086376e-06,
+      "loss": 0.5322,
+      "step": 3074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.952252009900968e-06,
+      "loss": 0.5594,
+      "step": 3075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949523643512081e-06,
+      "loss": 0.5387,
+      "step": 3076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94679529215445e-06,
+      "loss": 0.4874,
+      "step": 3077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.944066956640544e-06,
+      "loss": 0.5416,
+      "step": 3078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941338637782831e-06,
+      "loss": 0.4612,
+      "step": 3079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938610336393772e-06,
+      "loss": 0.5393,
+      "step": 3080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9358820532858224e-06,
+      "loss": 0.4983,
+      "step": 3081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933153789271429e-06,
+      "loss": 0.5756,
+      "step": 3082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9304255451630395e-06,
+      "loss": 0.5477,
+      "step": 3083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927697321773092e-06,
+      "loss": 0.4729,
+      "step": 3084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92496911991402e-06,
+      "loss": 0.5391,
+      "step": 3085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.922240940398246e-06,
+      "loss": 0.4592,
+      "step": 3086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919512784038192e-06,
+      "loss": 0.5325,
+      "step": 3087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9167846516462705e-06,
+      "loss": 0.5205,
+      "step": 3088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914056544034885e-06,
+      "loss": 0.5739,
+      "step": 3089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911328462016433e-06,
+      "loss": 0.5419,
+      "step": 3090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.908600406403308e-06,
+      "loss": 0.4644,
+      "step": 3091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90587237800789e-06,
+      "loss": 0.5342,
+      "step": 3092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903144377642553e-06,
+      "loss": 0.5576,
+      "step": 3093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.900416406119663e-06,
+      "loss": 0.5247,
+      "step": 3094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89768846425158e-06,
+      "loss": 0.4795,
+      "step": 3095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894960552850649e-06,
+      "loss": 0.5881,
+      "step": 3096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892232672729209e-06,
+      "loss": 0.6246,
+      "step": 3097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889504824699594e-06,
+      "loss": 0.5038,
+      "step": 3098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.886777009574124e-06,
+      "loss": 0.6357,
+      "step": 3099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.884049228165107e-06,
+      "loss": 0.5458,
+      "step": 3100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881321481284843e-06,
+      "loss": 0.2156,
+      "step": 3101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8785937697456295e-06,
+      "loss": 0.1649,
+      "step": 3102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875866094359737e-06,
+      "loss": 0.5355,
+      "step": 3103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87313845593944e-06,
+      "loss": 0.5218,
+      "step": 3104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870410855296994e-06,
+      "loss": 0.6162,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.867683293244651e-06,
+      "loss": 0.1753,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.864955770594641e-06,
+      "loss": 0.5624,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862228288159191e-06,
+      "loss": 0.6244,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8595008467505115e-06,
+      "loss": 0.5049,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856773447180799e-06,
+      "loss": 0.5372,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854046090262245e-06,
+      "loss": 0.5144,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8513187768070225e-06,
+      "loss": 0.5111,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848591507627293e-06,
+      "loss": 0.5409,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.845864283535203e-06,
+      "loss": 0.5373,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.843137105342889e-06,
+      "loss": 0.5361,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840409973862472e-06,
+      "loss": 0.506,
+      "step": 3116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837682889906059e-06,
+      "loss": 0.1626,
+      "step": 3117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83495585428574e-06,
+      "loss": 0.5968,
+      "step": 3118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8322288678136e-06,
+      "loss": 0.544,
+      "step": 3119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8295019313016975e-06,
+      "loss": 0.5304,
+      "step": 3120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826775045562084e-06,
+      "loss": 0.5619,
+      "step": 3121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824048211406792e-06,
+      "loss": 0.5614,
+      "step": 3122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821321429647843e-06,
+      "loss": 0.5658,
+      "step": 3123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818594701097239e-06,
+      "loss": 0.5321,
+      "step": 3124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815868026566963e-06,
+      "loss": 0.5186,
+      "step": 3125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813141406868992e-06,
+      "loss": 0.5266,
+      "step": 3126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8104148428152785e-06,
+      "loss": 0.5903,
+      "step": 3127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80768833521776e-06,
+      "loss": 0.5881,
+      "step": 3128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804961884888357e-06,
+      "loss": 0.4663,
+      "step": 3129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802235492638976e-06,
+      "loss": 0.1821,
+      "step": 3130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.799509159281506e-06,
+      "loss": 0.5435,
+      "step": 3131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7967828856278115e-06,
+      "loss": 0.5252,
+      "step": 3132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794056672489746e-06,
+      "loss": 0.5531,
+      "step": 3133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791330520679146e-06,
+      "loss": 0.5321,
+      "step": 3134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.788604431007823e-06,
+      "loss": 0.599,
+      "step": 3135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785878404287576e-06,
+      "loss": 0.1764,
+      "step": 3136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783152441330182e-06,
+      "loss": 0.5568,
+      "step": 3137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780426542947403e-06,
+      "loss": 0.4953,
+      "step": 3138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7777007099509765e-06,
+      "loss": 0.4712,
+      "step": 3139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774974943152623e-06,
+      "loss": 0.5521,
+      "step": 3140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772249243364045e-06,
+      "loss": 0.5954,
+      "step": 3141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7695236113969205e-06,
+      "loss": 0.5702,
+      "step": 3142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766798048062913e-06,
+      "loss": 0.557,
+      "step": 3143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76407255417366e-06,
+      "loss": 0.4862,
+      "step": 3144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761347130540787e-06,
+      "loss": 0.5867,
+      "step": 3145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758621777975884e-06,
+      "loss": 0.5635,
+      "step": 3146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.755896497290534e-06,
+      "loss": 0.5118,
+      "step": 3147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753171289296292e-06,
+      "loss": 0.5957,
+      "step": 3148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750446154804692e-06,
+      "loss": 0.5952,
+      "step": 3149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747721094627246e-06,
+      "loss": 0.5393,
+      "step": 3150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744996109575445e-06,
+      "loss": 0.5414,
+      "step": 3151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742271200460759e-06,
+      "loss": 0.5118,
+      "step": 3152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739546368094628e-06,
+      "loss": 0.5877,
+      "step": 3153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73682161328848e-06,
+      "loss": 0.6244,
+      "step": 3154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734096936853713e-06,
+      "loss": 0.5553,
+      "step": 3155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731372339601701e-06,
+      "loss": 0.4964,
+      "step": 3156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.728647822343798e-06,
+      "loss": 0.5362,
+      "step": 3157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725923385891333e-06,
+      "loss": 0.5488,
+      "step": 3158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723199031055611e-06,
+      "loss": 0.5355,
+      "step": 3159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720474758647911e-06,
+      "loss": 0.5279,
+      "step": 3160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717750569479488e-06,
+      "loss": 0.1926,
+      "step": 3161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715026464361576e-06,
+      "loss": 0.5979,
+      "step": 3162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712302444105381e-06,
+      "loss": 0.5234,
+      "step": 3163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70957850952208e-06,
+      "loss": 0.5584,
+      "step": 3164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70685466142283e-06,
+      "loss": 0.6044,
+      "step": 3165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704130900618762e-06,
+      "loss": 0.54,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7014072279209786e-06,
+      "loss": 0.4979,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698683644140555e-06,
+      "loss": 0.1558,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.695960150088543e-06,
+      "loss": 0.4869,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.693236746575969e-06,
+      "loss": 0.5666,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690513434413828e-06,
+      "loss": 0.6328,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6877902144130875e-06,
+      "loss": 0.5933,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685067087384696e-06,
+      "loss": 0.4736,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682344054139561e-06,
+      "loss": 0.5591,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679621115488575e-06,
+      "loss": 0.6067,
+      "step": 3175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6768982722425925e-06,
+      "loss": 0.559,
+      "step": 3176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6741755252124495e-06,
+      "loss": 0.5546,
+      "step": 3177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.67145287520894e-06,
+      "loss": 0.5396,
+      "step": 3178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668730323042842e-06,
+      "loss": 0.6125,
+      "step": 3179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666007869524898e-06,
+      "loss": 0.5813,
+      "step": 3180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663285515465818e-06,
+      "loss": 0.5708,
+      "step": 3181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660563261676293e-06,
+      "loss": 0.4662,
+      "step": 3182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657841108966975e-06,
+      "loss": 0.5421,
+      "step": 3183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655119058148488e-06,
+      "loss": 0.5356,
+      "step": 3184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6523971100314245e-06,
+      "loss": 0.5208,
+      "step": 3185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649675265426352e-06,
+      "loss": 0.5141,
+      "step": 3186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646953525143801e-06,
+      "loss": 0.5476,
+      "step": 3187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644231889994274e-06,
+      "loss": 0.5711,
+      "step": 3188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.641510360788239e-06,
+      "loss": 0.5532,
+      "step": 3189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63878893833614e-06,
+      "loss": 0.5879,
+      "step": 3190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.636067623448381e-06,
+      "loss": 0.5531,
+      "step": 3191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.633346416935335e-06,
+      "loss": 0.5627,
+      "step": 3192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6306253196073466e-06,
+      "loss": 0.5258,
+      "step": 3193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6279043322747294e-06,
+      "loss": 0.5127,
+      "step": 3194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6251834557477565e-06,
+      "loss": 0.5541,
+      "step": 3195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622462690836674e-06,
+      "loss": 0.54,
+      "step": 3196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6197420383516926e-06,
+      "loss": 0.5322,
+      "step": 3197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617021499102992e-06,
+      "loss": 0.5606,
+      "step": 3198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614301073900714e-06,
+      "loss": 0.5035,
+      "step": 3199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611580763554969e-06,
+      "loss": 0.507,
+      "step": 3200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608860568875834e-06,
+      "loss": 0.5506,
+      "step": 3201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606140490673352e-06,
+      "loss": 0.4519,
+      "step": 3202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603420529757526e-06,
+      "loss": 0.5799,
+      "step": 3203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6007006869383295e-06,
+      "loss": 0.5667,
+      "step": 3204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597980963025702e-06,
+      "loss": 0.507,
+      "step": 3205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59526135882954e-06,
+      "loss": 0.6009,
+      "step": 3206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5925418751597115e-06,
+      "loss": 0.4827,
+      "step": 3207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.589822512826046e-06,
+      "loss": 0.5317,
+      "step": 3208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.587103272638339e-06,
+      "loss": 0.4899,
+      "step": 3209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.584384155406345e-06,
+      "loss": 0.5858,
+      "step": 3210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5816651619397854e-06,
+      "loss": 0.4717,
+      "step": 3211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5789462930483465e-06,
+      "loss": 0.5411,
+      "step": 3212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.576227549541669e-06,
+      "loss": 0.1875,
+      "step": 3213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.573508932229368e-06,
+      "loss": 0.5464,
+      "step": 3214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5707904419210135e-06,
+      "loss": 0.5135,
+      "step": 3215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568072079426141e-06,
+      "loss": 0.5015,
+      "step": 3216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565353845554241e-06,
+      "loss": 0.5584,
+      "step": 3217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5626357411147765e-06,
+      "loss": 0.5376,
+      "step": 3218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559917766917166e-06,
+      "loss": 0.4999,
+      "step": 3219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5571999237707864e-06,
+      "loss": 0.5475,
+      "step": 3220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.554482212484981e-06,
+      "loss": 0.5706,
+      "step": 3221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5517646338690525e-06,
+      "loss": 0.6251,
+      "step": 3222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549047188732264e-06,
+      "loss": 0.552,
+      "step": 3223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546329877883835e-06,
+      "loss": 0.571,
+      "step": 3224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54361270213295e-06,
+      "loss": 0.5368,
+      "step": 3225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540895662288752e-06,
+      "loss": 0.5341,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.538178759160343e-06,
+      "loss": 0.6176,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535461993556783e-06,
+      "loss": 0.5854,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532745366287094e-06,
+      "loss": 0.4492,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530028878160255e-06,
+      "loss": 0.4984,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527312529985202e-06,
+      "loss": 0.5655,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5245963225708314e-06,
+      "loss": 0.5372,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5218802567260015e-06,
+      "loss": 0.5864,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.519164333259517e-06,
+      "loss": 0.5573,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516448552980153e-06,
+      "loss": 0.5676,
+      "step": 3235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513732916696633e-06,
+      "loss": 0.1729,
+      "step": 3236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511017425217646e-06,
+      "loss": 0.6117,
+      "step": 3237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508302079351827e-06,
+      "loss": 0.5364,
+      "step": 3238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505586879907777e-06,
+      "loss": 0.5188,
+      "step": 3239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.502871827694049e-06,
+      "loss": 0.5257,
+      "step": 3240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.500156923519154e-06,
+      "loss": 0.5541,
+      "step": 3241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.497442168191558e-06,
+      "loss": 0.5912,
+      "step": 3242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494727562519681e-06,
+      "loss": 0.4606,
+      "step": 3243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4920131073119035e-06,
+      "loss": 0.4993,
+      "step": 3244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.489298803376553e-06,
+      "loss": 0.5024,
+      "step": 3245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4865846515219195e-06,
+      "loss": 0.5603,
+      "step": 3246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.483870652556245e-06,
+      "loss": 0.4758,
+      "step": 3247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.481156807287729e-06,
+      "loss": 0.5313,
+      "step": 3248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.478443116524515e-06,
+      "loss": 0.5497,
+      "step": 3249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475729581074714e-06,
+      "loss": 0.5986,
+      "step": 3250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4730162017463826e-06,
+      "loss": 0.5059,
+      "step": 3251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470302979347532e-06,
+      "loss": 0.5543,
+      "step": 3252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467589914686125e-06,
+      "loss": 0.584,
+      "step": 3253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464877008570086e-06,
+      "loss": 0.5607,
+      "step": 3254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462164261807283e-06,
+      "loss": 0.5877,
+      "step": 3255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459451675205537e-06,
+      "loss": 0.5635,
+      "step": 3256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456739249572628e-06,
+      "loss": 0.4885,
+      "step": 3257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454026985716282e-06,
+      "loss": 0.5027,
+      "step": 3258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451314884444179e-06,
+      "loss": 0.2017,
+      "step": 3259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.448602946563948e-06,
+      "loss": 0.1782,
+      "step": 3260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.445891172883176e-06,
+      "loss": 0.5769,
+      "step": 3261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.443179564209396e-06,
+      "loss": 0.213,
+      "step": 3262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4404681213500905e-06,
+      "loss": 0.5103,
+      "step": 3263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.437756845112694e-06,
+      "loss": 0.5673,
+      "step": 3264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.435045736304597e-06,
+      "loss": 0.5117,
+      "step": 3265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432334795733131e-06,
+      "loss": 0.4518,
+      "step": 3266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.429624024205582e-06,
+      "loss": 0.605,
+      "step": 3267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.426913422529187e-06,
+      "loss": 0.5787,
+      "step": 3268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.424202991511132e-06,
+      "loss": 0.5229,
+      "step": 3269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421492731958547e-06,
+      "loss": 0.5427,
+      "step": 3270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418782644678518e-06,
+      "loss": 0.56,
+      "step": 3271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416072730478078e-06,
+      "loss": 0.5415,
+      "step": 3272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413362990164201e-06,
+      "loss": 0.5264,
+      "step": 3273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.410653424543822e-06,
+      "loss": 0.4728,
+      "step": 3274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.407944034423811e-06,
+      "loss": 0.5187,
+      "step": 3275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405234820611001e-06,
+      "loss": 0.5789,
+      "step": 3276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4025257839121535e-06,
+      "loss": 0.5215,
+      "step": 3277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399816925133993e-06,
+      "loss": 0.5271,
+      "step": 3278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397108245083183e-06,
+      "loss": 0.584,
+      "step": 3279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.39439974456634e-06,
+      "loss": 0.602,
+      "step": 3280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391691424390018e-06,
+      "loss": 0.5101,
+      "step": 3281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388983285360725e-06,
+      "loss": 0.584,
+      "step": 3282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386275328284913e-06,
+      "loss": 0.5549,
+      "step": 3283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.383567553968976e-06,
+      "loss": 0.5676,
+      "step": 3284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.380859963219262e-06,
+      "loss": 0.5685,
+      "step": 3285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.378152556842055e-06,
+      "loss": 0.5377,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37544533564359e-06,
+      "loss": 0.5257,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.372738300430044e-06,
+      "loss": 0.5129,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370031452007542e-06,
+      "loss": 0.5386,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367324791182149e-06,
+      "loss": 0.503,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364618318759877e-06,
+      "loss": 0.5273,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36191203554668e-06,
+      "loss": 0.5642,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3592059423484595e-06,
+      "loss": 0.5157,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356500039971057e-06,
+      "loss": 0.5152,
+      "step": 3294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3537943292202555e-06,
+      "loss": 0.5264,
+      "step": 3295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351088810901785e-06,
+      "loss": 0.4712,
+      "step": 3296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348383485821318e-06,
+      "loss": 0.5064,
+      "step": 3297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345678354784465e-06,
+      "loss": 0.5059,
+      "step": 3298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3429734185967835e-06,
+      "loss": 0.5205,
+      "step": 3299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34026867806377e-06,
+      "loss": 0.5459,
+      "step": 3300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.337564133990867e-06,
+      "loss": 0.5091,
+      "step": 3301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334859787183451e-06,
+      "loss": 0.5577,
+      "step": 3302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332155638446846e-06,
+      "loss": 0.5223,
+      "step": 3303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.32945168858632e-06,
+      "loss": 0.5074,
+      "step": 3304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326747938407068e-06,
+      "loss": 0.5564,
+      "step": 3305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.324044388714241e-06,
+      "loss": 0.5219,
+      "step": 3306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.321341040312921e-06,
+      "loss": 0.5642,
+      "step": 3307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318637894008137e-06,
+      "loss": 0.5414,
+      "step": 3308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315934950604847e-06,
+      "loss": 0.5699,
+      "step": 3309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313232210907959e-06,
+      "loss": 0.6279,
+      "step": 3310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310529675722319e-06,
+      "loss": 0.5851,
+      "step": 3311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.307827345852704e-06,
+      "loss": 0.5755,
+      "step": 3312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305125222103839e-06,
+      "loss": 0.4806,
+      "step": 3313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3024233052803855e-06,
+      "loss": 0.6225,
+      "step": 3314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299721596186941e-06,
+      "loss": 0.5674,
+      "step": 3315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297020095628038e-06,
+      "loss": 0.4528,
+      "step": 3316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294318804408157e-06,
+      "loss": 0.5321,
+      "step": 3317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291617723331707e-06,
+      "loss": 0.5101,
+      "step": 3318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28891685320304e-06,
+      "loss": 0.4915,
+      "step": 3319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2862161948264405e-06,
+      "loss": 0.5616,
+      "step": 3320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283515749006133e-06,
+      "loss": 0.5536,
+      "step": 3321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280815516546279e-06,
+      "loss": 0.6068,
+      "step": 3322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278115498250975e-06,
+      "loss": 0.5943,
+      "step": 3323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275415694924252e-06,
+      "loss": 0.5712,
+      "step": 3324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272716107370083e-06,
+      "loss": 0.4672,
+      "step": 3325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270016736392371e-06,
+      "loss": 0.1866,
+      "step": 3326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.267317582794957e-06,
+      "loss": 0.5819,
+      "step": 3327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264618647381614e-06,
+      "loss": 0.4891,
+      "step": 3328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.261919930956057e-06,
+      "loss": 0.5783,
+      "step": 3329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259221434321929e-06,
+      "loss": 0.1743,
+      "step": 3330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.256523158282809e-06,
+      "loss": 0.5118,
+      "step": 3331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253825103642214e-06,
+      "loss": 0.5903,
+      "step": 3332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251127271203593e-06,
+      "loss": 0.1796,
+      "step": 3333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248429661770325e-06,
+      "loss": 0.4822,
+      "step": 3334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245732276145727e-06,
+      "loss": 0.4793,
+      "step": 3335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.243035115133052e-06,
+      "loss": 0.5163,
+      "step": 3336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240338179535476e-06,
+      "loss": 0.5601,
+      "step": 3337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.237641470156117e-06,
+      "loss": 0.6106,
+      "step": 3338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234944987798023e-06,
+      "loss": 0.5519,
+      "step": 3339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232248733264177e-06,
+      "loss": 0.4634,
+      "step": 3340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2295527073574875e-06,
+      "loss": 0.4944,
+      "step": 3341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2268569108808e-06,
+      "loss": 0.5413,
+      "step": 3342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224161344636891e-06,
+      "loss": 0.5701,
+      "step": 3343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2214660094284665e-06,
+      "loss": 0.5737,
+      "step": 3344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218770906058167e-06,
+      "loss": 0.6189,
+      "step": 3345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2160760353285604e-06,
+      "loss": 0.5929,
+      "step": 3346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21338139804215e-06,
+      "loss": 0.53,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210686995001361e-06,
+      "loss": 0.5612,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20799282700856e-06,
+      "loss": 0.4841,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205298894866037e-06,
+      "loss": 0.5382,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20260519937601e-06,
+      "loss": 0.6295,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199911741340631e-06,
+      "loss": 0.5277,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.19721852156198e-06,
+      "loss": 0.6021,
+      "step": 3353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.194525540842067e-06,
+      "loss": 0.5365,
+      "step": 3354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191832799982827e-06,
+      "loss": 0.522,
+      "step": 3355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.189140299786126e-06,
+      "loss": 0.552,
+      "step": 3356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18644804105376e-06,
+      "loss": 0.5246,
+      "step": 3357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183756024587452e-06,
+      "loss": 0.5371,
+      "step": 3358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18106425118885e-06,
+      "loss": 0.5024,
+      "step": 3359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.178372721659534e-06,
+      "loss": 0.5612,
+      "step": 3360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.175681436801009e-06,
+      "loss": 0.56,
+      "step": 3361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172990397414706e-06,
+      "loss": 0.561,
+      "step": 3362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170299604301985e-06,
+      "loss": 0.5052,
+      "step": 3363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167609058264133e-06,
+      "loss": 0.5504,
+      "step": 3364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.164918760102363e-06,
+      "loss": 0.5651,
+      "step": 3365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.162228710617812e-06,
+      "loss": 0.5751,
+      "step": 3366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159538910611545e-06,
+      "loss": 0.5513,
+      "step": 3367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156849360884553e-06,
+      "loss": 0.5298,
+      "step": 3368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154160062237751e-06,
+      "loss": 0.5817,
+      "step": 3369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151471015471981e-06,
+      "loss": 0.558,
+      "step": 3370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.148782221388007e-06,
+      "loss": 0.5269,
+      "step": 3371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1460936807865225e-06,
+      "loss": 0.5562,
+      "step": 3372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1434053944681414e-06,
+      "loss": 0.5051,
+      "step": 3373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140717363233403e-06,
+      "loss": 0.5426,
+      "step": 3374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.138029587882772e-06,
+      "loss": 0.571,
+      "step": 3375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135342069216633e-06,
+      "loss": 0.5228,
+      "step": 3376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1326548080353e-06,
+      "loss": 0.5222,
+      "step": 3377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.129967805139005e-06,
+      "loss": 0.5237,
+      "step": 3378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12728106132791e-06,
+      "loss": 0.5579,
+      "step": 3379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.124594577402089e-06,
+      "loss": 0.5424,
+      "step": 3380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121908354161548e-06,
+      "loss": 0.5339,
+      "step": 3381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1192223924062135e-06,
+      "loss": 0.5227,
+      "step": 3382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11653669293593e-06,
+      "loss": 0.5144,
+      "step": 3383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113851256550466e-06,
+      "loss": 0.5671,
+      "step": 3384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111166084049517e-06,
+      "loss": 0.642,
+      "step": 3385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108481176232693e-06,
+      "loss": 0.5116,
+      "step": 3386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105796533899524e-06,
+      "loss": 0.5399,
+      "step": 3387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1031121578494695e-06,
+      "loss": 0.5299,
+      "step": 3388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1004280488819034e-06,
+      "loss": 0.4904,
+      "step": 3389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.097744207796119e-06,
+      "loss": 0.1799,
+      "step": 3390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095060635391334e-06,
+      "loss": 0.5337,
+      "step": 3391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.092377332466684e-06,
+      "loss": 0.5183,
+      "step": 3392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089694299821225e-06,
+      "loss": 0.5434,
+      "step": 3393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087011538253931e-06,
+      "loss": 0.4787,
+      "step": 3394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084329048563696e-06,
+      "loss": 0.5917,
+      "step": 3395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081646831549335e-06,
+      "loss": 0.546,
+      "step": 3396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.07896488800958e-06,
+      "loss": 0.5618,
+      "step": 3397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076283218743081e-06,
+      "loss": 0.5375,
+      "step": 3398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073601824548406e-06,
+      "loss": 0.5604,
+      "step": 3399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070920706224045e-06,
+      "loss": 0.5145,
+      "step": 3400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068239864568401e-06,
+      "loss": 0.5218,
+      "step": 3401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065559300379797e-06,
+      "loss": 0.565,
+      "step": 3402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062879014456472e-06,
+      "loss": 0.5931,
+      "step": 3403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060199007596586e-06,
+      "loss": 0.6275,
+      "step": 3404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05751928059821e-06,
+      "loss": 0.499,
+      "step": 3405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054839834259335e-06,
+      "loss": 0.188,
+      "step": 3406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052160669377873e-06,
+      "loss": 0.6021,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04948178675164e-06,
+      "loss": 0.6014,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04680318717838e-06,
+      "loss": 0.5359,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0441248714557445e-06,
+      "loss": 0.564,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.041446840381309e-06,
+      "loss": 0.5266,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038769094752554e-06,
+      "loss": 0.5339,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0360916353668835e-06,
+      "loss": 0.5742,
+      "step": 3413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033414463021612e-06,
+      "loss": 0.5057,
+      "step": 3414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030737578513968e-06,
+      "loss": 0.5045,
+      "step": 3415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028060982641097e-06,
+      "loss": 0.458,
+      "step": 3416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0253846762000585e-06,
+      "loss": 0.5815,
+      "step": 3417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022708659987823e-06,
+      "loss": 0.5174,
+      "step": 3418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.020032934801275e-06,
+      "loss": 0.568,
+      "step": 3419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017357501437216e-06,
+      "loss": 0.5412,
+      "step": 3420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014682360692359e-06,
+      "loss": 0.5358,
+      "step": 3421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.012007513363325e-06,
+      "loss": 0.5555,
+      "step": 3422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.009332960246653e-06,
+      "loss": 0.5612,
+      "step": 3423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006658702138794e-06,
+      "loss": 0.5727,
+      "step": 3424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003984739836112e-06,
+      "loss": 0.5405,
+      "step": 3425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0013110741348776e-06,
+      "loss": 0.4987,
+      "step": 3426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9986377058312765e-06,
+      "loss": 0.5655,
+      "step": 3427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.995964635721409e-06,
+      "loss": 0.5731,
+      "step": 3428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99329186460128e-06,
+      "loss": 0.619,
+      "step": 3429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.990619393266811e-06,
+      "loss": 0.5295,
+      "step": 3430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98794722251383e-06,
+      "loss": 0.5464,
+      "step": 3431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985275353138082e-06,
+      "loss": 0.6284,
+      "step": 3432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982603785935213e-06,
+      "loss": 0.521,
+      "step": 3433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979932521700784e-06,
+      "loss": 0.5205,
+      "step": 3434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977261561230268e-06,
+      "loss": 0.5311,
+      "step": 3435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9745909053190455e-06,
+      "loss": 0.5013,
+      "step": 3436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.971920554762404e-06,
+      "loss": 0.6424,
+      "step": 3437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.969250510355542e-06,
+      "loss": 0.5121,
+      "step": 3438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.96658077289357e-06,
+      "loss": 0.5284,
+      "step": 3439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963911343171498e-06,
+      "loss": 0.5235,
+      "step": 3440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961242221984256e-06,
+      "loss": 0.5159,
+      "step": 3441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.958573410126672e-06,
+      "loss": 0.5807,
+      "step": 3442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955904908393491e-06,
+      "loss": 0.4912,
+      "step": 3443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953236717579357e-06,
+      "loss": 0.6202,
+      "step": 3444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950568838478827e-06,
+      "loss": 0.5884,
+      "step": 3445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.947901271886364e-06,
+      "loss": 0.5258,
+      "step": 3446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.945234018596335e-06,
+      "loss": 0.5297,
+      "step": 3447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.942567079403019e-06,
+      "loss": 0.5446,
+      "step": 3448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.939900455100597e-06,
+      "loss": 0.5427,
+      "step": 3449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9372341464831605e-06,
+      "loss": 0.5179,
+      "step": 3450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934568154344699e-06,
+      "loss": 0.594,
+      "step": 3451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931902479479118e-06,
+      "loss": 0.5218,
+      "step": 3452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9292371226802236e-06,
+      "loss": 0.5213,
+      "step": 3453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.926572084741724e-06,
+      "loss": 0.1832,
+      "step": 3454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9239073664572365e-06,
+      "loss": 0.5336,
+      "step": 3455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921242968620284e-06,
+      "loss": 0.5774,
+      "step": 3456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918578892024293e-06,
+      "loss": 0.5426,
+      "step": 3457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.915915137462591e-06,
+      "loss": 0.5264,
+      "step": 3458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913251705728413e-06,
+      "loss": 0.5873,
+      "step": 3459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910588597614899e-06,
+      "loss": 0.5243,
+      "step": 3460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9079258139150885e-06,
+      "loss": 0.6005,
+      "step": 3461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905263355421927e-06,
+      "loss": 0.4576,
+      "step": 3462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902601222928265e-06,
+      "loss": 0.5075,
+      "step": 3463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899939417226852e-06,
+      "loss": 0.5594,
+      "step": 3464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897277939110343e-06,
+      "loss": 0.4728,
+      "step": 3465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8946167893712916e-06,
+      "loss": 0.5867,
+      "step": 3466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891955968802161e-06,
+      "loss": 0.4342,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889295478195307e-06,
+      "loss": 0.4184,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.886635318342995e-06,
+      "loss": 0.5785,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8839754900373874e-06,
+      "loss": 0.5597,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.881315994070552e-06,
+      "loss": 0.4632,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878656831234451e-06,
+      "loss": 0.4922,
+      "step": 3472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8759980023209535e-06,
+      "loss": 0.6387,
+      "step": 3473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873339508121825e-06,
+      "loss": 0.5311,
+      "step": 3474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870681349428738e-06,
+      "loss": 0.5176,
+      "step": 3475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868023527033256e-06,
+      "loss": 0.5336,
+      "step": 3476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.865366041726848e-06,
+      "loss": 0.5663,
+      "step": 3477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.862708894300883e-06,
+      "loss": 0.5949,
+      "step": 3478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860052085546624e-06,
+      "loss": 0.5388,
+      "step": 3479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.85739561625524e-06,
+      "loss": 0.5696,
+      "step": 3480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854739487217794e-06,
+      "loss": 0.5765,
+      "step": 3481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852083699225254e-06,
+      "loss": 0.5204,
+      "step": 3482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849428253068475e-06,
+      "loss": 0.6034,
+      "step": 3483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846773149538222e-06,
+      "loss": 0.4809,
+      "step": 3484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.844118389425154e-06,
+      "loss": 0.5225,
+      "step": 3485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841463973519822e-06,
+      "loss": 0.512,
+      "step": 3486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.838809902612681e-06,
+      "loss": 0.5792,
+      "step": 3487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836156177494084e-06,
+      "loss": 0.5313,
+      "step": 3488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833502798954278e-06,
+      "loss": 0.5725,
+      "step": 3489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.830849767783406e-06,
+      "loss": 0.5303,
+      "step": 3490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.828197084771509e-06,
+      "loss": 0.536,
+      "step": 3491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.825544750708526e-06,
+      "loss": 0.5366,
+      "step": 3492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.82289276638429e-06,
+      "loss": 0.6206,
+      "step": 3493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820241132588528e-06,
+      "loss": 0.5305,
+      "step": 3494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817589850110866e-06,
+      "loss": 0.5471,
+      "step": 3495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8149389197408285e-06,
+      "loss": 0.5654,
+      "step": 3496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8122883422678244e-06,
+      "loss": 0.6046,
+      "step": 3497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.809638118481166e-06,
+      "loss": 0.5625,
+      "step": 3498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8069882491700615e-06,
+      "loss": 0.5031,
+      "step": 3499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8043387351236073e-06,
+      "loss": 0.5057,
+      "step": 3500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8016895771307973e-06,
+      "loss": 0.5492,
+      "step": 3501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7990407759805183e-06,
+      "loss": 0.5218,
+      "step": 3502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796392332461555e-06,
+      "loss": 0.5062,
+      "step": 3503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7937442473625787e-06,
+      "loss": 0.5206,
+      "step": 3504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7910965214721594e-06,
+      "loss": 0.6083,
+      "step": 3505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7884491555787583e-06,
+      "loss": 0.508,
+      "step": 3506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785802150470727e-06,
+      "loss": 0.5434,
+      "step": 3507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.783155506936315e-06,
+      "loss": 0.5303,
+      "step": 3508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.780509225763659e-06,
+      "loss": 0.1766,
+      "step": 3509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777863307740794e-06,
+      "loss": 0.5591,
+      "step": 3510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7752177536556363e-06,
+      "loss": 0.5239,
+      "step": 3511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7725725642960047e-06,
+      "loss": 0.6154,
+      "step": 3512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7699277404496033e-06,
+      "loss": 0.6175,
+      "step": 3513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7672832829040324e-06,
+      "loss": 0.543,
+      "step": 3514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.764639192446774e-06,
+      "loss": 0.5129,
+      "step": 3515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7619954698652112e-06,
+      "loss": 0.577,
+      "step": 3516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7593521159466127e-06,
+      "loss": 0.5445,
+      "step": 3517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7567091314781346e-06,
+      "loss": 0.4948,
+      "step": 3518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7540665172468294e-06,
+      "loss": 0.4805,
+      "step": 3519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7514242740396343e-06,
+      "loss": 0.1846,
+      "step": 3520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7487824026433795e-06,
+      "loss": 0.5697,
+      "step": 3521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74614090384478e-06,
+      "loss": 0.5128,
+      "step": 3522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743499778430445e-06,
+      "loss": 0.5028,
+      "step": 3523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74085902718687e-06,
+      "loss": 0.1656,
+      "step": 3524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738218650900438e-06,
+      "loss": 0.5895,
+      "step": 3525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7355786503574205e-06,
+      "loss": 0.5713,
+      "step": 3526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7329390263439803e-06,
+      "loss": 0.211,
+      "step": 3527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.730299779646167e-06,
+      "loss": 0.5453,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7276609110499136e-06,
+      "loss": 0.6063,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7250224213410437e-06,
+      "loss": 0.4904,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7223843113052716e-06,
+      "loss": 0.5157,
+      "step": 3531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7197465817281917e-06,
+      "loss": 0.5618,
+      "step": 3532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.71710923339529e-06,
+      "loss": 0.553,
+      "step": 3533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.714472267091935e-06,
+      "loss": 0.4971,
+      "step": 3534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7118356836033874e-06,
+      "loss": 0.598,
+      "step": 3535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7091994837147877e-06,
+      "loss": 0.5792,
+      "step": 3536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7065636682111638e-06,
+      "loss": 0.5376,
+      "step": 3537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7039282378774356e-06,
+      "loss": 0.4791,
+      "step": 3538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7012931934983953e-06,
+      "loss": 0.2053,
+      "step": 3539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6986585358587325e-06,
+      "loss": 0.5129,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6960242657430136e-06,
+      "loss": 0.6093,
+      "step": 3541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6933903839356983e-06,
+      "loss": 0.5088,
+      "step": 3542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6907568912211183e-06,
+      "loss": 0.5841,
+      "step": 3543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6881237883835007e-06,
+      "loss": 0.6253,
+      "step": 3544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6854910762069517e-06,
+      "loss": 0.5856,
+      "step": 3545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6828587554754586e-06,
+      "loss": 0.5236,
+      "step": 3546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680226826972898e-06,
+      "loss": 0.5661,
+      "step": 3547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6775952914830266e-06,
+      "loss": 0.4946,
+      "step": 3548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.674964149789484e-06,
+      "loss": 0.5387,
+      "step": 3549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6723334026757906e-06,
+      "loss": 0.5263,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.669703050925354e-06,
+      "loss": 0.502,
+      "step": 3551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6670730953214608e-06,
+      "loss": 0.5044,
+      "step": 3552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.664443536647282e-06,
+      "loss": 0.6182,
+      "step": 3553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6618143756858637e-06,
+      "loss": 0.5732,
+      "step": 3554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659185613220144e-06,
+      "loss": 0.537,
+      "step": 3555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656557250032936e-06,
+      "loss": 0.6213,
+      "step": 3556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6539292869069326e-06,
+      "loss": 0.4962,
+      "step": 3557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6513017246247097e-06,
+      "loss": 0.5119,
+      "step": 3558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648674563968727e-06,
+      "loss": 0.6062,
+      "step": 3559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646047805721321e-06,
+      "loss": 0.5438,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6434214506647064e-06,
+      "loss": 0.6202,
+      "step": 3561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640795499580981e-06,
+      "loss": 0.5332,
+      "step": 3562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6381699532521243e-06,
+      "loss": 0.2046,
+      "step": 3563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6355448124599902e-06,
+      "loss": 0.5552,
+      "step": 3564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632920077986314e-06,
+      "loss": 0.4754,
+      "step": 3565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6302957506127113e-06,
+      "loss": 0.5733,
+      "step": 3566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627671831120677e-06,
+      "loss": 0.5044,
+      "step": 3567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6250483202915805e-06,
+      "loss": 0.6218,
+      "step": 3568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6224252189066713e-06,
+      "loss": 0.5249,
+      "step": 3569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6198025277470826e-06,
+      "loss": 0.516,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617180247593814e-06,
+      "loss": 0.579,
+      "step": 3571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6145583792277527e-06,
+      "loss": 0.5328,
+      "step": 3572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.611936923429659e-06,
+      "loss": 0.5162,
+      "step": 3573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6093158809801737e-06,
+      "loss": 0.542,
+      "step": 3574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.606695252659809e-06,
+      "loss": 0.5491,
+      "step": 3575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6040750392489567e-06,
+      "loss": 0.5587,
+      "step": 3576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601455241527887e-06,
+      "loss": 0.6136,
+      "step": 3577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5988358602767407e-06,
+      "loss": 0.5698,
+      "step": 3578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5962168962755426e-06,
+      "loss": 0.5146,
+      "step": 3579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5935983503041864e-06,
+      "loss": 0.5076,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5909802231424463e-06,
+      "loss": 0.6123,
+      "step": 3581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.588362515569964e-06,
+      "loss": 0.5411,
+      "step": 3582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5857452283662676e-06,
+      "loss": 0.4742,
+      "step": 3583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5831283623107516e-06,
+      "loss": 0.5618,
+      "step": 3584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5805119181826865e-06,
+      "loss": 0.585,
+      "step": 3585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5778958967612175e-06,
+      "loss": 0.5593,
+      "step": 3586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.575280298825367e-06,
+      "loss": 0.7005,
+      "step": 3587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.572665125154029e-06,
+      "loss": 0.5099,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.570050376525969e-06,
+      "loss": 0.2035,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.567436053719826e-06,
+      "loss": 0.5202,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5648221575141197e-06,
+      "loss": 0.5862,
+      "step": 3591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5622086886872344e-06,
+      "loss": 0.5523,
+      "step": 3592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.559595648017427e-06,
+      "loss": 0.5877,
+      "step": 3593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556983036282834e-06,
+      "loss": 0.6082,
+      "step": 3594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5543708542614598e-06,
+      "loss": 0.575,
+      "step": 3595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.551759102731178e-06,
+      "loss": 0.5349,
+      "step": 3596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5491477824697386e-06,
+      "loss": 0.5574,
+      "step": 3597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5465368942547617e-06,
+      "loss": 0.5851,
+      "step": 3598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5439264388637407e-06,
+      "loss": 0.5252,
+      "step": 3599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5413164170740345e-06,
+      "loss": 0.4981,
+      "step": 3600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5387068296628763e-06,
+      "loss": 0.5807,
+      "step": 3601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.536097677407374e-06,
+      "loss": 0.5236,
+      "step": 3602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.533488961084499e-06,
+      "loss": 0.5465,
+      "step": 3603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530880681471096e-06,
+      "loss": 0.6031,
+      "step": 3604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5282728393438793e-06,
+      "loss": 0.5322,
+      "step": 3605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.525665435479435e-06,
+      "loss": 0.5248,
+      "step": 3606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5230584706542157e-06,
+      "loss": 0.5689,
+      "step": 3607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.520451945644544e-06,
+      "loss": 0.6214,
+      "step": 3608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517845861226613e-06,
+      "loss": 0.1683,
+      "step": 3609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5152402181764812e-06,
+      "loss": 0.5196,
+      "step": 3610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51263501727008e-06,
+      "loss": 0.5831,
+      "step": 3611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5100302592832068e-06,
+      "loss": 0.594,
+      "step": 3612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507425944991529e-06,
+      "loss": 0.5736,
+      "step": 3613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504822075170576e-06,
+      "loss": 0.521,
+      "step": 3614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5022186505957524e-06,
+      "loss": 0.5378,
+      "step": 3615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499615672042328e-06,
+      "loss": 0.6083,
+      "step": 3616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4970131402854357e-06,
+      "loss": 0.5358,
+      "step": 3617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4944110561000785e-06,
+      "loss": 0.5672,
+      "step": 3618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.491809420261128e-06,
+      "loss": 0.5273,
+      "step": 3619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.489208233543321e-06,
+      "loss": 0.2208,
+      "step": 3620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.486607496721256e-06,
+      "loss": 0.4871,
+      "step": 3621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4840072105694045e-06,
+      "loss": 0.495,
+      "step": 3622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4814073758621004e-06,
+      "loss": 0.5439,
+      "step": 3623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4788079933735425e-06,
+      "loss": 0.5335,
+      "step": 3624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4762090638777944e-06,
+      "loss": 0.5618,
+      "step": 3625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47361058814879e-06,
+      "loss": 0.557,
+      "step": 3626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4710125669603235e-06,
+      "loss": 0.5282,
+      "step": 3627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468415001086052e-06,
+      "loss": 0.5294,
+      "step": 3628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4658178912995e-06,
+      "loss": 0.5401,
+      "step": 3629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4632212383740577e-06,
+      "loss": 0.4923,
+      "step": 3630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4606250430829773e-06,
+      "loss": 0.5846,
+      "step": 3631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4580293061993737e-06,
+      "loss": 0.5836,
+      "step": 3632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.455434028496225e-06,
+      "loss": 0.4733,
+      "step": 3633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452839210746377e-06,
+      "loss": 0.5035,
+      "step": 3634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4502448537225343e-06,
+      "loss": 0.618,
+      "step": 3635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447650958197264e-06,
+      "loss": 0.5707,
+      "step": 3636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4450575249429975e-06,
+      "loss": 0.5539,
+      "step": 3637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4424645547320305e-06,
+      "loss": 0.5927,
+      "step": 3638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4398720483365166e-06,
+      "loss": 0.5664,
+      "step": 3639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437280006528474e-06,
+      "loss": 0.5653,
+      "step": 3640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4346884300797817e-06,
+      "loss": 0.5606,
+      "step": 3641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4320973197621787e-06,
+      "loss": 0.5782,
+      "step": 3642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4295066763472683e-06,
+      "loss": 0.5394,
+      "step": 3643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.426916500606512e-06,
+      "loss": 0.5402,
+      "step": 3644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4243267933112377e-06,
+      "loss": 0.5202,
+      "step": 3645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4217375552326225e-06,
+      "loss": 0.1791,
+      "step": 3646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4191487871417138e-06,
+      "loss": 0.1814,
+      "step": 3647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4165604898094185e-06,
+      "loss": 0.5337,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.413972664006494e-06,
+      "loss": 0.5527,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.41138531050357e-06,
+      "loss": 0.5302,
+      "step": 3650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4087984300711274e-06,
+      "loss": 0.1978,
+      "step": 3651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406212023479509e-06,
+      "loss": 0.5394,
+      "step": 3652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4036260914989128e-06,
+      "loss": 0.5594,
+      "step": 3653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401040634899401e-06,
+      "loss": 0.6007,
+      "step": 3654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3984556544508934e-06,
+      "loss": 0.4907,
+      "step": 3655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3958711509231627e-06,
+      "loss": 0.5305,
+      "step": 3656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3932871250858442e-06,
+      "loss": 0.5661,
+      "step": 3657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3907035777084306e-06,
+      "loss": 0.5839,
+      "step": 3658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3881205095602733e-06,
+      "loss": 0.5798,
+      "step": 3659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3855379214105753e-06,
+      "loss": 0.504,
+      "step": 3660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3829558140284024e-06,
+      "loss": 0.5405,
+      "step": 3661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.380374188182677e-06,
+      "loss": 0.5495,
+      "step": 3662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377793044642173e-06,
+      "loss": 0.5295,
+      "step": 3663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3752123841755263e-06,
+      "loss": 0.529,
+      "step": 3664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.372632207551225e-06,
+      "loss": 0.5985,
+      "step": 3665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3700525155376185e-06,
+      "loss": 0.5722,
+      "step": 3666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3674733089029055e-06,
+      "loss": 0.5479,
+      "step": 3667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3648945884151427e-06,
+      "loss": 0.5538,
+      "step": 3668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3623163548422424e-06,
+      "loss": 0.5487,
+      "step": 3669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.359738608951974e-06,
+      "loss": 0.523,
+      "step": 3670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357161351511957e-06,
+      "loss": 0.4849,
+      "step": 3671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354584583289667e-06,
+      "loss": 0.2273,
+      "step": 3672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.352008305052441e-06,
+      "loss": 0.5089,
+      "step": 3673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349432517567455e-06,
+      "loss": 0.6172,
+      "step": 3674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3468572216017536e-06,
+      "loss": 0.5205,
+      "step": 3675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3442824179222266e-06,
+      "loss": 0.5885,
+      "step": 3676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3417081072956247e-06,
+      "loss": 0.4922,
+      "step": 3677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3391342904885394e-06,
+      "loss": 0.5109,
+      "step": 3678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3365609682674268e-06,
+      "loss": 0.5328,
+      "step": 3679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.333988141398592e-06,
+      "loss": 0.5445,
+      "step": 3680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33141581064819e-06,
+      "loss": 0.5195,
+      "step": 3681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.328843976782231e-06,
+      "loss": 0.5365,
+      "step": 3682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.326272640566578e-06,
+      "loss": 0.5989,
+      "step": 3683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.323701802766943e-06,
+      "loss": 0.5916,
+      "step": 3684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.321131464148889e-06,
+      "loss": 0.4992,
+      "step": 3685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.318561625477834e-06,
+      "loss": 0.5806,
+      "step": 3686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315992287519047e-06,
+      "loss": 0.535,
+      "step": 3687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.313423451037643e-06,
+      "loss": 0.2116,
+      "step": 3688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3108551167985913e-06,
+      "loss": 0.5448,
+      "step": 3689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3082872855667126e-06,
+      "loss": 0.5489,
+      "step": 3690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3057199581066767e-06,
+      "loss": 0.5599,
+      "step": 3691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303153135183001e-06,
+      "loss": 0.5815,
+      "step": 3692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3005868175600554e-06,
+      "loss": 0.4835,
+      "step": 3693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.29802100600206e-06,
+      "loss": 0.6168,
+      "step": 3694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2954557012730815e-06,
+      "loss": 0.5827,
+      "step": 3695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292890904137037e-06,
+      "loss": 0.1901,
+      "step": 3696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.290326615357692e-06,
+      "loss": 0.5326,
+      "step": 3697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2877628356986633e-06,
+      "loss": 0.1685,
+      "step": 3698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2851995659234128e-06,
+      "loss": 0.4943,
+      "step": 3699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.282636806795249e-06,
+      "loss": 0.516,
+      "step": 3700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2800745590773365e-06,
+      "loss": 0.6344,
+      "step": 3701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2775128235326757e-06,
+      "loss": 0.5258,
+      "step": 3702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2749516009241254e-06,
+      "loss": 0.6171,
+      "step": 3703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2723908920143845e-06,
+      "loss": 0.6168,
+      "step": 3704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2698306975660055e-06,
+      "loss": 0.611,
+      "step": 3705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.267271018341378e-06,
+      "loss": 0.4979,
+      "step": 3706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264711855102748e-06,
+      "loss": 0.4936,
+      "step": 3707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262153208612203e-06,
+      "loss": 0.5568,
+      "step": 3708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2595950796316754e-06,
+      "loss": 0.4872,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2570374689229477e-06,
+      "loss": 0.5892,
+      "step": 3710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.254480377247646e-06,
+      "loss": 0.1915,
+      "step": 3711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251923805367241e-06,
+      "loss": 0.5187,
+      "step": 3712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.249367754043047e-06,
+      "loss": 0.5792,
+      "step": 3713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2468122240362287e-06,
+      "loss": 0.5157,
+      "step": 3714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.244257216107791e-06,
+      "loss": 0.4971,
+      "step": 3715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2417027310185865e-06,
+      "loss": 0.5215,
+      "step": 3716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.239148769529307e-06,
+      "loss": 0.5299,
+      "step": 3717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2365953324004938e-06,
+      "loss": 0.5168,
+      "step": 3718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234042420392531e-06,
+      "loss": 0.5223,
+      "step": 3719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2314900342656443e-06,
+      "loss": 0.5274,
+      "step": 3720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2289381747799012e-06,
+      "loss": 0.5585,
+      "step": 3721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2263868426952193e-06,
+      "loss": 0.5455,
+      "step": 3722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2238360387713535e-06,
+      "loss": 0.5909,
+      "step": 3723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221285763767901e-06,
+      "loss": 0.5435,
+      "step": 3724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218736018444304e-06,
+      "loss": 0.5716,
+      "step": 3725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.216186803559849e-06,
+      "loss": 0.5187,
+      "step": 3726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2136381198736572e-06,
+      "loss": 0.5274,
+      "step": 3727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211089968144697e-06,
+      "loss": 0.5905,
+      "step": 3728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2085423491317814e-06,
+      "loss": 0.5469,
+      "step": 3729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.205995263593559e-06,
+      "loss": 0.5611,
+      "step": 3730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2034487122885187e-06,
+      "loss": 0.5531,
+      "step": 3731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.200902695974995e-06,
+      "loss": 0.5334,
+      "step": 3732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1983572154111643e-06,
+      "loss": 0.6131,
+      "step": 3733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1958122713550344e-06,
+      "loss": 0.4591,
+      "step": 3734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193267864564463e-06,
+      "loss": 0.4723,
+      "step": 3735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1907239957971424e-06,
+      "loss": 0.5409,
+      "step": 3736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1881806658106096e-06,
+      "loss": 0.5983,
+      "step": 3737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.185637875362234e-06,
+      "loss": 0.5662,
+      "step": 3738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1830956252092305e-06,
+      "loss": 0.5442,
+      "step": 3739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.18055391610865e-06,
+      "loss": 0.5542,
+      "step": 3740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1780127488173817e-06,
+      "loss": 0.6,
+      "step": 3741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.175472124092157e-06,
+      "loss": 0.6044,
+      "step": 3742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1729320426895434e-06,
+      "loss": 0.5287,
+      "step": 3743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1703925053659466e-06,
+      "loss": 0.5742,
+      "step": 3744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.167853512877609e-06,
+      "loss": 0.5477,
+      "step": 3745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165315065980614e-06,
+      "loss": 0.5072,
+      "step": 3746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1627771654308815e-06,
+      "loss": 0.5711,
+      "step": 3747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160239811984166e-06,
+      "loss": 0.6054,
+      "step": 3748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1577030063960603e-06,
+      "loss": 0.4985,
+      "step": 3749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.155166749421997e-06,
+      "loss": 0.5316,
+      "step": 3750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152631041817244e-06,
+      "loss": 0.5438,
+      "step": 3751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1500958843369023e-06,
+      "loss": 0.6033,
+      "step": 3752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1475612777359105e-06,
+      "loss": 0.5457,
+      "step": 3753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1450272227690477e-06,
+      "loss": 0.5749,
+      "step": 3754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1424937201909243e-06,
+      "loss": 0.6193,
+      "step": 3755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1399607707559843e-06,
+      "loss": 0.579,
+      "step": 3756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1374283752185133e-06,
+      "loss": 0.5321,
+      "step": 3757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.134896534332627e-06,
+      "loss": 0.2074,
+      "step": 3758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.132365248852277e-06,
+      "loss": 0.4983,
+      "step": 3759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.12983451953125e-06,
+      "loss": 0.5816,
+      "step": 3760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127304347123167e-06,
+      "loss": 0.5551,
+      "step": 3761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.124774732381485e-06,
+      "loss": 0.5867,
+      "step": 3762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.12224567605949e-06,
+      "loss": 0.594,
+      "step": 3763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.119717178910305e-06,
+      "loss": 0.221,
+      "step": 3764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1171892416868898e-06,
+      "loss": 0.5258,
+      "step": 3765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11466186514203e-06,
+      "loss": 0.5716,
+      "step": 3766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1121350500283503e-06,
+      "loss": 0.5365,
+      "step": 3767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109608797098305e-06,
+      "loss": 0.5583,
+      "step": 3768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107083107104183e-06,
+      "loss": 0.1646,
+      "step": 3769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104557980798104e-06,
+      "loss": 0.5494,
+      "step": 3770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.10203341893202e-06,
+      "loss": 0.5805,
+      "step": 3771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0995094222577176e-06,
+      "loss": 0.5781,
+      "step": 3772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.096985991526809e-06,
+      "loss": 0.6003,
+      "step": 3773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.094463127490745e-06,
+      "loss": 0.5075,
+      "step": 3774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0919408309008027e-06,
+      "loss": 0.5574,
+      "step": 3775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0894191025080945e-06,
+      "loss": 0.5977,
+      "step": 3776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0868979430635566e-06,
+      "loss": 0.5727,
+      "step": 3777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0843773533179634e-06,
+      "loss": 0.5001,
+      "step": 3778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.081857334021917e-06,
+      "loss": 0.6637,
+      "step": 3779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.079337885925846e-06,
+      "loss": 0.5553,
+      "step": 3780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0768190097800133e-06,
+      "loss": 0.6037,
+      "step": 3781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.074300706334511e-06,
+      "loss": 0.5147,
+      "step": 3782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0717829763392605e-06,
+      "loss": 0.5357,
+      "step": 3783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0692658205440083e-06,
+      "loss": 0.2084,
+      "step": 3784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066749239698336e-06,
+      "loss": 0.5322,
+      "step": 3785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0642332345516525e-06,
+      "loss": 0.5756,
+      "step": 3786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0617178058531917e-06,
+      "loss": 0.5214,
+      "step": 3787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.059202954352017e-06,
+      "loss": 0.5366,
+      "step": 3788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056688680797024e-06,
+      "loss": 0.6184,
+      "step": 3789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.054174985936934e-06,
+      "loss": 0.6488,
+      "step": 3790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.051661870520293e-06,
+      "loss": 0.5079,
+      "step": 3791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0491493352954752e-06,
+      "loss": 0.5467,
+      "step": 3792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0466373810106876e-06,
+      "loss": 0.5007,
+      "step": 3793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.04412600841396e-06,
+      "loss": 0.5828,
+      "step": 3794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0416152182531457e-06,
+      "loss": 0.572,
+      "step": 3795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.039105011275929e-06,
+      "loss": 0.5964,
+      "step": 3796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0365953882298226e-06,
+      "loss": 0.5195,
+      "step": 3797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.03408634986216e-06,
+      "loss": 0.5534,
+      "step": 3798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0315778969201016e-06,
+      "loss": 0.544,
+      "step": 3799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0290700301506358e-06,
+      "loss": 0.4956,
+      "step": 3800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026562750300578e-06,
+      "loss": 0.5807,
+      "step": 3801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0240560581165624e-06,
+      "loss": 0.2267,
+      "step": 3802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0215499543450524e-06,
+      "loss": 0.5451,
+      "step": 3803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0190444397323398e-06,
+      "loss": 0.5446,
+      "step": 3804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0165395150245315e-06,
+      "loss": 0.4956,
+      "step": 3805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.014035180967567e-06,
+      "loss": 0.1941,
+      "step": 3806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.011531438307206e-06,
+      "loss": 0.5875,
+      "step": 3807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0090282877890376e-06,
+      "loss": 0.5924,
+      "step": 3808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.006525730158463e-06,
+      "loss": 0.5252,
+      "step": 3809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0040237661607173e-06,
+      "loss": 0.5911,
+      "step": 3810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.001522396540858e-06,
+      "loss": 0.585,
+      "step": 3811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.999021622043757e-06,
+      "loss": 0.6037,
+      "step": 3812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9965214434141216e-06,
+      "loss": 0.5974,
+      "step": 3813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.994021861396471e-06,
+      "loss": 0.5465,
+      "step": 3814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.991522876735154e-06,
+      "loss": 0.5325,
+      "step": 3815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9890244901743344e-06,
+      "loss": 0.4906,
+      "step": 3816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.986526702458005e-06,
+      "loss": 0.525,
+      "step": 3817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9840295143299768e-06,
+      "loss": 0.5143,
+      "step": 3818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.981532926533881e-06,
+      "loss": 0.1936,
+      "step": 3819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.979036939813172e-06,
+      "loss": 0.2249,
+      "step": 3820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9765415549111266e-06,
+      "loss": 0.579,
+      "step": 3821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.97404677257084e-06,
+      "loss": 0.4818,
+      "step": 3822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.971552593535227e-06,
+      "loss": 0.5252,
+      "step": 3823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9690590185470243e-06,
+      "loss": 0.5674,
+      "step": 3824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9665660483487923e-06,
+      "loss": 0.5825,
+      "step": 3825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964073683682904e-06,
+      "loss": 0.5781,
+      "step": 3826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.961581925291557e-06,
+      "loss": 0.5997,
+      "step": 3827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9590907739167663e-06,
+      "loss": 0.5213,
+      "step": 3828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9566002303003697e-06,
+      "loss": 0.5756,
+      "step": 3829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9541102951840196e-06,
+      "loss": 0.5124,
+      "step": 3830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951620969309187e-06,
+      "loss": 0.5963,
+      "step": 3831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9491322534171674e-06,
+      "loss": 0.4972,
+      "step": 3832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.946644148249069e-06,
+      "loss": 0.567,
+      "step": 3833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.944156654545819e-06,
+      "loss": 0.5315,
+      "step": 3834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9416697730481624e-06,
+      "loss": 0.5507,
+      "step": 3835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.939183504496667e-06,
+      "loss": 0.4873,
+      "step": 3836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936697849631708e-06,
+      "loss": 0.5738,
+      "step": 3837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9342128091934886e-06,
+      "loss": 0.5615,
+      "step": 3838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.931728383922021e-06,
+      "loss": 0.5375,
+      "step": 3839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9292445745571405e-06,
+      "loss": 0.5553,
+      "step": 3840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9267613818384937e-06,
+      "loss": 0.5529,
+      "step": 3841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.924278806505546e-06,
+      "loss": 0.53,
+      "step": 3842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9217968492975796e-06,
+      "loss": 0.5843,
+      "step": 3843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9193155109536887e-06,
+      "loss": 0.5385,
+      "step": 3844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9168347922127905e-06,
+      "loss": 0.5702,
+      "step": 3845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9143546938136093e-06,
+      "loss": 0.5701,
+      "step": 3846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9118752164946934e-06,
+      "loss": 0.5146,
+      "step": 3847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909396360994399e-06,
+      "loss": 0.5318,
+      "step": 3848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9069181280508975e-06,
+      "loss": 0.6214,
+      "step": 3849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9044405184021817e-06,
+      "loss": 0.5305,
+      "step": 3850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.901963532786049e-06,
+      "loss": 0.5491,
+      "step": 3851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.899487171940121e-06,
+      "loss": 0.5295,
+      "step": 3852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.897011436601825e-06,
+      "loss": 0.5497,
+      "step": 3853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8945363275084084e-06,
+      "loss": 0.5002,
+      "step": 3854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8920618453969245e-06,
+      "loss": 0.5917,
+      "step": 3855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88958799100425e-06,
+      "loss": 0.532,
+      "step": 3856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8871147650670663e-06,
+      "loss": 0.6022,
+      "step": 3857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8846421683218695e-06,
+      "loss": 0.5457,
+      "step": 3858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88217020150497e-06,
+      "loss": 0.5309,
+      "step": 3859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8796988653524925e-06,
+      "loss": 0.2249,
+      "step": 3860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8772281606003704e-06,
+      "loss": 0.5658,
+      "step": 3861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8747580879843472e-06,
+      "loss": 0.5317,
+      "step": 3862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8722886482399826e-06,
+      "loss": 0.5498,
+      "step": 3863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.869819842102649e-06,
+      "loss": 0.5363,
+      "step": 3864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8673516703075247e-06,
+      "loss": 0.5381,
+      "step": 3865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8648841335896004e-06,
+      "loss": 0.5718,
+      "step": 3866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8624172326836815e-06,
+      "loss": 0.6084,
+      "step": 3867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8599509683243824e-06,
+      "loss": 0.5168,
+      "step": 3868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8574853412461256e-06,
+      "loss": 0.1938,
+      "step": 3869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8550203521831445e-06,
+      "loss": 0.5696,
+      "step": 3870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8525560018694835e-06,
+      "loss": 0.5364,
+      "step": 3871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.850092291039e-06,
+      "loss": 0.5713,
+      "step": 3872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.847629220425353e-06,
+      "loss": 0.5211,
+      "step": 3873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.845166790762021e-06,
+      "loss": 0.5286,
+      "step": 3874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8427050027822816e-06,
+      "loss": 0.5183,
+      "step": 3875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8402438572192258e-06,
+      "loss": 0.5914,
+      "step": 3876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8377833548057546e-06,
+      "loss": 0.5544,
+      "step": 3877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.835323496274579e-06,
+      "loss": 0.6455,
+      "step": 3878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8328642823582126e-06,
+      "loss": 0.6695,
+      "step": 3879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8304057137889785e-06,
+      "loss": 0.2165,
+      "step": 3880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82794779129901e-06,
+      "loss": 0.5276,
+      "step": 3881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8254905156202504e-06,
+      "loss": 0.5221,
+      "step": 3882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.823033887484444e-06,
+      "loss": 0.5936,
+      "step": 3883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.820577907623145e-06,
+      "loss": 0.5375,
+      "step": 3884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8181225767677146e-06,
+      "loss": 0.5696,
+      "step": 3885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8156678956493246e-06,
+      "loss": 0.5504,
+      "step": 3886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.813213864998947e-06,
+      "loss": 0.5243,
+      "step": 3887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8107604855473615e-06,
+      "loss": 0.5383,
+      "step": 3888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8083077580251584e-06,
+      "loss": 0.5533,
+      "step": 3889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.805855683162727e-06,
+      "loss": 0.5249,
+      "step": 3890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8034042616902703e-06,
+      "loss": 0.5516,
+      "step": 3891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8009534943377874e-06,
+      "loss": 0.5791,
+      "step": 3892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7985033818350926e-06,
+      "loss": 0.5624,
+      "step": 3893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7960539249117953e-06,
+      "loss": 0.5794,
+      "step": 3894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7936051242973185e-06,
+      "loss": 0.4824,
+      "step": 3895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.791156980720885e-06,
+      "loss": 0.5809,
+      "step": 3896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.788709494911519e-06,
+      "loss": 0.4927,
+      "step": 3897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7862626675980566e-06,
+      "loss": 0.4552,
+      "step": 3898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.783816499509133e-06,
+      "loss": 0.5228,
+      "step": 3899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7813709913731884e-06,
+      "loss": 0.5984,
+      "step": 3900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7789261439184627e-06,
+      "loss": 0.5898,
+      "step": 3901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7764819578730048e-06,
+      "loss": 0.5831,
+      "step": 3902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7740384339646655e-06,
+      "loss": 0.5632,
+      "step": 3903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7715955729210953e-06,
+      "loss": 0.5525,
+      "step": 3904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7691533754697463e-06,
+      "loss": 0.5175,
+      "step": 3905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76671184233788e-06,
+      "loss": 0.5703,
+      "step": 3906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7642709742525555e-06,
+      "loss": 0.578,
+      "step": 3907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.761830771940633e-06,
+      "loss": 0.5038,
+      "step": 3908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7593912361287733e-06,
+      "loss": 0.1794,
+      "step": 3909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7569523675434438e-06,
+      "loss": 0.5253,
+      "step": 3910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.754514166910912e-06,
+      "loss": 0.5628,
+      "step": 3911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7520766349572426e-06,
+      "loss": 0.563,
+      "step": 3912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.749639772408302e-06,
+      "loss": 0.5913,
+      "step": 3913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7472035799897623e-06,
+      "loss": 0.5312,
+      "step": 3914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7447680584270888e-06,
+      "loss": 0.5093,
+      "step": 3915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7423332084455543e-06,
+      "loss": 0.5332,
+      "step": 3916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.739899030770225e-06,
+      "loss": 0.5179,
+      "step": 3917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737465526125973e-06,
+      "loss": 0.5809,
+      "step": 3918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7350326952374638e-06,
+      "loss": 0.608,
+      "step": 3919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7326005388291664e-06,
+      "loss": 0.6163,
+      "step": 3920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7301690576253526e-06,
+      "loss": 0.5702,
+      "step": 3921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7277382523500804e-06,
+      "loss": 0.5453,
+      "step": 3922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7253081237272183e-06,
+      "loss": 0.5765,
+      "step": 3923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7228786724804294e-06,
+      "loss": 0.5472,
+      "step": 3924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7204498993331797e-06,
+      "loss": 0.5647,
+      "step": 3925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.718021805008721e-06,
+      "loss": 0.5292,
+      "step": 3926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.715594390230114e-06,
+      "loss": 0.2045,
+      "step": 3927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.713167655720217e-06,
+      "loss": 0.5495,
+      "step": 3928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7107416022016774e-06,
+      "loss": 0.5492,
+      "step": 3929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7083162303969503e-06,
+      "loss": 0.6301,
+      "step": 3930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7058915410282773e-06,
+      "loss": 0.5908,
+      "step": 3931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7034675348177055e-06,
+      "loss": 0.6351,
+      "step": 3932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.701044212487072e-06,
+      "loss": 0.6178,
+      "step": 3933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.698621574758017e-06,
+      "loss": 0.5302,
+      "step": 3934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6961996223519716e-06,
+      "loss": 0.5441,
+      "step": 3935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.693778355990161e-06,
+      "loss": 0.5851,
+      "step": 3936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6913577763936127e-06,
+      "loss": 0.5604,
+      "step": 3937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6889378842831483e-06,
+      "loss": 0.6145,
+      "step": 3938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6865186803793797e-06,
+      "loss": 0.5607,
+      "step": 3939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6841001654027154e-06,
+      "loss": 0.5444,
+      "step": 3940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6816823400733628e-06,
+      "loss": 0.5492,
+      "step": 3941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6792652051113232e-06,
+      "loss": 0.6521,
+      "step": 3942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.676848761236388e-06,
+      "loss": 0.5307,
+      "step": 3943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.674433009168144e-06,
+      "loss": 0.5685,
+      "step": 3944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6720179496259753e-06,
+      "loss": 0.5713,
+      "step": 3945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669603583329059e-06,
+      "loss": 0.5519,
+      "step": 3946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.667189910996363e-06,
+      "loss": 0.5437,
+      "step": 3947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.664776933346648e-06,
+      "loss": 0.5245,
+      "step": 3948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6623646510984726e-06,
+      "loss": 0.5752,
+      "step": 3949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6599530649701872e-06,
+      "loss": 0.5071,
+      "step": 3950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6575421756799313e-06,
+      "loss": 0.5051,
+      "step": 3951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6551319839456363e-06,
+      "loss": 0.639,
+      "step": 3952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6527224904850336e-06,
+      "loss": 0.2074,
+      "step": 3953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6503136960156366e-06,
+      "loss": 0.579,
+      "step": 3954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6479056012547604e-06,
+      "loss": 0.5165,
+      "step": 3955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6454982069195025e-06,
+      "loss": 0.57,
+      "step": 3956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.64309151372676e-06,
+      "loss": 0.5772,
+      "step": 3957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6406855223932136e-06,
+      "loss": 0.6347,
+      "step": 3958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6382802336353425e-06,
+      "loss": 0.5134,
+      "step": 3959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6358756481694115e-06,
+      "loss": 0.5307,
+      "step": 3960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.633471766711475e-06,
+      "loss": 0.5677,
+      "step": 3961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6310685899773824e-06,
+      "loss": 0.5515,
+      "step": 3962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.628666118682773e-06,
+      "loss": 0.5137,
+      "step": 3963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.626264353543072e-06,
+      "loss": 0.5648,
+      "step": 3964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.623863295273494e-06,
+      "loss": 0.573,
+      "step": 3965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6214629445890484e-06,
+      "loss": 0.5949,
+      "step": 3966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.619063302204532e-06,
+      "loss": 0.5864,
+      "step": 3967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6166643688345284e-06,
+      "loss": 0.5395,
+      "step": 3968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.614266145193408e-06,
+      "loss": 0.5106,
+      "step": 3969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.611868631995337e-06,
+      "loss": 0.5122,
+      "step": 3970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6094718299542677e-06,
+      "loss": 0.6637,
+      "step": 3971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6070757397839364e-06,
+      "loss": 0.5419,
+      "step": 3972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6046803621978677e-06,
+      "loss": 0.5469,
+      "step": 3973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.602285697909382e-06,
+      "loss": 0.5589,
+      "step": 3974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5998917476315767e-06,
+      "loss": 0.5578,
+      "step": 3975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5974985120773445e-06,
+      "loss": 0.54,
+      "step": 3976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5951059919593624e-06,
+      "loss": 0.556,
+      "step": 3977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.592714187990094e-06,
+      "loss": 0.5593,
+      "step": 3978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5903231008817888e-06,
+      "loss": 0.2199,
+      "step": 3979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5879327313464836e-06,
+      "loss": 0.5954,
+      "step": 3980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.585543080096007e-06,
+      "loss": 0.5624,
+      "step": 3981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.583154147841961e-06,
+      "loss": 0.5185,
+      "step": 3982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5807659352957446e-06,
+      "loss": 0.6393,
+      "step": 3983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.57837844316854e-06,
+      "loss": 0.5302,
+      "step": 3984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575991672171314e-06,
+      "loss": 0.1607,
+      "step": 3985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5736056230148176e-06,
+      "loss": 0.2053,
+      "step": 3986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5712202964095857e-06,
+      "loss": 0.6188,
+      "step": 3987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.568835693065942e-06,
+      "loss": 0.5079,
+      "step": 3988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.566451813693995e-06,
+      "loss": 0.5513,
+      "step": 3989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5640686590036334e-06,
+      "loss": 0.5617,
+      "step": 3990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5616862297045297e-06,
+      "loss": 0.5813,
+      "step": 3991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559304526506148e-06,
+      "loss": 0.576,
+      "step": 3992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.556923550117726e-06,
+      "loss": 0.5205,
+      "step": 3993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5545433012482947e-06,
+      "loss": 0.4972,
+      "step": 3994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5521637806066596e-06,
+      "loss": 0.5376,
+      "step": 3995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.549784988901417e-06,
+      "loss": 0.5493,
+      "step": 3996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5474069268409397e-06,
+      "loss": 0.5542,
+      "step": 3997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5450295951333896e-06,
+      "loss": 0.5659,
+      "step": 3998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5426529944867057e-06,
+      "loss": 0.1821,
+      "step": 3999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.540277125608609e-06,
+      "loss": 0.561,
+      "step": 4000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537901989206608e-06,
+      "loss": 0.4945,
+      "step": 4001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5355275859879904e-06,
+      "loss": 0.6468,
+      "step": 4002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.533153916659824e-06,
+      "loss": 0.5562,
+      "step": 4003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.530780981928957e-06,
+      "loss": 0.5922,
+      "step": 4004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528408782502023e-06,
+      "loss": 0.5495,
+      "step": 4005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5260373190854372e-06,
+      "loss": 0.6203,
+      "step": 4006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.52366659238539e-06,
+      "loss": 0.5097,
+      "step": 4007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5212966031078544e-06,
+      "loss": 0.5371,
+      "step": 4008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5189273519585867e-06,
+      "loss": 0.5637,
+      "step": 4009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.516558839643123e-06,
+      "loss": 0.5441,
+      "step": 4010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.514191066866777e-06,
+      "loss": 0.5675,
+      "step": 4011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5118240343346402e-06,
+      "loss": 0.5033,
+      "step": 4012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.509457742751591e-06,
+      "loss": 0.4968,
+      "step": 4013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.507092192822279e-06,
+      "loss": 0.5303,
+      "step": 4014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.504727385251141e-06,
+      "loss": 0.6041,
+      "step": 4015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5023633207423836e-06,
+      "loss": 0.5128,
+      "step": 4016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.5457,
+      "step": 4017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4976374237277588e-06,
+      "loss": 0.5302,
+      "step": 4018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.495275592629207e-06,
+      "loss": 0.5446,
+      "step": 4019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4929145074076694e-06,
+      "loss": 0.5157,
+      "step": 4020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4905541687662473e-06,
+      "loss": 0.6223,
+      "step": 4021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.488194577407822e-06,
+      "loss": 0.5203,
+      "step": 4022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4858357340350526e-06,
+      "loss": 0.567,
+      "step": 4023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.483477639350378e-06,
+      "loss": 0.5851,
+      "step": 4024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4811202940560033e-06,
+      "loss": 0.5479,
+      "step": 4025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4787636988539216e-06,
+      "loss": 0.5326,
+      "step": 4026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476407854445898e-06,
+      "loss": 0.6415,
+      "step": 4027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4740527615334798e-06,
+      "loss": 0.6222,
+      "step": 4028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4716984208179774e-06,
+      "loss": 0.2105,
+      "step": 4029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.469344833000489e-06,
+      "loss": 0.505,
+      "step": 4030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.466991998781886e-06,
+      "loss": 0.5059,
+      "step": 4031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.464639918862813e-06,
+      "loss": 0.5553,
+      "step": 4032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4622885939436923e-06,
+      "loss": 0.5199,
+      "step": 4033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459938024724719e-06,
+      "loss": 0.6044,
+      "step": 4034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4575882119058663e-06,
+      "loss": 0.6295,
+      "step": 4035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4552391561868783e-06,
+      "loss": 0.5957,
+      "step": 4036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.452890858267278e-06,
+      "loss": 0.6059,
+      "step": 4037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4505433188463603e-06,
+      "loss": 0.1876,
+      "step": 4038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4481965386231906e-06,
+      "loss": 0.1941,
+      "step": 4039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4458505182966153e-06,
+      "loss": 0.6066,
+      "step": 4040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4435052585652524e-06,
+      "loss": 0.6441,
+      "step": 4041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4411607601274907e-06,
+      "loss": 0.5538,
+      "step": 4042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4388170236814925e-06,
+      "loss": 0.5491,
+      "step": 4043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4364740499251947e-06,
+      "loss": 0.195,
+      "step": 4044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4341318395563106e-06,
+      "loss": 0.5869,
+      "step": 4045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4317903932723198e-06,
+      "loss": 0.5773,
+      "step": 4046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4294497117704747e-06,
+      "loss": 0.532,
+      "step": 4047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4271097957478052e-06,
+      "loss": 0.5985,
+      "step": 4048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.424770645901111e-06,
+      "loss": 0.6019,
+      "step": 4049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4224322629269626e-06,
+      "loss": 0.6052,
+      "step": 4050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4200946475216995e-06,
+      "loss": 0.4924,
+      "step": 4051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.41775780038144e-06,
+      "loss": 0.585,
+      "step": 4052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4154217222020652e-06,
+      "loss": 0.5704,
+      "step": 4053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4130864136792353e-06,
+      "loss": 0.5591,
+      "step": 4054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410751875508373e-06,
+      "loss": 0.6196,
+      "step": 4055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.40841810838468e-06,
+      "loss": 0.5505,
+      "step": 4056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.406085113003121e-06,
+      "loss": 0.4581,
+      "step": 4057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4037528900584366e-06,
+      "loss": 0.5963,
+      "step": 4058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4014214402451346e-06,
+      "loss": 0.5581,
+      "step": 4059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.39909076425749e-06,
+      "loss": 0.5016,
+      "step": 4060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3967608627895526e-06,
+      "loss": 0.5735,
+      "step": 4061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3944317365351413e-06,
+      "loss": 0.5724,
+      "step": 4062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.392103386187839e-06,
+      "loss": 0.5494,
+      "step": 4063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3897758124410004e-06,
+      "loss": 0.63,
+      "step": 4064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3874490159877495e-06,
+      "loss": 0.5676,
+      "step": 4065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3851229975209817e-06,
+      "loss": 0.5911,
+      "step": 4066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.382797757733355e-06,
+      "loss": 0.5946,
+      "step": 4067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3804732973172968e-06,
+      "loss": 0.5769,
+      "step": 4068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3781496169650044e-06,
+      "loss": 0.5511,
+      "step": 4069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.375826717368444e-06,
+      "loss": 0.609,
+      "step": 4070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.373504599219347e-06,
+      "loss": 0.536,
+      "step": 4071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3711832632092085e-06,
+      "loss": 0.5788,
+      "step": 4072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3688627100292984e-06,
+      "loss": 0.539,
+      "step": 4073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3665429403706506e-06,
+      "loss": 0.559,
+      "step": 4074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.364223954924063e-06,
+      "loss": 0.5642,
+      "step": 4075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3619057543801e-06,
+      "loss": 0.5318,
+      "step": 4076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.359588339429098e-06,
+      "loss": 0.5086,
+      "step": 4077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3572717107611514e-06,
+      "loss": 0.5276,
+      "step": 4078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.354955869066126e-06,
+      "loss": 0.6371,
+      "step": 4079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352640815033655e-06,
+      "loss": 0.5615,
+      "step": 4080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.350326549353131e-06,
+      "loss": 0.5346,
+      "step": 4081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.348013072713713e-06,
+      "loss": 0.5194,
+      "step": 4082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.345700385804328e-06,
+      "loss": 0.5251,
+      "step": 4083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3433884893136715e-06,
+      "loss": 0.5527,
+      "step": 4084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3410773839301905e-06,
+      "loss": 0.5913,
+      "step": 4085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338767070342108e-06,
+      "loss": 0.2008,
+      "step": 4086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.336457549237408e-06,
+      "loss": 0.539,
+      "step": 4087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.334148821303841e-06,
+      "loss": 0.5969,
+      "step": 4088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3318408872289154e-06,
+      "loss": 0.552,
+      "step": 4089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329533747699906e-06,
+      "loss": 0.5625,
+      "step": 4090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3272274034038538e-06,
+      "loss": 0.5728,
+      "step": 4091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3249218550275575e-06,
+      "loss": 0.5296,
+      "step": 4092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3226171032575856e-06,
+      "loss": 0.4962,
+      "step": 4093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3203131487802616e-06,
+      "loss": 0.6279,
+      "step": 4094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3180099922816794e-06,
+      "loss": 0.5762,
+      "step": 4095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.315707634447688e-06,
+      "loss": 0.5464,
+      "step": 4096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.313406075963906e-06,
+      "loss": 0.22,
+      "step": 4097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311105317515708e-06,
+      "loss": 0.5645,
+      "step": 4098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3088053597882305e-06,
+      "loss": 0.565,
+      "step": 4099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3065062034663742e-06,
+      "loss": 0.5457,
+      "step": 4100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.304207849234804e-06,
+      "loss": 0.5728,
+      "step": 4101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30191029777794e-06,
+      "loss": 0.6074,
+      "step": 4102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.299613549779963e-06,
+      "loss": 0.5461,
+      "step": 4103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2973176059248196e-06,
+      "loss": 0.5803,
+      "step": 4104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2950224668962168e-06,
+      "loss": 0.4684,
+      "step": 4105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292728133377616e-06,
+      "loss": 0.5479,
+      "step": 4106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2904346060522425e-06,
+      "loss": 0.5162,
+      "step": 4107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288141885603082e-06,
+      "loss": 0.5578,
+      "step": 4108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2858499727128814e-06,
+      "loss": 0.5633,
+      "step": 4109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2835588680641428e-06,
+      "loss": 0.5511,
+      "step": 4110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2812685723391287e-06,
+      "loss": 0.5511,
+      "step": 4111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278979086219863e-06,
+      "loss": 0.5415,
+      "step": 4112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2766904103881294e-06,
+      "loss": 0.5304,
+      "step": 4113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2744025455254666e-06,
+      "loss": 0.6039,
+      "step": 4114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272115492313171e-06,
+      "loss": 0.5051,
+      "step": 4115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2698292514323038e-06,
+      "loss": 0.5402,
+      "step": 4116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2675438235636773e-06,
+      "loss": 0.5098,
+      "step": 4117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.265259209387867e-06,
+      "loss": 0.5545,
+      "step": 4118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2629754095852e-06,
+      "loss": 0.5988,
+      "step": 4119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26069242483577e-06,
+      "loss": 0.61,
+      "step": 4120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.258410255819417e-06,
+      "loss": 0.4941,
+      "step": 4121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2561289032157484e-06,
+      "loss": 0.6093,
+      "step": 4122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2538483677041217e-06,
+      "loss": 0.5715,
+      "step": 4123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2515686499636514e-06,
+      "loss": 0.1747,
+      "step": 4124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2492897506732113e-06,
+      "loss": 0.5683,
+      "step": 4125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247011670511432e-06,
+      "loss": 0.6102,
+      "step": 4126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2447344101567002e-06,
+      "loss": 0.4756,
+      "step": 4127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2424579702871513e-06,
+      "loss": 0.531,
+      "step": 4128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.240182351580685e-06,
+      "loss": 0.5592,
+      "step": 4129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.237907554714954e-06,
+      "loss": 0.5962,
+      "step": 4130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2356335803673655e-06,
+      "loss": 0.5438,
+      "step": 4131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233360429215079e-06,
+      "loss": 0.5976,
+      "step": 4132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2310881019350143e-06,
+      "loss": 0.5652,
+      "step": 4133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2288165992038447e-06,
+      "loss": 0.5733,
+      "step": 4134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2265459216979935e-06,
+      "loss": 0.522,
+      "step": 4135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.224276070093644e-06,
+      "loss": 0.5345,
+      "step": 4136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.22200704506673e-06,
+      "loss": 0.5617,
+      "step": 4137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2197388472929376e-06,
+      "loss": 0.5553,
+      "step": 4138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2174714774477113e-06,
+      "loss": 0.5771,
+      "step": 4139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2152049362062484e-06,
+      "loss": 0.5779,
+      "step": 4140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2129392242434953e-06,
+      "loss": 0.5721,
+      "step": 4141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2106743422341535e-06,
+      "loss": 0.5278,
+      "step": 4142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208410290852679e-06,
+      "loss": 0.5904,
+      "step": 4143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2061470707732807e-06,
+      "loss": 0.5088,
+      "step": 4144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2038846826699174e-06,
+      "loss": 0.5907,
+      "step": 4145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2016231272162987e-06,
+      "loss": 0.6233,
+      "step": 4146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.199362405085891e-06,
+      "loss": 0.5463,
+      "step": 4147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197102516951912e-06,
+      "loss": 0.5825,
+      "step": 4148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1948434634873276e-06,
+      "loss": 0.5639,
+      "step": 4149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.192585245364856e-06,
+      "loss": 0.5575,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.190327863256968e-06,
+      "loss": 0.5306,
+      "step": 4151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.188071317835888e-06,
+      "loss": 0.5209,
+      "step": 4152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1858156097735855e-06,
+      "loss": 0.1992,
+      "step": 4153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1835607397417825e-06,
+      "loss": 0.5674,
+      "step": 4154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.181306708411955e-06,
+      "loss": 0.553,
+      "step": 4155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.179053516455324e-06,
+      "loss": 0.5302,
+      "step": 4156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.176801164542866e-06,
+      "loss": 0.516,
+      "step": 4157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1745496533453016e-06,
+      "loss": 0.6023,
+      "step": 4158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1722989835331066e-06,
+      "loss": 0.5574,
+      "step": 4159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1700491557765006e-06,
+      "loss": 0.6059,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1678001707454595e-06,
+      "loss": 0.5103,
+      "step": 4161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1655520291097014e-06,
+      "loss": 0.5052,
+      "step": 4162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.163304731538695e-06,
+      "loss": 0.5422,
+      "step": 4163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1610582787016597e-06,
+      "loss": 0.5741,
+      "step": 4164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158812671267565e-06,
+      "loss": 0.5975,
+      "step": 4165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.156567909905124e-06,
+      "loss": 0.5386,
+      "step": 4166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.154323995282797e-06,
+      "loss": 0.6509,
+      "step": 4167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.152080928068798e-06,
+      "loss": 0.5874,
+      "step": 4168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.149838708931087e-06,
+      "loss": 0.5362,
+      "step": 4169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.147597338537368e-06,
+      "loss": 0.6378,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.145356817555093e-06,
+      "loss": 0.5988,
+      "step": 4171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1431171466514645e-06,
+      "loss": 0.4941,
+      "step": 4172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.14087832649343e-06,
+      "loss": 0.5497,
+      "step": 4173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1386403577476835e-06,
+      "loss": 0.5688,
+      "step": 4174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1364032410806623e-06,
+      "loss": 0.6095,
+      "step": 4175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134166977158557e-06,
+      "loss": 0.5853,
+      "step": 4176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1319315666472973e-06,
+      "loss": 0.5785,
+      "step": 4177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1296970102125643e-06,
+      "loss": 0.2047,
+      "step": 4178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1274633085197786e-06,
+      "loss": 0.5533,
+      "step": 4179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1252304622341146e-06,
+      "loss": 0.576,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1229984720204827e-06,
+      "loss": 0.5587,
+      "step": 4181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1207673385435457e-06,
+      "loss": 0.5312,
+      "step": 4182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118537062467709e-06,
+      "loss": 0.5722,
+      "step": 4183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1163076444571214e-06,
+      "loss": 0.5039,
+      "step": 4184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1140790851756743e-06,
+      "loss": 0.5603,
+      "step": 4185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111851385287009e-06,
+      "loss": 0.2341,
+      "step": 4186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1096245454545108e-06,
+      "loss": 0.6161,
+      "step": 4187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1073985663412984e-06,
+      "loss": 0.4723,
+      "step": 4188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105173448610246e-06,
+      "loss": 0.5161,
+      "step": 4189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1029491929239664e-06,
+      "loss": 0.595,
+      "step": 4190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1007257999448192e-06,
+      "loss": 0.5533,
+      "step": 4191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.098503270334902e-06,
+      "loss": 0.5223,
+      "step": 4192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0962816047560564e-06,
+      "loss": 0.6068,
+      "step": 4193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.09406080386987e-06,
+      "loss": 0.5716,
+      "step": 4194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0918408683376685e-06,
+      "loss": 0.5495,
+      "step": 4195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0896217988205254e-06,
+      "loss": 0.5127,
+      "step": 4196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.087403595979249e-06,
+      "loss": 0.5843,
+      "step": 4197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085186260474398e-06,
+      "loss": 0.5629,
+      "step": 4198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.082969792966264e-06,
+      "loss": 0.2357,
+      "step": 4199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0807541941148883e-06,
+      "loss": 0.5418,
+      "step": 4200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0785394645800487e-06,
+      "loss": 0.4623,
+      "step": 4201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0763256050212634e-06,
+      "loss": 0.5605,
+      "step": 4202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0741126160977944e-06,
+      "loss": 0.5847,
+      "step": 4203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.071900498468646e-06,
+      "loss": 0.5395,
+      "step": 4204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.069689252792557e-06,
+      "loss": 0.5268,
+      "step": 4205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.06747887972801e-06,
+      "loss": 0.5717,
+      "step": 4206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0652693799332286e-06,
+      "loss": 0.5092,
+      "step": 4207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0630607540661773e-06,
+      "loss": 0.6265,
+      "step": 4208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.060853002784557e-06,
+      "loss": 0.5202,
+      "step": 4209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0586461267458073e-06,
+      "loss": 0.5308,
+      "step": 4210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0564401266071125e-06,
+      "loss": 0.5922,
+      "step": 4211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0542350030253934e-06,
+      "loss": 0.5826,
+      "step": 4212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0520307566573088e-06,
+      "loss": 0.5915,
+      "step": 4213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0498273881592546e-06,
+      "loss": 0.6265,
+      "step": 4214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.047624898187371e-06,
+      "loss": 0.5885,
+      "step": 4215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0454232873975297e-06,
+      "loss": 0.5077,
+      "step": 4216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.043222556445349e-06,
+      "loss": 0.5076,
+      "step": 4217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0410227059861747e-06,
+      "loss": 0.5998,
+      "step": 4218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0388237366751005e-06,
+      "loss": 0.6343,
+      "step": 4219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0366256491669505e-06,
+      "loss": 0.5381,
+      "step": 4220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.034428444116291e-06,
+      "loss": 0.6095,
+      "step": 4221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0322321221774204e-06,
+      "loss": 0.5539,
+      "step": 4222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.030036684004381e-06,
+      "loss": 0.556,
+      "step": 4223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0278421302509446e-06,
+      "loss": 0.5528,
+      "step": 4224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.025648461570626e-06,
+      "loss": 0.5616,
+      "step": 4225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0234556786166715e-06,
+      "loss": 0.5674,
+      "step": 4226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021263782042065e-06,
+      "loss": 0.602,
+      "step": 4227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0190727724995286e-06,
+      "loss": 0.4755,
+      "step": 4228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0168826506415183e-06,
+      "loss": 0.5162,
+      "step": 4229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0146934171202305e-06,
+      "loss": 0.5715,
+      "step": 4230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0125050725875846e-06,
+      "loss": 0.2503,
+      "step": 4231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0103176176952475e-06,
+      "loss": 0.2466,
+      "step": 4232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0081310530946197e-06,
+      "loss": 0.5378,
+      "step": 4233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0059453794368315e-06,
+      "loss": 0.2214,
+      "step": 4234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0037605973727486e-06,
+      "loss": 0.6074,
+      "step": 4235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0015767075529752e-06,
+      "loss": 0.5694,
+      "step": 4236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999393710627849e-06,
+      "loss": 0.5789,
+      "step": 4237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.997211607247437e-06,
+      "loss": 0.5696,
+      "step": 4238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9950303980615466e-06,
+      "loss": 0.5366,
+      "step": 4239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9928500837197153e-06,
+      "loss": 0.5126,
+      "step": 4240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9906706648712117e-06,
+      "loss": 0.5889,
+      "step": 4241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988492142165043e-06,
+      "loss": 0.6328,
+      "step": 4242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9863145162499485e-06,
+      "loss": 0.5396,
+      "step": 4243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9841377877743983e-06,
+      "loss": 0.5006,
+      "step": 4244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9819619573865932e-06,
+      "loss": 0.5396,
+      "step": 4245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979787025734473e-06,
+      "loss": 0.513,
+      "step": 4246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9776129934657067e-06,
+      "loss": 0.4555,
+      "step": 4247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9754398612276937e-06,
+      "loss": 0.5571,
+      "step": 4248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9732676296675653e-06,
+      "loss": 0.5443,
+      "step": 4249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9710962994321874e-06,
+      "loss": 0.5757,
+      "step": 4250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9689258711681587e-06,
+      "loss": 0.6105,
+      "step": 4251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.966756345521805e-06,
+      "loss": 0.572,
+      "step": 4252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9645877231391817e-06,
+      "loss": 0.6057,
+      "step": 4253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9624200046660845e-06,
+      "loss": 0.5188,
+      "step": 4254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.96025319074803e-06,
+      "loss": 0.5801,
+      "step": 4255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9580872820302714e-06,
+      "loss": 0.2636,
+      "step": 4256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9559222791577885e-06,
+      "loss": 0.5249,
+      "step": 4257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9537581827752973e-06,
+      "loss": 0.5207,
+      "step": 4258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9515949935272344e-06,
+      "loss": 0.2255,
+      "step": 4259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9494327120577765e-06,
+      "loss": 0.5897,
+      "step": 4260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9472713390108216e-06,
+      "loss": 0.5687,
+      "step": 4261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945110875030004e-06,
+      "loss": 0.6202,
+      "step": 4262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.942951320758681e-06,
+      "loss": 0.529,
+      "step": 4263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9407926768399456e-06,
+      "loss": 0.5057,
+      "step": 4264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9386349439166135e-06,
+      "loss": 0.6107,
+      "step": 4265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9364781226312306e-06,
+      "loss": 0.6754,
+      "step": 4266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.934322213626075e-06,
+      "loss": 0.2364,
+      "step": 4267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932167217543151e-06,
+      "loss": 0.5777,
+      "step": 4268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93001313502419e-06,
+      "loss": 0.2385,
+      "step": 4269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9278599667106505e-06,
+      "loss": 0.5938,
+      "step": 4270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9257077132437214e-06,
+      "loss": 0.5917,
+      "step": 4271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9235563752643195e-06,
+      "loss": 0.6157,
+      "step": 4272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9214059534130864e-06,
+      "loss": 0.5456,
+      "step": 4273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91925644833039e-06,
+      "loss": 0.575,
+      "step": 4274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91710786065633e-06,
+      "loss": 0.5903,
+      "step": 4275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9149601910307303e-06,
+      "loss": 0.5573,
+      "step": 4276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9128134400931404e-06,
+      "loss": 0.5227,
+      "step": 4277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910667608482835e-06,
+      "loss": 0.5312,
+      "step": 4278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9085226968388216e-06,
+      "loss": 0.6262,
+      "step": 4279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.906378705799824e-06,
+      "loss": 0.5801,
+      "step": 4280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9042356360043023e-06,
+      "loss": 0.5424,
+      "step": 4281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.902093488090433e-06,
+      "loss": 0.593,
+      "step": 4282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8999522626961254e-06,
+      "loss": 0.6149,
+      "step": 4283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8978119604590078e-06,
+      "loss": 0.6087,
+      "step": 4284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.895672582016438e-06,
+      "loss": 0.2251,
+      "step": 4285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8935341280055003e-06,
+      "loss": 0.5436,
+      "step": 4286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8913965990629952e-06,
+      "loss": 0.6055,
+      "step": 4287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.889259995825456e-06,
+      "loss": 0.6286,
+      "step": 4288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8871243189291378e-06,
+      "loss": 0.5097,
+      "step": 4289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8849895690100222e-06,
+      "loss": 0.5915,
+      "step": 4290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8828557467038073e-06,
+      "loss": 0.6262,
+      "step": 4291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8807228526459214e-06,
+      "loss": 0.6184,
+      "step": 4292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8785908874715176e-06,
+      "loss": 0.4878,
+      "step": 4293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8764598518154658e-06,
+      "loss": 0.5006,
+      "step": 4294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8743297463123666e-06,
+      "loss": 0.6087,
+      "step": 4295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8722005715965368e-06,
+      "loss": 0.2103,
+      "step": 4296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8700723283020222e-06,
+      "loss": 0.4901,
+      "step": 4297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8679450170625852e-06,
+      "loss": 0.6143,
+      "step": 4298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8658186385117166e-06,
+      "loss": 0.6374,
+      "step": 4299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.863693193282623e-06,
+      "loss": 0.5819,
+      "step": 4300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8615686820082412e-06,
+      "loss": 0.4875,
+      "step": 4301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.859445105321221e-06,
+      "loss": 0.5512,
+      "step": 4302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8573224638539412e-06,
+      "loss": 0.5381,
+      "step": 4303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8552007582384985e-06,
+      "loss": 0.5578,
+      "step": 4304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8530799891067086e-06,
+      "loss": 0.5509,
+      "step": 4305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8509601570901137e-06,
+      "loss": 0.5747,
+      "step": 4306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.848841262819976e-06,
+      "loss": 0.5939,
+      "step": 4307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8467233069272744e-06,
+      "loss": 0.5674,
+      "step": 4308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8446062900427108e-06,
+      "loss": 0.6032,
+      "step": 4309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8424902127967082e-06,
+      "loss": 0.6137,
+      "step": 4310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8403750758194105e-06,
+      "loss": 0.5523,
+      "step": 4311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8382608797406792e-06,
+      "loss": 0.4814,
+      "step": 4312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.836147625190095e-06,
+      "loss": 0.2033,
+      "step": 4313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8340353127969611e-06,
+      "loss": 0.5476,
+      "step": 4314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8319239431903007e-06,
+      "loss": 0.639,
+      "step": 4315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8298135169988534e-06,
+      "loss": 0.6253,
+      "step": 4316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8277040348510762e-06,
+      "loss": 0.2077,
+      "step": 4317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8255954973751515e-06,
+      "loss": 0.5462,
+      "step": 4318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8234879051989734e-06,
+      "loss": 0.5478,
+      "step": 4319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8213812589501611e-06,
+      "loss": 0.5412,
+      "step": 4320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8192755592560446e-06,
+      "loss": 0.5278,
+      "step": 4321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8171708067436804e-06,
+      "loss": 0.5297,
+      "step": 4322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8150670020398353e-06,
+      "loss": 0.544,
+      "step": 4323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8129641457709996e-06,
+      "loss": 0.5244,
+      "step": 4324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8108622385633778e-06,
+      "loss": 0.4817,
+      "step": 4325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8087612810428912e-06,
+      "loss": 0.564,
+      "step": 4326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8066612738351813e-06,
+      "loss": 0.5055,
+      "step": 4327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8045622175656064e-06,
+      "loss": 0.5922,
+      "step": 4328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8024641128592397e-06,
+      "loss": 0.644,
+      "step": 4329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800366960340869e-06,
+      "loss": 0.6024,
+      "step": 4330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7982707606350041e-06,
+      "loss": 0.5358,
+      "step": 4331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7961755143658689e-06,
+      "loss": 0.6202,
+      "step": 4332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7940812221574016e-06,
+      "loss": 0.567,
+      "step": 4333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.791987884633255e-06,
+      "loss": 0.5504,
+      "step": 4334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7898955024168029e-06,
+      "loss": 0.5817,
+      "step": 4335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7878040761311322e-06,
+      "loss": 0.5179,
+      "step": 4336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7857136063990437e-06,
+      "loss": 0.5643,
+      "step": 4337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7836240938430527e-06,
+      "loss": 0.63,
+      "step": 4338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7815355390853945e-06,
+      "loss": 0.6423,
+      "step": 4339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7794479427480115e-06,
+      "loss": 0.5889,
+      "step": 4340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7773613054525674e-06,
+      "loss": 0.2256,
+      "step": 4341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7752756278204397e-06,
+      "loss": 0.5577,
+      "step": 4342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7731909104727163e-06,
+      "loss": 0.5877,
+      "step": 4343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711071540301995e-06,
+      "loss": 0.6104,
+      "step": 4344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690243591134083e-06,
+      "loss": 0.5779,
+      "step": 4345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7669425263425766e-06,
+      "loss": 0.5658,
+      "step": 4346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7648616563376469e-06,
+      "loss": 0.5889,
+      "step": 4347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7627817497182764e-06,
+      "loss": 0.5621,
+      "step": 4348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7607028071038378e-06,
+      "loss": 0.2193,
+      "step": 4349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7586248291134179e-06,
+      "loss": 0.5247,
+      "step": 4350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7565478163658107e-06,
+      "loss": 0.5228,
+      "step": 4351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7544717694795243e-06,
+      "loss": 0.5646,
+      "step": 4352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7523966890727833e-06,
+      "loss": 0.5829,
+      "step": 4353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7503225757635227e-06,
+      "loss": 0.5312,
+      "step": 4354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7482494301693875e-06,
+      "loss": 0.6,
+      "step": 4355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7461772529077343e-06,
+      "loss": 0.5504,
+      "step": 4356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.744106044595635e-06,
+      "loss": 0.6226,
+      "step": 4357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7420358058498682e-06,
+      "loss": 0.6404,
+      "step": 4358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739966537286929e-06,
+      "loss": 0.5514,
+      "step": 4359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7378982395230183e-06,
+      "loss": 0.55,
+      "step": 4360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.735830913174053e-06,
+      "loss": 0.5614,
+      "step": 4361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7337645588556557e-06,
+      "loss": 0.5425,
+      "step": 4362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.731699177183165e-06,
+      "loss": 0.6136,
+      "step": 4363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7296347687716248e-06,
+      "loss": 0.2196,
+      "step": 4364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7275713342357902e-06,
+      "loss": 0.5344,
+      "step": 4365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7255088741901289e-06,
+      "loss": 0.5792,
+      "step": 4366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.723447389248818e-06,
+      "loss": 0.5174,
+      "step": 4367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7213868800257422e-06,
+      "loss": 0.5968,
+      "step": 4368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7193273471344946e-06,
+      "loss": 0.6034,
+      "step": 4369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7172687911883806e-06,
+      "loss": 0.6216,
+      "step": 4370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7152112128004162e-06,
+      "loss": 0.5076,
+      "step": 4371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7131546125833216e-06,
+      "loss": 0.5553,
+      "step": 4372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7110989911495257e-06,
+      "loss": 0.582,
+      "step": 4373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7090443491111702e-06,
+      "loss": 0.6273,
+      "step": 4374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7069906870801045e-06,
+      "loss": 0.4887,
+      "step": 4375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7049380056678832e-06,
+      "loss": 0.5785,
+      "step": 4376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7028863054857681e-06,
+      "loss": 0.5665,
+      "step": 4377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7008355871447345e-06,
+      "loss": 0.5829,
+      "step": 4378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.698785851255459e-06,
+      "loss": 0.5548,
+      "step": 4379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6967370984283311e-06,
+      "loss": 0.5713,
+      "step": 4380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6946893292734417e-06,
+      "loss": 0.5283,
+      "step": 4381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6926425444005956e-06,
+      "loss": 0.5518,
+      "step": 4382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6905967444192978e-06,
+      "loss": 0.4674,
+      "step": 4383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.688551929938766e-06,
+      "loss": 0.5974,
+      "step": 4384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686508101567918e-06,
+      "loss": 0.5155,
+      "step": 4385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6844652599153855e-06,
+      "loss": 0.4895,
+      "step": 4386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6824234055894984e-06,
+      "loss": 0.6106,
+      "step": 4387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6803825391982986e-06,
+      "loss": 0.6215,
+      "step": 4388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6783426613495347e-06,
+      "loss": 0.5629,
+      "step": 4389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.676303772650652e-06,
+      "loss": 0.5569,
+      "step": 4390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6742658737088096e-06,
+      "loss": 0.4856,
+      "step": 4391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6722289651308698e-06,
+      "loss": 0.5931,
+      "step": 4392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6701930475234035e-06,
+      "loss": 0.5747,
+      "step": 4393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668158121492675e-06,
+      "loss": 0.2238,
+      "step": 4394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6661241876446655e-06,
+      "loss": 0.6093,
+      "step": 4395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6640912465850572e-06,
+      "loss": 0.5645,
+      "step": 4396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6620592989192318e-06,
+      "loss": 0.6131,
+      "step": 4397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6600283452522837e-06,
+      "loss": 0.5231,
+      "step": 4398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657998386189002e-06,
+      "loss": 0.5162,
+      "step": 4399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6559694223338885e-06,
+      "loss": 0.537,
+      "step": 4400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6539414542911402e-06,
+      "loss": 0.5601,
+      "step": 4401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.651914482664666e-06,
+      "loss": 0.5002,
+      "step": 4402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6498885080580718e-06,
+      "loss": 0.5318,
+      "step": 4403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.647863531074667e-06,
+      "loss": 0.5836,
+      "step": 4404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6458395523174674e-06,
+      "loss": 0.5124,
+      "step": 4405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6438165723891907e-06,
+      "loss": 0.5274,
+      "step": 4406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6417945918922557e-06,
+      "loss": 0.5897,
+      "step": 4407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.639773611428781e-06,
+      "loss": 0.5938,
+      "step": 4408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6377536316005932e-06,
+      "loss": 0.5501,
+      "step": 4409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6357346530092188e-06,
+      "loss": 0.6213,
+      "step": 4410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.633716676255885e-06,
+      "loss": 0.6042,
+      "step": 4411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6316997019415187e-06,
+      "loss": 0.5923,
+      "step": 4412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6296837306667529e-06,
+      "loss": 0.5596,
+      "step": 4413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.627668763031921e-06,
+      "loss": 0.5556,
+      "step": 4414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6256547996370547e-06,
+      "loss": 0.544,
+      "step": 4415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6236418410818872e-06,
+      "loss": 0.5449,
+      "step": 4416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.621629887965856e-06,
+      "loss": 0.5795,
+      "step": 4417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6196189408880935e-06,
+      "loss": 0.5953,
+      "step": 4418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6176090004474399e-06,
+      "loss": 0.6132,
+      "step": 4419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6156000672424271e-06,
+      "loss": 0.5899,
+      "step": 4420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6135921418712959e-06,
+      "loss": 0.5074,
+      "step": 4421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6115852249319774e-06,
+      "loss": 0.5318,
+      "step": 4422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6095793170221125e-06,
+      "loss": 0.5816,
+      "step": 4423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6075744187390325e-06,
+      "loss": 0.6403,
+      "step": 4424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6055705306797752e-06,
+      "loss": 0.5787,
+      "step": 4425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6035676534410728e-06,
+      "loss": 0.5776,
+      "step": 4426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.601565787619359e-06,
+      "loss": 0.5104,
+      "step": 4427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.599564933810766e-06,
+      "loss": 0.5338,
+      "step": 4428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.597565092611122e-06,
+      "loss": 0.6425,
+      "step": 4429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5955662646159569e-06,
+      "loss": 0.2138,
+      "step": 4430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5935684504205002e-06,
+      "loss": 0.207,
+      "step": 4431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5915716506196743e-06,
+      "loss": 0.5414,
+      "step": 4432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5895758658081023e-06,
+      "loss": 0.5957,
+      "step": 4433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5875810965801052e-06,
+      "loss": 0.5608,
+      "step": 4434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5855873435297042e-06,
+      "loss": 0.5767,
+      "step": 4435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5835946072506131e-06,
+      "loss": 0.5905,
+      "step": 4436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5816028883362428e-06,
+      "loss": 0.575,
+      "step": 4437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.579612187379706e-06,
+      "loss": 0.5721,
+      "step": 4438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577622504973811e-06,
+      "loss": 0.5597,
+      "step": 4439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5756338417110594e-06,
+      "loss": 0.5765,
+      "step": 4440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5736461981836498e-06,
+      "loss": 0.6255,
+      "step": 4441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5716595749834828e-06,
+      "loss": 0.6303,
+      "step": 4442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5696739727021465e-06,
+      "loss": 0.5038,
+      "step": 4443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5676893919309322e-06,
+      "loss": 0.5815,
+      "step": 4444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5657058332608256e-06,
+      "loss": 0.5918,
+      "step": 4445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563723297282505e-06,
+      "loss": 0.585,
+      "step": 4446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5617417845863452e-06,
+      "loss": 0.5382,
+      "step": 4447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5597612957624175e-06,
+      "loss": 0.544,
+      "step": 4448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5577818314004911e-06,
+      "loss": 0.2235,
+      "step": 4449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5558033920900212e-06,
+      "loss": 0.5254,
+      "step": 4450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5538259784201659e-06,
+      "loss": 0.4837,
+      "step": 4451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.551849590979776e-06,
+      "loss": 0.2186,
+      "step": 4452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5498742303573972e-06,
+      "loss": 0.5102,
+      "step": 4453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5478998971412669e-06,
+      "loss": 0.5887,
+      "step": 4454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545926591919316e-06,
+      "loss": 0.519,
+      "step": 4455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5439543152791752e-06,
+      "loss": 0.6484,
+      "step": 4456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5419830678081614e-06,
+      "loss": 0.6366,
+      "step": 4457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5400128500932915e-06,
+      "loss": 0.651,
+      "step": 4458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5380436627212692e-06,
+      "loss": 0.613,
+      "step": 4459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5360755062784992e-06,
+      "loss": 0.6211,
+      "step": 4460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.534108381351071e-06,
+      "loss": 0.557,
+      "step": 4461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5321422885247744e-06,
+      "loss": 0.5778,
+      "step": 4462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5301772283850853e-06,
+      "loss": 0.5713,
+      "step": 4463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.528213201517178e-06,
+      "loss": 0.5796,
+      "step": 4464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5262502085059134e-06,
+      "loss": 0.6456,
+      "step": 4465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5242882499358507e-06,
+      "loss": 0.5693,
+      "step": 4466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5223273263912352e-06,
+      "loss": 0.5763,
+      "step": 4467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5203674384560062e-06,
+      "loss": 0.2342,
+      "step": 4468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5184085867137955e-06,
+      "loss": 0.5839,
+      "step": 4469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5164507717479276e-06,
+      "loss": 0.6354,
+      "step": 4470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.514493994141415e-06,
+      "loss": 0.5812,
+      "step": 4471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5125382544769613e-06,
+      "loss": 0.5845,
+      "step": 4472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.510583553336964e-06,
+      "loss": 0.5708,
+      "step": 4473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5086298913035103e-06,
+      "loss": 0.5952,
+      "step": 4474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.506677268958377e-06,
+      "loss": 0.5968,
+      "step": 4475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5047256868830301e-06,
+      "loss": 0.5661,
+      "step": 4476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5027751456586282e-06,
+      "loss": 0.5622,
+      "step": 4477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5008256458660214e-06,
+      "loss": 0.2324,
+      "step": 4478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4988771880857462e-06,
+      "loss": 0.5229,
+      "step": 4479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4969297728980275e-06,
+      "loss": 0.5399,
+      "step": 4480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4949834008827858e-06,
+      "loss": 0.5203,
+      "step": 4481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4930380726196247e-06,
+      "loss": 0.5222,
+      "step": 4482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4910937886878423e-06,
+      "loss": 0.6018,
+      "step": 4483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4891505496664205e-06,
+      "loss": 0.6213,
+      "step": 4484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4872083561340355e-06,
+      "loss": 0.6497,
+      "step": 4485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4852672086690462e-06,
+      "loss": 0.5444,
+      "step": 4486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4833271078495065e-06,
+      "loss": 0.5369,
+      "step": 4487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.481388054253154e-06,
+      "loss": 0.6109,
+      "step": 4488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.479450048457413e-06,
+      "loss": 0.6692,
+      "step": 4489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4775130910394008e-06,
+      "loss": 0.5626,
+      "step": 4490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4755771825759207e-06,
+      "loss": 0.5878,
+      "step": 4491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.473642323643465e-06,
+      "loss": 0.511,
+      "step": 4492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4717085148182065e-06,
+      "loss": 0.563,
+      "step": 4493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4697757566760124e-06,
+      "loss": 0.5613,
+      "step": 4494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4678440497924374e-06,
+      "loss": 0.5965,
+      "step": 4495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.465913394742719e-06,
+      "loss": 0.4901,
+      "step": 4496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.463983792101782e-06,
+      "loss": 0.6023,
+      "step": 4497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4620552424442391e-06,
+      "loss": 0.5395,
+      "step": 4498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.460127746344393e-06,
+      "loss": 0.6051,
+      "step": 4499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4582013043762244e-06,
+      "loss": 0.6241,
+      "step": 4500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4562759171134083e-06,
+      "loss": 0.5839,
+      "step": 4501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4543515851292988e-06,
+      "loss": 0.6062,
+      "step": 4502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4524283089969426e-06,
+      "loss": 0.5396,
+      "step": 4503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4505060892890643e-06,
+      "loss": 0.6293,
+      "step": 4504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4485849265780821e-06,
+      "loss": 0.5811,
+      "step": 4505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4466648214360928e-06,
+      "loss": 0.561,
+      "step": 4506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4447457744348802e-06,
+      "loss": 0.6685,
+      "step": 4507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4428277861459145e-06,
+      "loss": 0.6499,
+      "step": 4508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4409108571403518e-06,
+      "loss": 0.5632,
+      "step": 4509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.438994987989028e-06,
+      "loss": 0.5476,
+      "step": 4510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4370801792624656e-06,
+      "loss": 0.5278,
+      "step": 4511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4351664315308727e-06,
+      "loss": 0.5891,
+      "step": 4512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4332537453641426e-06,
+      "loss": 0.5597,
+      "step": 4513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4313421213318484e-06,
+      "loss": 0.5264,
+      "step": 4514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4294315600032477e-06,
+      "loss": 0.5743,
+      "step": 4515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4275220619472841e-06,
+      "loss": 0.5754,
+      "step": 4516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.425613627732585e-06,
+      "loss": 0.5553,
+      "step": 4517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4237062579274574e-06,
+      "loss": 0.6064,
+      "step": 4518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4217999530998927e-06,
+      "loss": 0.6133,
+      "step": 4519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198947138175678e-06,
+      "loss": 0.5817,
+      "step": 4520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4179905406478373e-06,
+      "loss": 0.5416,
+      "step": 4521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4160874341577447e-06,
+      "loss": 0.567,
+      "step": 4522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4141853949140088e-06,
+      "loss": 0.6298,
+      "step": 4523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4122844234830374e-06,
+      "loss": 0.5775,
+      "step": 4524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4103845204309147e-06,
+      "loss": 0.5136,
+      "step": 4525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4084856863234115e-06,
+      "loss": 0.5322,
+      "step": 4526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.406587921725976e-06,
+      "loss": 0.6015,
+      "step": 4527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4046912272037395e-06,
+      "loss": 0.5231,
+      "step": 4528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4027956033215157e-06,
+      "loss": 0.5509,
+      "step": 4529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4009010506437997e-06,
+      "loss": 0.4997,
+      "step": 4530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3990075697347667e-06,
+      "loss": 0.5609,
+      "step": 4531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.39711516115827e-06,
+      "loss": 0.567,
+      "step": 4532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3952238254778483e-06,
+      "loss": 0.6127,
+      "step": 4533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3933335632567195e-06,
+      "loss": 0.6233,
+      "step": 4534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.391444375057781e-06,
+      "loss": 0.6016,
+      "step": 4535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3895562614436075e-06,
+      "loss": 0.5649,
+      "step": 4536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3876692229764594e-06,
+      "loss": 0.6249,
+      "step": 4537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3857832602182753e-06,
+      "loss": 0.611,
+      "step": 4538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3838983737306704e-06,
+      "loss": 0.6024,
+      "step": 4539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.38201456407494e-06,
+      "loss": 0.5777,
+      "step": 4540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3801318318120622e-06,
+      "loss": 0.4913,
+      "step": 4541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3782501775026935e-06,
+      "loss": 0.6073,
+      "step": 4542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3763696017071664e-06,
+      "loss": 0.5641,
+      "step": 4543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3744901049854926e-06,
+      "loss": 0.5839,
+      "step": 4544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3726116878973661e-06,
+      "loss": 0.6597,
+      "step": 4545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3707343510021547e-06,
+      "loss": 0.5838,
+      "step": 4546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3688580948589087e-06,
+      "loss": 0.5059,
+      "step": 4547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3669829200263557e-06,
+      "loss": 0.5634,
+      "step": 4548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3651088270628992e-06,
+      "loss": 0.6001,
+      "step": 4549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3632358165266201e-06,
+      "loss": 0.6279,
+      "step": 4550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3613638889752795e-06,
+      "loss": 0.564,
+      "step": 4551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3594930449663192e-06,
+      "loss": 0.5481,
+      "step": 4552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3576232850568471e-06,
+      "loss": 0.5371,
+      "step": 4553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3557546098036596e-06,
+      "loss": 0.5535,
+      "step": 4554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3538870197632242e-06,
+      "loss": 0.6322,
+      "step": 4555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3520205154916898e-06,
+      "loss": 0.5064,
+      "step": 4556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3501550975448769e-06,
+      "loss": 0.5793,
+      "step": 4557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3482907664782835e-06,
+      "loss": 0.5325,
+      "step": 4558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3464275228470885e-06,
+      "loss": 0.5333,
+      "step": 4559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.34456536720614e-06,
+      "loss": 0.5534,
+      "step": 4560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3427043001099682e-06,
+      "loss": 0.6016,
+      "step": 4561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3408443221127748e-06,
+      "loss": 0.5811,
+      "step": 4562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3389854337684416e-06,
+      "loss": 0.6059,
+      "step": 4563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3371276356305195e-06,
+      "loss": 0.6712,
+      "step": 4564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.335270928252243e-06,
+      "loss": 0.539,
+      "step": 4565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3334153121865157e-06,
+      "loss": 0.6353,
+      "step": 4566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3315607879859155e-06,
+      "loss": 0.6162,
+      "step": 4567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3297073562026992e-06,
+      "loss": 0.5063,
+      "step": 4568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.327855017388799e-06,
+      "loss": 0.5452,
+      "step": 4569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3260037720958168e-06,
+      "loss": 0.2188,
+      "step": 4570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3241536208750306e-06,
+      "loss": 0.633,
+      "step": 4571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.322304564277394e-06,
+      "loss": 0.5144,
+      "step": 4572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3204566028535365e-06,
+      "loss": 0.5831,
+      "step": 4573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.318609737153757e-06,
+      "loss": 0.6391,
+      "step": 4574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.316763967728028e-06,
+      "loss": 0.5906,
+      "step": 4575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3149192951260003e-06,
+      "loss": 0.5685,
+      "step": 4576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3130757198969957e-06,
+      "loss": 0.5663,
+      "step": 4577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3112332425900077e-06,
+      "loss": 0.5652,
+      "step": 4578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3093918637537028e-06,
+      "loss": 0.5213,
+      "step": 4579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3075515839364229e-06,
+      "loss": 0.5964,
+      "step": 4580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.305712403686183e-06,
+      "loss": 0.5402,
+      "step": 4581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3038743235506674e-06,
+      "loss": 0.5792,
+      "step": 4582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3020373440772332e-06,
+      "loss": 0.5487,
+      "step": 4583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3002014658129132e-06,
+      "loss": 0.2027,
+      "step": 4584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2983666893044079e-06,
+      "loss": 0.5887,
+      "step": 4585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2965330150980943e-06,
+      "loss": 0.5971,
+      "step": 4586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2947004437400161e-06,
+      "loss": 0.5609,
+      "step": 4587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2928689757758938e-06,
+      "loss": 0.5778,
+      "step": 4588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2910386117511137e-06,
+      "loss": 0.5662,
+      "step": 4589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2892093522107396e-06,
+      "loss": 0.5442,
+      "step": 4590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.287381197699501e-06,
+      "loss": 0.5959,
+      "step": 4591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2855541487618e-06,
+      "loss": 0.6128,
+      "step": 4592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2837282059417105e-06,
+      "loss": 0.5704,
+      "step": 4593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.281903369782977e-06,
+      "loss": 0.2146,
+      "step": 4594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2800796408290173e-06,
+      "loss": 0.5706,
+      "step": 4595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2782570196229088e-06,
+      "loss": 0.5995,
+      "step": 4596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2764355067074102e-06,
+      "loss": 0.5382,
+      "step": 4597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274615102624948e-06,
+      "loss": 0.599,
+      "step": 4598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.272795807917615e-06,
+      "loss": 0.5601,
+      "step": 4599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2709776231271742e-06,
+      "loss": 0.5751,
+      "step": 4600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2691605487950604e-06,
+      "loss": 0.5944,
+      "step": 4601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.267344585462378e-06,
+      "loss": 0.6192,
+      "step": 4602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2655297336698963e-06,
+      "loss": 0.593,
+      "step": 4603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.26371599395806e-06,
+      "loss": 0.5788,
+      "step": 4604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2619033668669773e-06,
+      "loss": 0.5874,
+      "step": 4605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2600918529364253e-06,
+      "loss": 0.6813,
+      "step": 4606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2582814527058528e-06,
+      "loss": 0.672,
+      "step": 4607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.256472166714377e-06,
+      "loss": 0.6243,
+      "step": 4608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.25466399550078e-06,
+      "loss": 0.6892,
+      "step": 4609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2528569396035123e-06,
+      "loss": 0.5485,
+      "step": 4610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2510509995606945e-06,
+      "loss": 0.5836,
+      "step": 4611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.249246175910116e-06,
+      "loss": 0.579,
+      "step": 4612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24744246918923e-06,
+      "loss": 0.6149,
+      "step": 4613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2456398799351577e-06,
+      "loss": 0.6453,
+      "step": 4614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2438384086846894e-06,
+      "loss": 0.5964,
+      "step": 4615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.242038055974284e-06,
+      "loss": 0.6018,
+      "step": 4616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2402388223400636e-06,
+      "loss": 0.5397,
+      "step": 4617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2384407083178158e-06,
+      "loss": 0.6472,
+      "step": 4618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2366437144430005e-06,
+      "loss": 0.4906,
+      "step": 4619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2348478412507415e-06,
+      "loss": 0.6239,
+      "step": 4620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2330530892758268e-06,
+      "loss": 0.6443,
+      "step": 4621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2312594590527117e-06,
+      "loss": 0.5384,
+      "step": 4622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2294669511155193e-06,
+      "loss": 0.5622,
+      "step": 4623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2276755659980355e-06,
+      "loss": 0.6821,
+      "step": 4624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.225885304233716e-06,
+      "loss": 0.5854,
+      "step": 4625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2240961663556756e-06,
+      "loss": 0.5441,
+      "step": 4626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2223081528967018e-06,
+      "loss": 0.5901,
+      "step": 4627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2205212643892411e-06,
+      "loss": 0.5927,
+      "step": 4628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2187355013654096e-06,
+      "loss": 0.698,
+      "step": 4629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2169508643569854e-06,
+      "loss": 0.5574,
+      "step": 4630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2151673538954107e-06,
+      "loss": 0.2424,
+      "step": 4631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.213384970511794e-06,
+      "loss": 0.5045,
+      "step": 4632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2116037147369103e-06,
+      "loss": 0.5073,
+      "step": 4633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2098235871011948e-06,
+      "loss": 0.552,
+      "step": 4634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2080445881347463e-06,
+      "loss": 0.2361,
+      "step": 4635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2062667183673315e-06,
+      "loss": 0.5852,
+      "step": 4636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2044899783283798e-06,
+      "loss": 0.3042,
+      "step": 4637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2027143685469817e-06,
+      "loss": 0.2352,
+      "step": 4638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2009398895518909e-06,
+      "loss": 0.6282,
+      "step": 4639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1991665418715282e-06,
+      "loss": 0.5456,
+      "step": 4640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1973943260339765e-06,
+      "loss": 0.5433,
+      "step": 4641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.195623242566979e-06,
+      "loss": 0.6183,
+      "step": 4642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1938532919979412e-06,
+      "loss": 0.5893,
+      "step": 4643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1920844748539373e-06,
+      "loss": 0.5278,
+      "step": 4644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1903167916616959e-06,
+      "loss": 0.5794,
+      "step": 4645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1885502429476158e-06,
+      "loss": 0.6401,
+      "step": 4646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1867848292377509e-06,
+      "loss": 0.6413,
+      "step": 4647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1850205510578227e-06,
+      "loss": 0.225,
+      "step": 4648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.18325740893321e-06,
+      "loss": 0.229,
+      "step": 4649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1814954033889558e-06,
+      "loss": 0.5551,
+      "step": 4650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1797345349497685e-06,
+      "loss": 0.6108,
+      "step": 4651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1779748041400074e-06,
+      "loss": 0.647,
+      "step": 4652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1762162114837023e-06,
+      "loss": 0.2642,
+      "step": 4653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1744587575045413e-06,
+      "loss": 0.5944,
+      "step": 4654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1727024427258748e-06,
+      "loss": 0.5685,
+      "step": 4655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1709472676707083e-06,
+      "loss": 0.6496,
+      "step": 4656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.169193232861714e-06,
+      "loss": 0.5652,
+      "step": 4657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1674403388212224e-06,
+      "loss": 0.6583,
+      "step": 4658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1656885860712264e-06,
+      "loss": 0.5497,
+      "step": 4659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.163937975133375e-06,
+      "loss": 0.5469,
+      "step": 4660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1621885065289783e-06,
+      "loss": 0.5424,
+      "step": 4661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.16044018077901e-06,
+      "loss": 0.6237,
+      "step": 4662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1586929984040974e-06,
+      "loss": 0.6313,
+      "step": 4663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1569469599245332e-06,
+      "loss": 0.6465,
+      "step": 4664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1552020658602648e-06,
+      "loss": 0.59,
+      "step": 4665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1534583167309027e-06,
+      "loss": 0.2415,
+      "step": 4666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1517157130557128e-06,
+      "loss": 0.5442,
+      "step": 4667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1499742553536237e-06,
+      "loss": 0.5504,
+      "step": 4668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1482339441432195e-06,
+      "loss": 0.6134,
+      "step": 4669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1464947799427423e-06,
+      "loss": 0.5228,
+      "step": 4670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1447567632700968e-06,
+      "loss": 0.5968,
+      "step": 4671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430198946428439e-06,
+      "loss": 0.6366,
+      "step": 4672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1412841745782016e-06,
+      "loss": 0.5499,
+      "step": 4673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1395496035930454e-06,
+      "loss": 0.6479,
+      "step": 4674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1378161822039107e-06,
+      "loss": 0.5121,
+      "step": 4675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136083910926991e-06,
+      "loss": 0.603,
+      "step": 4676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1343527902781353e-06,
+      "loss": 0.5434,
+      "step": 4677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1326228207728485e-06,
+      "loss": 0.5661,
+      "step": 4678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1308940029262971e-06,
+      "loss": 0.6018,
+      "step": 4679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.129166337253303e-06,
+      "loss": 0.5794,
+      "step": 4680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1274398242683433e-06,
+      "loss": 0.6459,
+      "step": 4681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.125714464485551e-06,
+      "loss": 0.5737,
+      "step": 4682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1239902584187218e-06,
+      "loss": 0.584,
+      "step": 4683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1222672065812995e-06,
+      "loss": 0.5717,
+      "step": 4684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1205453094863922e-06,
+      "loss": 0.546,
+      "step": 4685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1188245676467563e-06,
+      "loss": 0.6368,
+      "step": 4686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1171049815748125e-06,
+      "loss": 0.5629,
+      "step": 4687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1153865517826283e-06,
+      "loss": 0.545,
+      "step": 4688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1136692787819365e-06,
+      "loss": 0.6058,
+      "step": 4689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1119531630841173e-06,
+      "loss": 0.5862,
+      "step": 4690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1102382052002091e-06,
+      "loss": 0.5183,
+      "step": 4691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1085244056409072e-06,
+      "loss": 0.5869,
+      "step": 4692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1068117649165617e-06,
+      "loss": 0.5702,
+      "step": 4693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1051002835371748e-06,
+      "loss": 0.5717,
+      "step": 4694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.103389962012405e-06,
+      "loss": 0.6168,
+      "step": 4695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1016808008515656e-06,
+      "loss": 0.5761,
+      "step": 4696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0999728005636251e-06,
+      "loss": 0.5064,
+      "step": 4697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0982659616572089e-06,
+      "loss": 0.6441,
+      "step": 4698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0965602846405866e-06,
+      "loss": 0.6155,
+      "step": 4699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0948557700216923e-06,
+      "loss": 0.6456,
+      "step": 4700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0931524183081105e-06,
+      "loss": 0.5527,
+      "step": 4701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0914502300070777e-06,
+      "loss": 0.6193,
+      "step": 4702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.089749205625485e-06,
+      "loss": 0.5913,
+      "step": 4703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088049345669877e-06,
+      "loss": 0.5416,
+      "step": 4704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0863506506464539e-06,
+      "loss": 0.6002,
+      "step": 4705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0846531210610644e-06,
+      "loss": 0.5431,
+      "step": 4706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0829567574192146e-06,
+      "loss": 0.6008,
+      "step": 4707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0812615602260602e-06,
+      "loss": 0.5346,
+      "step": 4708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795675299864089e-06,
+      "loss": 0.5613,
+      "step": 4709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0778746672047247e-06,
+      "loss": 0.6507,
+      "step": 4710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.076182972385123e-06,
+      "loss": 0.6164,
+      "step": 4711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0744924460313694e-06,
+      "loss": 0.5775,
+      "step": 4712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0728030886468798e-06,
+      "loss": 0.5895,
+      "step": 4713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0711149007347277e-06,
+      "loss": 0.5509,
+      "step": 4714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.069427882797635e-06,
+      "loss": 0.6201,
+      "step": 4715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0677420353379758e-06,
+      "loss": 0.2263,
+      "step": 4716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0660573588577726e-06,
+      "loss": 0.6163,
+      "step": 4717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.064373853858704e-06,
+      "loss": 0.5874,
+      "step": 4718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.062691520842099e-06,
+      "loss": 0.6308,
+      "step": 4719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0610103603089345e-06,
+      "loss": 0.5593,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0593303727598392e-06,
+      "loss": 0.6148,
+      "step": 4721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0576515586950959e-06,
+      "loss": 0.5209,
+      "step": 4722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0559739186146323e-06,
+      "loss": 0.245,
+      "step": 4723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0542974530180327e-06,
+      "loss": 0.5822,
+      "step": 4724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0526221624045258e-06,
+      "loss": 0.5562,
+      "step": 4725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0509480472729954e-06,
+      "loss": 0.5932,
+      "step": 4726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.049275108121971e-06,
+      "loss": 0.5735,
+      "step": 4727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0476033454496359e-06,
+      "loss": 0.561,
+      "step": 4728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0459327597538199e-06,
+      "loss": 0.5672,
+      "step": 4729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0442633515320028e-06,
+      "loss": 0.5083,
+      "step": 4730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.042595121281315e-06,
+      "loss": 0.5526,
+      "step": 4731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0409280694985368e-06,
+      "loss": 0.576,
+      "step": 4732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.039262196680096e-06,
+      "loss": 0.4871,
+      "step": 4733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.037597503322067e-06,
+      "loss": 0.6007,
+      "step": 4734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0359339899201775e-06,
+      "loss": 0.5446,
+      "step": 4735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034271656969803e-06,
+      "loss": 0.489,
+      "step": 4736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0326105049659662e-06,
+      "loss": 0.5661,
+      "step": 4737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0309505344033356e-06,
+      "loss": 0.6096,
+      "step": 4738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0292917457762325e-06,
+      "loss": 0.5574,
+      "step": 4739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0276341395786255e-06,
+      "loss": 0.6022,
+      "step": 4740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025977716304129e-06,
+      "loss": 0.5687,
+      "step": 4741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0243224764460042e-06,
+      "loss": 0.5936,
+      "step": 4742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0226684204971638e-06,
+      "loss": 0.5229,
+      "step": 4743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0210155489501661e-06,
+      "loss": 0.6236,
+      "step": 4744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0193638622972164e-06,
+      "loss": 0.6129,
+      "step": 4745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0177133610301655e-06,
+      "loss": 0.5624,
+      "step": 4746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0160640456405151e-06,
+      "loss": 0.2466,
+      "step": 4747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0144159166194101e-06,
+      "loss": 0.6258,
+      "step": 4748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0127689744576447e-06,
+      "loss": 0.5604,
+      "step": 4749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0111232196456572e-06,
+      "loss": 0.5924,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0094786526735361e-06,
+      "loss": 0.5893,
+      "step": 4751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0078352740310115e-06,
+      "loss": 0.5616,
+      "step": 4752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0061930842074635e-06,
+      "loss": 0.6233,
+      "step": 4753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0045520836919165e-06,
+      "loss": 0.5303,
+      "step": 4754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0029122729730384e-06,
+      "loss": 0.5456,
+      "step": 4755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0012736525391475e-06,
+      "loss": 0.5153,
+      "step": 4756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.996362228782046e-07,
+      "loss": 0.5823,
+      "step": 4757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.979999844778203e-07,
+      "loss": 0.5087,
+      "step": 4758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.96364937825241e-07,
+      "loss": 0.5498,
+      "step": 4759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.94731083407367e-07,
+      "loss": 0.4965,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.930984217107415e-07,
+      "loss": 0.2422,
+      "step": 4761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.914669532215487e-07,
+      "loss": 0.5755,
+      "step": 4762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.898366784256242e-07,
+      "loss": 0.551,
+      "step": 4763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.88207597808441e-07,
+      "loss": 0.6184,
+      "step": 4764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865797118551223e-07,
+      "loss": 0.5535,
+      "step": 4765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.849530210504315e-07,
+      "loss": 0.6429,
+      "step": 4766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.833275258787789e-07,
+      "loss": 0.5846,
+      "step": 4767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.817032268242172e-07,
+      "loss": 0.5178,
+      "step": 4768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800801243704417e-07,
+      "loss": 0.5804,
+      "step": 4769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78458219000794e-07,
+      "loss": 0.6154,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.768375111982603e-07,
+      "loss": 0.6361,
+      "step": 4771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.752180014454654e-07,
+      "loss": 0.599,
+      "step": 4772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.735996902246797e-07,
+      "loss": 0.5545,
+      "step": 4773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.719825780178176e-07,
+      "loss": 0.6151,
+      "step": 4774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703666653064376e-07,
+      "loss": 0.5718,
+      "step": 4775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.687519525717376e-07,
+      "loss": 0.5961,
+      "step": 4776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.671384402945588e-07,
+      "loss": 0.5989,
+      "step": 4777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65526128955387e-07,
+      "loss": 0.5958,
+      "step": 4778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639150190343505e-07,
+      "loss": 0.5895,
+      "step": 4779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62305111011217e-07,
+      "loss": 0.5983,
+      "step": 4780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.606964053653973e-07,
+      "loss": 0.6127,
+      "step": 4781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590889025759454e-07,
+      "loss": 0.5746,
+      "step": 4782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.574826031215577e-07,
+      "loss": 0.5009,
+      "step": 4783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.558775074805704e-07,
+      "loss": 0.5489,
+      "step": 4784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5427361613096e-07,
+      "loss": 0.5851,
+      "step": 4785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.526709295503489e-07,
+      "loss": 0.5799,
+      "step": 4786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.510694482159954e-07,
+      "loss": 0.5964,
+      "step": 4787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.494691726048038e-07,
+      "loss": 0.5948,
+      "step": 4788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.478701031933152e-07,
+      "loss": 0.6035,
+      "step": 4789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.462722404577152e-07,
+      "loss": 0.5458,
+      "step": 4790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.44675584873826e-07,
+      "loss": 0.5011,
+      "step": 4791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.430801369171149e-07,
+      "loss": 0.6651,
+      "step": 4792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.414858970626867e-07,
+      "loss": 0.6348,
+      "step": 4793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398928657852851e-07,
+      "loss": 0.5907,
+      "step": 4794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383010435592965e-07,
+      "loss": 0.589,
+      "step": 4795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.367104308587493e-07,
+      "loss": 0.5949,
+      "step": 4796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.351210281573064e-07,
+      "loss": 0.5734,
+      "step": 4797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.335328359282724e-07,
+      "loss": 0.5643,
+      "step": 4798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.319458546445925e-07,
+      "loss": 0.5665,
+      "step": 4799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.30360084778853e-07,
+      "loss": 0.6203,
+      "step": 4800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.287755268032755e-07,
+      "loss": 0.5073,
+      "step": 4801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.271921811897211e-07,
+      "loss": 0.5866,
+      "step": 4802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.256100484096925e-07,
+      "loss": 0.5678,
+      "step": 4803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.240291289343312e-07,
+      "loss": 0.5869,
+      "step": 4804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22449423234415e-07,
+      "loss": 0.5281,
+      "step": 4805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.208709317803599e-07,
+      "loss": 0.5882,
+      "step": 4806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.192936550422249e-07,
+      "loss": 0.5673,
+      "step": 4807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.177175934897009e-07,
+      "loss": 0.6069,
+      "step": 4808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.161427475921231e-07,
+      "loss": 0.5622,
+      "step": 4809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.14569117818459e-07,
+      "loss": 0.5429,
+      "step": 4810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.129967046373201e-07,
+      "loss": 0.5927,
+      "step": 4811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.114255085169482e-07,
+      "loss": 0.5265,
+      "step": 4812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.098555299252299e-07,
+      "loss": 0.5487,
+      "step": 4813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.08286769329686e-07,
+      "loss": 0.5855,
+      "step": 4814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.06719227197474e-07,
+      "loss": 0.5894,
+      "step": 4815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.051529039953877e-07,
+      "loss": 0.6264,
+      "step": 4816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.035878001898601e-07,
+      "loss": 0.5642,
+      "step": 4817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.020239162469635e-07,
+      "loss": 0.5388,
+      "step": 4818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.004612526324008e-07,
+      "loss": 0.6027,
+      "step": 4819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.988998098115137e-07,
+      "loss": 0.5736,
+      "step": 4820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.973395882492831e-07,
+      "loss": 0.5264,
+      "step": 4821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.957805884103255e-07,
+      "loss": 0.2369,
+      "step": 4822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942228107588907e-07,
+      "loss": 0.5341,
+      "step": 4823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.92666255758865e-07,
+      "loss": 0.5356,
+      "step": 4824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.911109238737748e-07,
+      "loss": 0.6403,
+      "step": 4825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.895568155667766e-07,
+      "loss": 0.515,
+      "step": 4826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.880039313006683e-07,
+      "loss": 0.5449,
+      "step": 4827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.86452271537877e-07,
+      "loss": 0.5329,
+      "step": 4828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.84901836740471e-07,
+      "loss": 0.5467,
+      "step": 4829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.833526273701498e-07,
+      "loss": 0.5659,
+      "step": 4830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.818046438882505e-07,
+      "loss": 0.245,
+      "step": 4831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.802578867557443e-07,
+      "loss": 0.2622,
+      "step": 4832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.787123564332345e-07,
+      "loss": 0.5903,
+      "step": 4833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.771680533809634e-07,
+      "loss": 0.5802,
+      "step": 4834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.756249780588077e-07,
+      "loss": 0.5968,
+      "step": 4835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.740831309262743e-07,
+      "loss": 0.5899,
+      "step": 4836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.725425124425069e-07,
+      "loss": 0.621,
+      "step": 4837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.710031230662841e-07,
+      "loss": 0.6124,
+      "step": 4838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.694649632560181e-07,
+      "loss": 0.5252,
+      "step": 4839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.679280334697549e-07,
+      "loss": 0.6074,
+      "step": 4840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.663923341651709e-07,
+      "loss": 0.5976,
+      "step": 4841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.648578657995815e-07,
+      "loss": 0.5434,
+      "step": 4842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.633246288299335e-07,
+      "loss": 0.4823,
+      "step": 4843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.61792623712806e-07,
+      "loss": 0.5426,
+      "step": 4844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.602618509044103e-07,
+      "loss": 0.5823,
+      "step": 4845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.587323108605944e-07,
+      "loss": 0.5625,
+      "step": 4846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.572040040368351e-07,
+      "loss": 0.612,
+      "step": 4847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.556769308882456e-07,
+      "loss": 0.5753,
+      "step": 4848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.541510918695683e-07,
+      "loss": 0.5946,
+      "step": 4849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.526264874351825e-07,
+      "loss": 0.5536,
+      "step": 4850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.511031180390944e-07,
+      "loss": 0.247,
+      "step": 4851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.495809841349473e-07,
+      "loss": 0.5658,
+      "step": 4852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.480600861760124e-07,
+      "loss": 0.6043,
+      "step": 4853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.465404246151976e-07,
+      "loss": 0.5024,
+      "step": 4854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.450219999050368e-07,
+      "loss": 0.6161,
+      "step": 4855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.435048124977019e-07,
+      "loss": 0.6289,
+      "step": 4856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.419888628449918e-07,
+      "loss": 0.5494,
+      "step": 4857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.404741513983372e-07,
+      "loss": 0.5902,
+      "step": 4858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.389606786088017e-07,
+      "loss": 0.5229,
+      "step": 4859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.374484449270798e-07,
+      "loss": 0.4832,
+      "step": 4860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35937450803499e-07,
+      "loss": 0.6345,
+      "step": 4861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.344276966880104e-07,
+      "loss": 0.5825,
+      "step": 4862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.32919183030203e-07,
+      "loss": 0.2353,
+      "step": 4863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.314119102792961e-07,
+      "loss": 0.6095,
+      "step": 4864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.299058788841358e-07,
+      "loss": 0.6527,
+      "step": 4865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.284010892931988e-07,
+      "loss": 0.6491,
+      "step": 4866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.268975419545943e-07,
+      "loss": 0.6033,
+      "step": 4867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.253952373160634e-07,
+      "loss": 0.5244,
+      "step": 4868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.238941758249708e-07,
+      "loss": 0.5552,
+      "step": 4869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.223943579283177e-07,
+      "loss": 0.5965,
+      "step": 4870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.208957840727305e-07,
+      "loss": 0.5599,
+      "step": 4871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.193984547044659e-07,
+      "loss": 0.5986,
+      "step": 4872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.179023702694111e-07,
+      "loss": 0.5323,
+      "step": 4873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.164075312130837e-07,
+      "loss": 0.5317,
+      "step": 4874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.149139379806281e-07,
+      "loss": 0.5304,
+      "step": 4875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.134215910168163e-07,
+      "loss": 0.5925,
+      "step": 4876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.119304907660541e-07,
+      "loss": 0.6383,
+      "step": 4877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.104406376723733e-07,
+      "loss": 0.5656,
+      "step": 4878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.089520321794336e-07,
+      "loss": 0.6181,
+      "step": 4879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.074646747305231e-07,
+      "loss": 0.536,
+      "step": 4880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.059785657685598e-07,
+      "loss": 0.4871,
+      "step": 4881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.044937057360902e-07,
+      "loss": 0.6162,
+      "step": 4882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.030100950752872e-07,
+      "loss": 0.569,
+      "step": 4883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.015277342279515e-07,
+      "loss": 0.5409,
+      "step": 4884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.000466236355137e-07,
+      "loss": 0.5374,
+      "step": 4885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.985667637390287e-07,
+      "loss": 0.65,
+      "step": 4886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.970881549791848e-07,
+      "loss": 0.5364,
+      "step": 4887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.956107977962901e-07,
+      "loss": 0.5942,
+      "step": 4888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.941346926302867e-07,
+      "loss": 0.4953,
+      "step": 4889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.926598399207392e-07,
+      "loss": 0.5625,
+      "step": 4890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.911862401068431e-07,
+      "loss": 0.2285,
+      "step": 4891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.897138936274168e-07,
+      "loss": 0.6065,
+      "step": 4892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.882428009209097e-07,
+      "loss": 0.6195,
+      "step": 4893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.867729624253929e-07,
+      "loss": 0.5724,
+      "step": 4894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.853043785785697e-07,
+      "loss": 0.5459,
+      "step": 4895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.838370498177649e-07,
+      "loss": 0.6083,
+      "step": 4896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.823709765799309e-07,
+      "loss": 0.5127,
+      "step": 4897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.809061593016481e-07,
+      "loss": 0.631,
+      "step": 4898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.794425984191223e-07,
+      "loss": 0.582,
+      "step": 4899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.779802943681836e-07,
+      "loss": 0.6445,
+      "step": 4900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76519247584287e-07,
+      "loss": 0.5219,
+      "step": 4901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.750594585025167e-07,
+      "loss": 0.65,
+      "step": 4902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.736009275575812e-07,
+      "loss": 0.5664,
+      "step": 4903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.721436551838125e-07,
+      "loss": 0.5663,
+      "step": 4904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.706876418151676e-07,
+      "loss": 0.5674,
+      "step": 4905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.692328878852312e-07,
+      "loss": 0.5403,
+      "step": 4906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.677793938272132e-07,
+      "loss": 0.6166,
+      "step": 4907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.663271600739447e-07,
+      "loss": 0.4886,
+      "step": 4908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.648761870578836e-07,
+      "loss": 0.633,
+      "step": 4909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.634264752111131e-07,
+      "loss": 0.5334,
+      "step": 4910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.619780249653391e-07,
+      "loss": 0.6082,
+      "step": 4911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.605308367518949e-07,
+      "loss": 0.6115,
+      "step": 4912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.590849110017329e-07,
+      "loss": 0.5428,
+      "step": 4913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.576402481454354e-07,
+      "loss": 0.6122,
+      "step": 4914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.561968486132027e-07,
+      "loss": 0.3014,
+      "step": 4915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547547128348643e-07,
+      "loss": 0.5606,
+      "step": 4916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.533138412398722e-07,
+      "loss": 0.5809,
+      "step": 4917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.518742342572971e-07,
+      "loss": 0.6628,
+      "step": 4918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.504358923158384e-07,
+      "loss": 0.5535,
+      "step": 4919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489988158438166e-07,
+      "loss": 0.5568,
+      "step": 4920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.475630052691796e-07,
+      "loss": 0.5809,
+      "step": 4921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.461284610194896e-07,
+      "loss": 0.5989,
+      "step": 4922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.446951835219385e-07,
+      "loss": 0.6388,
+      "step": 4923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.432631732033407e-07,
+      "loss": 0.5305,
+      "step": 4924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.418324304901292e-07,
+      "loss": 0.6402,
+      "step": 4925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.404029558083653e-07,
+      "loss": 0.5337,
+      "step": 4926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.389747495837263e-07,
+      "loss": 0.5025,
+      "step": 4927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.37547812241517e-07,
+      "loss": 0.5839,
+      "step": 4928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.361221442066607e-07,
+      "loss": 0.5968,
+      "step": 4929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.346977459037069e-07,
+      "loss": 0.5694,
+      "step": 4930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.332746177568212e-07,
+      "loss": 0.5832,
+      "step": 4931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.318527601897968e-07,
+      "loss": 0.5709,
+      "step": 4932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.304321736260439e-07,
+      "loss": 0.6022,
+      "step": 4933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.29012858488598e-07,
+      "loss": 0.6128,
+      "step": 4934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.275948152001139e-07,
+      "loss": 0.6093,
+      "step": 4935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.261780441828658e-07,
+      "loss": 0.6175,
+      "step": 4936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.247625458587526e-07,
+      "loss": 0.585,
+      "step": 4937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.233483206492948e-07,
+      "loss": 0.5338,
+      "step": 4938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.219353689756292e-07,
+      "loss": 0.5932,
+      "step": 4939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.205236912585156e-07,
+      "loss": 0.5507,
+      "step": 4940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.191132879183365e-07,
+      "loss": 0.6195,
+      "step": 4941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.177041593750933e-07,
+      "loss": 0.5766,
+      "step": 4942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.162963060484068e-07,
+      "loss": 0.6397,
+      "step": 4943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.148897283575179e-07,
+      "loss": 0.5566,
+      "step": 4944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.134844267212893e-07,
+      "loss": 0.54,
+      "step": 4945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.12080401558205e-07,
+      "loss": 0.5863,
+      "step": 4946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.106776532863657e-07,
+      "loss": 0.5319,
+      "step": 4947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.092761823234911e-07,
+      "loss": 0.5582,
+      "step": 4948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.078759890869258e-07,
+      "loss": 0.5862,
+      "step": 4949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.064770739936272e-07,
+      "loss": 0.6225,
+      "step": 4950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.050794374601788e-07,
+      "loss": 0.6285,
+      "step": 4951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.036830799027777e-07,
+      "loss": 0.5334,
+      "step": 4952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.022880017372441e-07,
+      "loss": 0.5297,
+      "step": 4953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.00894203379014e-07,
+      "loss": 0.5798,
+      "step": 4954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.995016852431458e-07,
+      "loss": 0.6177,
+      "step": 4955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.981104477443135e-07,
+      "loss": 0.5384,
+      "step": 4956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.967204912968107e-07,
+      "loss": 0.5781,
+      "step": 4957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.953318163145501e-07,
+      "loss": 0.5653,
+      "step": 4958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.939444232110642e-07,
+      "loss": 0.5774,
+      "step": 4959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.925583123995006e-07,
+      "loss": 0.5546,
+      "step": 4960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.911734842926254e-07,
+      "loss": 0.602,
+      "step": 4961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.897899393028252e-07,
+      "loss": 0.2912,
+      "step": 4962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.884076778421045e-07,
+      "loss": 0.6285,
+      "step": 4963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.870267003220826e-07,
+      "loss": 0.5471,
+      "step": 4964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.856470071539973e-07,
+      "loss": 0.5767,
+      "step": 4965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.842685987487057e-07,
+      "loss": 0.5586,
+      "step": 4966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.828914755166826e-07,
+      "loss": 0.5817,
+      "step": 4967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.815156378680182e-07,
+      "loss": 0.6116,
+      "step": 4968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.801410862124186e-07,
+      "loss": 0.62,
+      "step": 4969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.787678209592102e-07,
+      "loss": 0.5348,
+      "step": 4970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.773958425173366e-07,
+      "loss": 0.6331,
+      "step": 4971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.760251512953537e-07,
+      "loss": 0.5585,
+      "step": 4972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.746557477014398e-07,
+      "loss": 0.58,
+      "step": 4973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.732876321433851e-07,
+      "loss": 0.5518,
+      "step": 4974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.71920805028598e-07,
+      "loss": 0.5989,
+      "step": 4975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.705552667641036e-07,
+      "loss": 0.586,
+      "step": 4976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.691910177565436e-07,
+      "loss": 0.6116,
+      "step": 4977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.678280584121749e-07,
+      "loss": 0.6807,
+      "step": 4978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664663891368694e-07,
+      "loss": 0.6495,
+      "step": 4979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.651060103361162e-07,
+      "loss": 0.6167,
+      "step": 4980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.637469224150217e-07,
+      "loss": 0.5467,
+      "step": 4981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.62389125778305e-07,
+      "loss": 0.5866,
+      "step": 4982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.610326208302998e-07,
+      "loss": 0.5394,
+      "step": 4983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.596774079749591e-07,
+      "loss": 0.5612,
+      "step": 4984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.583234876158495e-07,
+      "loss": 0.5668,
+      "step": 4985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.569708601561515e-07,
+      "loss": 0.541,
+      "step": 4986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.556195259986603e-07,
+      "loss": 0.5972,
+      "step": 4987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.542694855457898e-07,
+      "loss": 0.5376,
+      "step": 4988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.529207391995628e-07,
+      "loss": 0.6008,
+      "step": 4989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.515732873616226e-07,
+      "loss": 0.6009,
+      "step": 4990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.50227130433222e-07,
+      "loss": 0.5348,
+      "step": 4991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.488822688152325e-07,
+      "loss": 0.583,
+      "step": 4992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.475387029081354e-07,
+      "loss": 0.6139,
+      "step": 4993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.461964331120313e-07,
+      "loss": 0.6226,
+      "step": 4994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.44855459826631e-07,
+      "loss": 0.6405,
+      "step": 4995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.43515783451259e-07,
+      "loss": 0.599,
+      "step": 4996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.421774043848561e-07,
+      "loss": 0.5432,
+      "step": 4997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408403230259763e-07,
+      "loss": 0.6098,
+      "step": 4998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.395045397727862e-07,
+      "loss": 0.6505,
+      "step": 4999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.381700550230641e-07,
+      "loss": 0.6647,
+      "step": 5000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.368368691742055e-07,
+      "loss": 0.6364,
+      "step": 5001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.355049826232179e-07,
+      "loss": 0.5846,
+      "step": 5002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.341743957667201e-07,
+      "loss": 0.634,
+      "step": 5003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.328451090009441e-07,
+      "loss": 0.5816,
+      "step": 5004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.315171227217365e-07,
+      "loss": 0.5596,
+      "step": 5005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301904373245565e-07,
+      "loss": 0.5587,
+      "step": 5006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.288650532044749e-07,
+      "loss": 0.5615,
+      "step": 5007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.275409707561741e-07,
+      "loss": 0.5745,
+      "step": 5008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.262181903739511e-07,
+      "loss": 0.5618,
+      "step": 5009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248967124517135e-07,
+      "loss": 0.6559,
+      "step": 5010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235765373829827e-07,
+      "loss": 0.2322,
+      "step": 5011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.222576655608891e-07,
+      "loss": 0.6146,
+      "step": 5012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.209400973781793e-07,
+      "loss": 0.583,
+      "step": 5013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.196238332272075e-07,
+      "loss": 0.5345,
+      "step": 5014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.183088734999426e-07,
+      "loss": 0.519,
+      "step": 5015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.169952185879629e-07,
+      "loss": 0.6264,
+      "step": 5016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.156828688824606e-07,
+      "loss": 0.5699,
+      "step": 5017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.143718247742364e-07,
+      "loss": 0.5982,
+      "step": 5018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.130620866537035e-07,
+      "loss": 0.6201,
+      "step": 5019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.117536549108893e-07,
+      "loss": 0.6811,
+      "step": 5020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.104465299354246e-07,
+      "loss": 0.5973,
+      "step": 5021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091407121165571e-07,
+      "loss": 0.5732,
+      "step": 5022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.078362018431449e-07,
+      "loss": 0.5666,
+      "step": 5023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.065329995036573e-07,
+      "loss": 0.2478,
+      "step": 5024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.052311054861676e-07,
+      "loss": 0.5563,
+      "step": 5025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.039305201783668e-07,
+      "loss": 0.6113,
+      "step": 5026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.026312439675553e-07,
+      "loss": 0.5799,
+      "step": 5027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.013332772406389e-07,
+      "loss": 0.5278,
+      "step": 5028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.00036620384139e-07,
+      "loss": 0.6136,
+      "step": 5029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.98741273784183e-07,
+      "loss": 0.4902,
+      "step": 5030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.974472378265112e-07,
+      "loss": 0.5603,
+      "step": 5031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.961545128964702e-07,
+      "loss": 0.265,
+      "step": 5032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9486309937902e-07,
+      "loss": 0.62,
+      "step": 5033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.935729976587274e-07,
+      "loss": 0.2468,
+      "step": 5034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922842081197672e-07,
+      "loss": 0.5922,
+      "step": 5035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.90996731145928e-07,
+      "loss": 0.61,
+      "step": 5036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897105671206049e-07,
+      "loss": 0.5693,
+      "step": 5037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.884257164268021e-07,
+      "loss": 0.4728,
+      "step": 5038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.871421794471311e-07,
+      "loss": 0.6722,
+      "step": 5039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.858599565638157e-07,
+      "loss": 0.5956,
+      "step": 5040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.845790481586866e-07,
+      "loss": 0.5985,
+      "step": 5041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.832994546131821e-07,
+      "loss": 0.6368,
+      "step": 5042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.820211763083494e-07,
+      "loss": 0.6192,
+      "step": 5043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807442136248459e-07,
+      "loss": 0.5658,
+      "step": 5044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.794685669429356e-07,
+      "loss": 0.5696,
+      "step": 5045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781942366424909e-07,
+      "loss": 0.5913,
+      "step": 5046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.769212231029903e-07,
+      "loss": 0.5511,
+      "step": 5047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.75649526703525e-07,
+      "loss": 0.6472,
+      "step": 5048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.743791478227883e-07,
+      "loss": 0.5475,
+      "step": 5049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.73110086839086e-07,
+      "loss": 0.5766,
+      "step": 5050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.718423441303273e-07,
+      "loss": 0.6185,
+      "step": 5051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.70575920074033e-07,
+      "loss": 0.6006,
+      "step": 5052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.69310815047327e-07,
+      "loss": 0.2302,
+      "step": 5053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.680470294269441e-07,
+      "loss": 0.4775,
+      "step": 5054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.667845635892233e-07,
+      "loss": 0.5411,
+      "step": 5055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.655234179101132e-07,
+      "loss": 0.5611,
+      "step": 5056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.642635927651657e-07,
+      "loss": 0.5459,
+      "step": 5057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.630050885295446e-07,
+      "loss": 0.5115,
+      "step": 5058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.617479055780162e-07,
+      "loss": 0.5663,
+      "step": 5059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.604920442849532e-07,
+      "loss": 0.5135,
+      "step": 5060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.59237505024337e-07,
+      "loss": 0.5929,
+      "step": 5061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.579842881697556e-07,
+      "loss": 0.5818,
+      "step": 5062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.567323940944009e-07,
+      "loss": 0.6185,
+      "step": 5063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.554818231710713e-07,
+      "loss": 0.6115,
+      "step": 5064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.542325757721728e-07,
+      "loss": 0.6143,
+      "step": 5065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.529846522697175e-07,
+      "loss": 0.597,
+      "step": 5066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.517380530353206e-07,
+      "loss": 0.5588,
+      "step": 5067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.504927784402037e-07,
+      "loss": 0.57,
+      "step": 5068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.492488288551962e-07,
+      "loss": 0.5713,
+      "step": 5069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480062046507312e-07,
+      "loss": 0.6422,
+      "step": 5070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.467649061968478e-07,
+      "loss": 0.6081,
+      "step": 5071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.455249338631879e-07,
+      "loss": 0.6106,
+      "step": 5072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.442862880190031e-07,
+      "loss": 0.5158,
+      "step": 5073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.430489690331447e-07,
+      "loss": 0.2771,
+      "step": 5074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.418129772740727e-07,
+      "loss": 0.2537,
+      "step": 5075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.405783131098519e-07,
+      "loss": 0.5866,
+      "step": 5076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.393449769081494e-07,
+      "loss": 0.6187,
+      "step": 5077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.381129690362369e-07,
+      "loss": 0.6559,
+      "step": 5078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.368822898609932e-07,
+      "loss": 0.592,
+      "step": 5079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.356529397489013e-07,
+      "loss": 0.574,
+      "step": 5080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344249190660427e-07,
+      "loss": 0.2698,
+      "step": 5081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3319822817811e-07,
+      "loss": 0.5934,
+      "step": 5082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.319728674503965e-07,
+      "loss": 0.6604,
+      "step": 5083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307488372478015e-07,
+      "loss": 0.5967,
+      "step": 5084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.295261379348254e-07,
+      "loss": 0.6049,
+      "step": 5085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.283047698755723e-07,
+      "loss": 0.5738,
+      "step": 5086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.270847334337525e-07,
+      "loss": 0.6155,
+      "step": 5087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.25866028972678e-07,
+      "loss": 0.5914,
+      "step": 5088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.246486568552644e-07,
+      "loss": 0.6456,
+      "step": 5089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.234326174440302e-07,
+      "loss": 0.5163,
+      "step": 5090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.222179111010983e-07,
+      "loss": 0.628,
+      "step": 5091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.210045381881923e-07,
+      "loss": 0.5786,
+      "step": 5092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19792499066642e-07,
+      "loss": 0.6208,
+      "step": 5093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.185817940973764e-07,
+      "loss": 0.5814,
+      "step": 5094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.173724236409306e-07,
+      "loss": 0.6776,
+      "step": 5095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.161643880574385e-07,
+      "loss": 0.5976,
+      "step": 5096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.149576877066414e-07,
+      "loss": 0.5826,
+      "step": 5097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.137523229478791e-07,
+      "loss": 0.6184,
+      "step": 5098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.125482941400933e-07,
+      "loss": 0.608,
+      "step": 5099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113456016418305e-07,
+      "loss": 0.6209,
+      "step": 5100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.101442458112399e-07,
+      "loss": 0.6097,
+      "step": 5101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.08944227006069e-07,
+      "loss": 0.5836,
+      "step": 5102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.077455455836682e-07,
+      "loss": 0.6555,
+      "step": 5103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.065482019009926e-07,
+      "loss": 0.5814,
+      "step": 5104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.053521963145963e-07,
+      "loss": 0.5974,
+      "step": 5105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041575291806356e-07,
+      "loss": 0.5615,
+      "step": 5106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.029642008548663e-07,
+      "loss": 0.5552,
+      "step": 5107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.017722116926488e-07,
+      "loss": 0.5764,
+      "step": 5108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005815620489446e-07,
+      "loss": 0.5615,
+      "step": 5109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.993922522783129e-07,
+      "loss": 0.6469,
+      "step": 5110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.98204282734916e-07,
+      "loss": 0.5741,
+      "step": 5111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.970176537725186e-07,
+      "loss": 0.6352,
+      "step": 5112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.958323657444824e-07,
+      "loss": 0.5733,
+      "step": 5113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946484190037743e-07,
+      "loss": 0.567,
+      "step": 5114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.934658139029574e-07,
+      "loss": 0.5379,
+      "step": 5115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.922845507941993e-07,
+      "loss": 0.6167,
+      "step": 5116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.911046300292644e-07,
+      "loss": 0.6068,
+      "step": 5117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.899260519595206e-07,
+      "loss": 0.6097,
+      "step": 5118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.88748816935934e-07,
+      "loss": 0.5218,
+      "step": 5119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.875729253090688e-07,
+      "loss": 0.525,
+      "step": 5120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.863983774290942e-07,
+      "loss": 0.5828,
+      "step": 5121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.852251736457752e-07,
+      "loss": 0.5433,
+      "step": 5122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840533143084802e-07,
+      "loss": 0.6055,
+      "step": 5123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.828827997661717e-07,
+      "loss": 0.4845,
+      "step": 5124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.817136303674158e-07,
+      "loss": 0.6874,
+      "step": 5125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.805458064603791e-07,
+      "loss": 0.581,
+      "step": 5126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.793793283928239e-07,
+      "loss": 0.5678,
+      "step": 5127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.782141965121129e-07,
+      "loss": 0.5588,
+      "step": 5128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.770504111652091e-07,
+      "loss": 0.6138,
+      "step": 5129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7588797269867535e-07,
+      "loss": 0.4867,
+      "step": 5130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7472688145866975e-07,
+      "loss": 0.5628,
+      "step": 5131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.735671377909534e-07,
+      "loss": 0.5379,
+      "step": 5132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7240874204088186e-07,
+      "loss": 0.5537,
+      "step": 5133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7125169455341347e-07,
+      "loss": 0.5632,
+      "step": 5134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.700959956731022e-07,
+      "loss": 0.5623,
+      "step": 5135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6894164574410203e-07,
+      "loss": 0.5738,
+      "step": 5136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.677886451101643e-07,
+      "loss": 0.5365,
+      "step": 5137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.666369941146376e-07,
+      "loss": 0.596,
+      "step": 5138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.654866931004709e-07,
+      "loss": 0.5542,
+      "step": 5139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.643377424102108e-07,
+      "loss": 0.6634,
+      "step": 5140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.631901423860008e-07,
+      "loss": 0.5948,
+      "step": 5141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6204389336958145e-07,
+      "loss": 0.5819,
+      "step": 5142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6089899570229245e-07,
+      "loss": 0.5832,
+      "step": 5143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597554497250717e-07,
+      "loss": 0.5764,
+      "step": 5144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.586132557784523e-07,
+      "loss": 0.4833,
+      "step": 5145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57472414202566e-07,
+      "loss": 0.5824,
+      "step": 5146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.563329253371418e-07,
+      "loss": 0.6319,
+      "step": 5147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5519478952150754e-07,
+      "loss": 0.5845,
+      "step": 5148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5405800709458494e-07,
+      "loss": 0.5994,
+      "step": 5149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5292257839489374e-07,
+      "loss": 0.5465,
+      "step": 5150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.517885037605524e-07,
+      "loss": 0.5837,
+      "step": 5151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.506557835292735e-07,
+      "loss": 0.5473,
+      "step": 5152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.49524418038369e-07,
+      "loss": 0.638,
+      "step": 5153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.483944076247448e-07,
+      "loss": 0.654,
+      "step": 5154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4726575262490523e-07,
+      "loss": 0.6808,
+      "step": 5155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4613845337494967e-07,
+      "loss": 0.5997,
+      "step": 5156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4501251021057566e-07,
+      "loss": 0.6116,
+      "step": 5157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4388792346707453e-07,
+      "loss": 0.6618,
+      "step": 5158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4276469347933394e-07,
+      "loss": 0.5459,
+      "step": 5159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.416428205818396e-07,
+      "loss": 0.5548,
+      "step": 5160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4052230510867213e-07,
+      "loss": 0.5777,
+      "step": 5161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3940314739350785e-07,
+      "loss": 0.5548,
+      "step": 5162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.382853477696164e-07,
+      "loss": 0.5952,
+      "step": 5163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.37168906569867e-07,
+      "loss": 0.5171,
+      "step": 5164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360538241267226e-07,
+      "loss": 0.6273,
+      "step": 5165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3494010077224147e-07,
+      "loss": 0.5959,
+      "step": 5166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3382773683807553e-07,
+      "loss": 0.5789,
+      "step": 5167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.327167326554749e-07,
+      "loss": 0.5587,
+      "step": 5168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316070885552837e-07,
+      "loss": 0.6152,
+      "step": 5169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3049880486793984e-07,
+      "loss": 0.5705,
+      "step": 5170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2939188192347656e-07,
+      "loss": 0.5539,
+      "step": 5171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.282863200515225e-07,
+      "loss": 0.6437,
+      "step": 5172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.271821195813025e-07,
+      "loss": 0.2963,
+      "step": 5173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.260792808416325e-07,
+      "loss": 0.603,
+      "step": 5174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2497780416092426e-07,
+      "loss": 0.6231,
+      "step": 5175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2387768986718644e-07,
+      "loss": 0.5314,
+      "step": 5176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.227789382880182e-07,
+      "loss": 0.5655,
+      "step": 5177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2168154975061435e-07,
+      "loss": 0.6025,
+      "step": 5178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2058552458176637e-07,
+      "loss": 0.588,
+      "step": 5179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194908631078559e-07,
+      "loss": 0.6679,
+      "step": 5180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1839756565485943e-07,
+      "loss": 0.6445,
+      "step": 5181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1730563254834875e-07,
+      "loss": 0.6009,
+      "step": 5182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1621506411348986e-07,
+      "loss": 0.6024,
+      "step": 5183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1512586067503804e-07,
+      "loss": 0.5947,
+      "step": 5184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140380225573459e-07,
+      "loss": 0.5666,
+      "step": 5185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1295155008436005e-07,
+      "loss": 0.5769,
+      "step": 5186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.118664435796188e-07,
+      "loss": 0.6348,
+      "step": 5187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.107827033662537e-07,
+      "loss": 0.6251,
+      "step": 5188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.097003297669883e-07,
+      "loss": 0.6409,
+      "step": 5189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0861932310414243e-07,
+      "loss": 0.5422,
+      "step": 5190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0753968369962483e-07,
+      "loss": 0.5958,
+      "step": 5191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0646141187494214e-07,
+      "loss": 0.5528,
+      "step": 5192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0538450795118754e-07,
+      "loss": 0.5395,
+      "step": 5193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0430897224905286e-07,
+      "loss": 0.5904,
+      "step": 5194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.03234805088818e-07,
+      "loss": 0.6204,
+      "step": 5195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.021620067903587e-07,
+      "loss": 0.5327,
+      "step": 5196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.010905776731405e-07,
+      "loss": 0.5082,
+      "step": 5197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.000205180562217e-07,
+      "loss": 0.5965,
+      "step": 5198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9895182825825386e-07,
+      "loss": 0.5799,
+      "step": 5199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9788450859748027e-07,
+      "loss": 0.6314,
+      "step": 5200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.968185593917362e-07,
+      "loss": 0.5813,
+      "step": 5201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9575398095844766e-07,
+      "loss": 0.5508,
+      "step": 5202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9469077361463325e-07,
+      "loss": 0.5248,
+      "step": 5203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9362893767690534e-07,
+      "loss": 0.595,
+      "step": 5204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9256847346146434e-07,
+      "loss": 0.5461,
+      "step": 5205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9150938128410366e-07,
+      "loss": 0.5984,
+      "step": 5206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.904516614602089e-07,
+      "loss": 0.6077,
+      "step": 5207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8939531430475687e-07,
+      "loss": 0.6034,
+      "step": 5208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.883403401323149e-07,
+      "loss": 0.5657,
+      "step": 5209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.872867392570401e-07,
+      "loss": 0.5716,
+      "step": 5210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.862345119926841e-07,
+      "loss": 0.5898,
+      "step": 5211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.851836586525876e-07,
+      "loss": 0.5867,
+      "step": 5212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8413417954968167e-07,
+      "loss": 0.6131,
+      "step": 5213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8308607499648765e-07,
+      "loss": 0.6224,
+      "step": 5214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.820393453051202e-07,
+      "loss": 0.6169,
+      "step": 5215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8099399078728117e-07,
+      "loss": 0.5843,
+      "step": 5216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7995001175426706e-07,
+      "loss": 0.6243,
+      "step": 5217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7890740851696027e-07,
+      "loss": 0.6106,
+      "step": 5218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.778661813858364e-07,
+      "loss": 0.602,
+      "step": 5219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.768263306709602e-07,
+      "loss": 0.5533,
+      "step": 5220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.757878566819878e-07,
+      "loss": 0.613,
+      "step": 5221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.74750759728163e-07,
+      "loss": 0.6857,
+      "step": 5222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737150401183209e-07,
+      "loss": 0.6213,
+      "step": 5223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7268069816088715e-07,
+      "loss": 0.5931,
+      "step": 5224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.716477341638758e-07,
+      "loss": 0.5343,
+      "step": 5225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.706161484348936e-07,
+      "loss": 0.6302,
+      "step": 5226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6958594128113035e-07,
+      "loss": 0.5565,
+      "step": 5227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.685571130093718e-07,
+      "loss": 0.6716,
+      "step": 5228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.675296639259912e-07,
+      "loss": 0.5413,
+      "step": 5229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6650359433695047e-07,
+      "loss": 0.5629,
+      "step": 5230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.654789045477991e-07,
+      "loss": 0.593,
+      "step": 5231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6445559486367855e-07,
+      "loss": 0.5893,
+      "step": 5232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.634336655893189e-07,
+      "loss": 0.6635,
+      "step": 5233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6241311702903735e-07,
+      "loss": 0.6403,
+      "step": 5234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6139394948674247e-07,
+      "loss": 0.5304,
+      "step": 5235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.603761632659297e-07,
+      "loss": 0.5951,
+      "step": 5236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5935975866968233e-07,
+      "loss": 0.5873,
+      "step": 5237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5834473600067423e-07,
+      "loss": 0.5858,
+      "step": 5238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5733109556116805e-07,
+      "loss": 0.556,
+      "step": 5239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5631883765301303e-07,
+      "loss": 0.6714,
+      "step": 5240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5530796257764676e-07,
+      "loss": 0.5725,
+      "step": 5241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5429847063609633e-07,
+      "loss": 0.5931,
+      "step": 5242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532903621289774e-07,
+      "loss": 0.5716,
+      "step": 5243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5228363735649196e-07,
+      "loss": 0.6757,
+      "step": 5244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5127829661842973e-07,
+      "loss": 0.5917,
+      "step": 5245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.502743402141695e-07,
+      "loss": 0.6431,
+      "step": 5246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4927176844267886e-07,
+      "loss": 0.613,
+      "step": 5247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.482705816025106e-07,
+      "loss": 0.6429,
+      "step": 5248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.472707799918057e-07,
+      "loss": 0.6517,
+      "step": 5249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.462723639082932e-07,
+      "loss": 0.2498,
+      "step": 5250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4527533364929145e-07,
+      "loss": 0.2465,
+      "step": 5251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4427968951170287e-07,
+      "loss": 0.6316,
+      "step": 5252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.43285431792017e-07,
+      "loss": 0.2332,
+      "step": 5253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4229256078631424e-07,
+      "loss": 0.5693,
+      "step": 5254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4130107679025803e-07,
+      "loss": 0.6161,
+      "step": 5255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403109800991017e-07,
+      "loss": 0.6104,
+      "step": 5256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3932227100768267e-07,
+      "loss": 0.6117,
+      "step": 5257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.383349498104288e-07,
+      "loss": 0.5832,
+      "step": 5258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3734901680135047e-07,
+      "loss": 0.5553,
+      "step": 5259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3636447227404835e-07,
+      "loss": 0.5868,
+      "step": 5260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.353813165217074e-07,
+      "loss": 0.6743,
+      "step": 5261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.343995498370989e-07,
+      "loss": 0.5445,
+      "step": 5262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.334191725125818e-07,
+      "loss": 0.6399,
+      "step": 5263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.324401848401021e-07,
+      "loss": 0.658,
+      "step": 5264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3146258711118927e-07,
+      "loss": 0.6531,
+      "step": 5265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3048637961695994e-07,
+      "loss": 0.5796,
+      "step": 5266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.295115626481171e-07,
+      "loss": 0.5892,
+      "step": 5267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.285381364949514e-07,
+      "loss": 0.6332,
+      "step": 5268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.275661014473358e-07,
+      "loss": 0.6318,
+      "step": 5269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2659545779473e-07,
+      "loss": 0.5866,
+      "step": 5270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.256262058261816e-07,
+      "loss": 0.6232,
+      "step": 5271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.246583458303226e-07,
+      "loss": 0.6291,
+      "step": 5272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2369187809536907e-07,
+      "loss": 0.6146,
+      "step": 5273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.227268029091235e-07,
+      "loss": 0.6534,
+      "step": 5274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2176312055897497e-07,
+      "loss": 0.5953,
+      "step": 5275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.208008313318944e-07,
+      "loss": 0.6169,
+      "step": 5276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.198399355144427e-07,
+      "loss": 0.6455,
+      "step": 5277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1888043339276086e-07,
+      "loss": 0.602,
+      "step": 5278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.179223252525787e-07,
+      "loss": 0.5709,
+      "step": 5279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.169656113792086e-07,
+      "loss": 0.6093,
+      "step": 5280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.160102920575481e-07,
+      "loss": 0.626,
+      "step": 5281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1505636757208247e-07,
+      "loss": 0.58,
+      "step": 5282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.141038382068756e-07,
+      "loss": 0.5451,
+      "step": 5283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.131527042455806e-07,
+      "loss": 0.5718,
+      "step": 5284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.122029659714343e-07,
+      "loss": 0.5863,
+      "step": 5285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.112546236672581e-07,
+      "loss": 0.5317,
+      "step": 5286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103076776154551e-07,
+      "loss": 0.2718,
+      "step": 5287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0936212809801527e-07,
+      "loss": 0.5949,
+      "step": 5288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.084179753965122e-07,
+      "loss": 0.5891,
+      "step": 5289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0747521979210436e-07,
+      "loss": 0.6162,
+      "step": 5290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0653386156553167e-07,
+      "loss": 0.59,
+      "step": 5291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.055939009971193e-07,
+      "loss": 0.5834,
+      "step": 5292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.046553383667772e-07,
+      "loss": 0.5855,
+      "step": 5293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.037181739539968e-07,
+      "loss": 0.5814,
+      "step": 5294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.027824080378561e-07,
+      "loss": 0.6419,
+      "step": 5295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0184804089701314e-07,
+      "loss": 0.5986,
+      "step": 5296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.009150728097138e-07,
+      "loss": 0.5844,
+      "step": 5297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9998350405378185e-07,
+      "loss": 0.5863,
+      "step": 5298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9905333490663044e-07,
+      "loss": 0.634,
+      "step": 5299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.981245656452508e-07,
+      "loss": 0.5878,
+      "step": 5300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.971971965462189e-07,
+      "loss": 0.577,
+      "step": 5301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.962712278856955e-07,
+      "loss": 0.5969,
+      "step": 5302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.953466599394228e-07,
+      "loss": 0.6094,
+      "step": 5303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9442349298272665e-07,
+      "loss": 0.6176,
+      "step": 5304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9350172729051276e-07,
+      "loss": 0.5149,
+      "step": 5305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9258136313727425e-07,
+      "loss": 0.6342,
+      "step": 5306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.916624007970842e-07,
+      "loss": 0.5968,
+      "step": 5307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9074484054359864e-07,
+      "loss": 0.6027,
+      "step": 5308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8982868265005457e-07,
+      "loss": 0.4773,
+      "step": 5309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.889139273892744e-07,
+      "loss": 0.634,
+      "step": 5310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.880005750336612e-07,
+      "loss": 0.6077,
+      "step": 5311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8708862585520025e-07,
+      "loss": 0.6232,
+      "step": 5312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8617808012545824e-07,
+      "loss": 0.5496,
+      "step": 5313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.852689381155865e-07,
+      "loss": 0.6339,
+      "step": 5314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8436120009631527e-07,
+      "loss": 0.6359,
+      "step": 5315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.834548663379588e-07,
+      "loss": 0.5865,
+      "step": 5316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825499371104118e-07,
+      "loss": 0.6491,
+      "step": 5317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.816464126831525e-07,
+      "loss": 0.6102,
+      "step": 5318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.80744293325238e-07,
+      "loss": 0.5899,
+      "step": 5319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.798435793053106e-07,
+      "loss": 0.5942,
+      "step": 5320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.789442708915907e-07,
+      "loss": 0.6287,
+      "step": 5321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7804636835188193e-07,
+      "loss": 0.5691,
+      "step": 5322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7714987195356923e-07,
+      "loss": 0.6193,
+      "step": 5323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.762547819636191e-07,
+      "loss": 0.5773,
+      "step": 5324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7536109864857774e-07,
+      "loss": 0.6847,
+      "step": 5325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7446882227457336e-07,
+      "loss": 0.5114,
+      "step": 5326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.735779531073152e-07,
+      "loss": 0.6361,
+      "step": 5327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.726884914120936e-07,
+      "loss": 0.6005,
+      "step": 5328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7180043745378084e-07,
+      "loss": 0.559,
+      "step": 5329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7091379149682683e-07,
+      "loss": 0.5562,
+      "step": 5330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7002855380526494e-07,
+      "loss": 0.7035,
+      "step": 5331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6914472464270925e-07,
+      "loss": 0.5624,
+      "step": 5332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.682623042723526e-07,
+      "loss": 0.5963,
+      "step": 5333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.673812929569686e-07,
+      "loss": 0.5968,
+      "step": 5334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6650169095891245e-07,
+      "loss": 0.625,
+      "step": 5335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6562349854012014e-07,
+      "loss": 0.6784,
+      "step": 5336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.647467159621053e-07,
+      "loss": 0.2714,
+      "step": 5337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.638713434859652e-07,
+      "loss": 0.5475,
+      "step": 5338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.629973813723741e-07,
+      "loss": 0.5751,
+      "step": 5339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621248298815865e-07,
+      "loss": 0.6414,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6125368927343963e-07,
+      "loss": 0.5732,
+      "step": 5341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.603839598073482e-07,
+      "loss": 0.5262,
+      "step": 5342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.595156417423067e-07,
+      "loss": 0.6225,
+      "step": 5343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5864873533688996e-07,
+      "loss": 0.5327,
+      "step": 5344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.577832408492531e-07,
+      "loss": 0.5795,
+      "step": 5345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5691915853712944e-07,
+      "loss": 0.6051,
+      "step": 5346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5605648865783315e-07,
+      "loss": 0.5943,
+      "step": 5347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551952314682554e-07,
+      "loss": 0.6401,
+      "step": 5348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.543353872248694e-07,
+      "loss": 0.5614,
+      "step": 5349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5347695618372694e-07,
+      "loss": 0.5414,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.52619938600458e-07,
+      "loss": 0.6969,
+      "step": 5351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5176433473027176e-07,
+      "loss": 0.4815,
+      "step": 5352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.509101448279577e-07,
+      "loss": 0.6351,
+      "step": 5353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5005736914788194e-07,
+      "loss": 0.5899,
+      "step": 5354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.492060079439929e-07,
+      "loss": 0.2782,
+      "step": 5355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4835606146981385e-07,
+      "loss": 0.5997,
+      "step": 5356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4750752997845075e-07,
+      "loss": 0.2552,
+      "step": 5357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4666041372258367e-07,
+      "loss": 0.5547,
+      "step": 5358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.458147129544758e-07,
+      "loss": 0.5837,
+      "step": 5359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4497042792596627e-07,
+      "loss": 0.6159,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4412755888847174e-07,
+      "loss": 0.6425,
+      "step": 5361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.432861060929897e-07,
+      "loss": 0.5489,
+      "step": 5362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4244606979009543e-07,
+      "loss": 0.584,
+      "step": 5363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4160745022994035e-07,
+      "loss": 0.6442,
+      "step": 5364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.407702476622559e-07,
+      "loss": 0.6627,
+      "step": 5365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.399344623363503e-07,
+      "loss": 0.5754,
+      "step": 5366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3910009450111236e-07,
+      "loss": 0.5969,
+      "step": 5367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3826714440500554e-07,
+      "loss": 0.6425,
+      "step": 5368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3743561229607137e-07,
+      "loss": 0.4975,
+      "step": 5369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.366054984219318e-07,
+      "loss": 0.6503,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3577680302978468e-07,
+      "loss": 0.5791,
+      "step": 5371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3494952636640533e-07,
+      "loss": 0.5756,
+      "step": 5372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3412366867814617e-07,
+      "loss": 0.2516,
+      "step": 5373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3329923021093825e-07,
+      "loss": 0.6034,
+      "step": 5374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3247621121029075e-07,
+      "loss": 0.5887,
+      "step": 5375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3165461192128757e-07,
+      "loss": 0.6066,
+      "step": 5376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.308344325885903e-07,
+      "loss": 0.5761,
+      "step": 5377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3001567345644072e-07,
+      "loss": 0.5648,
+      "step": 5378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2919833476865382e-07,
+      "loss": 0.5182,
+      "step": 5379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.283824167686244e-07,
+      "loss": 0.6504,
+      "step": 5380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275679196993219e-07,
+      "loss": 0.6088,
+      "step": 5381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2675484380329516e-07,
+      "loss": 0.4933,
+      "step": 5382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2594318932266711e-07,
+      "loss": 0.5371,
+      "step": 5383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2513295649914003e-07,
+      "loss": 0.5334,
+      "step": 5384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2432414557399197e-07,
+      "loss": 0.6164,
+      "step": 5385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.235167567880747e-07,
+      "loss": 0.4983,
+      "step": 5386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2271079038182042e-07,
+      "loss": 0.542,
+      "step": 5387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2190624659523652e-07,
+      "loss": 0.6249,
+      "step": 5388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2110312566790803e-07,
+      "loss": 0.6128,
+      "step": 5389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2030142783899145e-07,
+      "loss": 0.5912,
+      "step": 5390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1950115334722467e-07,
+      "loss": 0.5863,
+      "step": 5391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1870230243091984e-07,
+      "loss": 0.583,
+      "step": 5392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.17904875327965e-07,
+      "loss": 0.5571,
+      "step": 5393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1710887227582522e-07,
+      "loss": 0.5916,
+      "step": 5394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1631429351153865e-07,
+      "loss": 0.5626,
+      "step": 5395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155211392717238e-07,
+      "loss": 0.5732,
+      "step": 5396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1472940979257063e-07,
+      "loss": 0.6686,
+      "step": 5397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1393910530984828e-07,
+      "loss": 0.6567,
+      "step": 5398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1315022605889912e-07,
+      "loss": 0.5064,
+      "step": 5399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.123627722746413e-07,
+      "loss": 0.6127,
+      "step": 5400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1157674419157004e-07,
+      "loss": 0.5517,
+      "step": 5401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.107921420437553e-07,
+      "loss": 0.6035,
+      "step": 5402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1000896606484246e-07,
+      "loss": 0.62,
+      "step": 5403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0922721648805045e-07,
+      "loss": 0.2524,
+      "step": 5404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0844689354617587e-07,
+      "loss": 0.5987,
+      "step": 5405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0766799747159006e-07,
+      "loss": 0.5908,
+      "step": 5406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0689052849623913e-07,
+      "loss": 0.5371,
+      "step": 5407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0611448685164237e-07,
+      "loss": 0.6086,
+      "step": 5408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0533987276889712e-07,
+      "loss": 0.5169,
+      "step": 5409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.045666864786744e-07,
+      "loss": 0.6185,
+      "step": 5410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0379492821121948e-07,
+      "loss": 0.5588,
+      "step": 5411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0302459819635135e-07,
+      "loss": 0.5979,
+      "step": 5412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.02255696663467e-07,
+      "loss": 0.5876,
+      "step": 5413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.014882238415361e-07,
+      "loss": 0.6099,
+      "step": 5414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0072217995910192e-07,
+      "loss": 0.6005,
+      "step": 5415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9995756524428312e-07,
+      "loss": 0.5433,
+      "step": 5416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9919437992477363e-07,
+      "loss": 0.5563,
+      "step": 5417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9843262422784059e-07,
+      "loss": 0.5567,
+      "step": 5418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9767229838032586e-07,
+      "loss": 0.5456,
+      "step": 5419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9691340260864443e-07,
+      "loss": 0.651,
+      "step": 5420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961559371387878e-07,
+      "loss": 0.6247,
+      "step": 5421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9539990219631932e-07,
+      "loss": 0.608,
+      "step": 5422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9464529800637731e-07,
+      "loss": 0.5899,
+      "step": 5423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.938921247936748e-07,
+      "loss": 0.6103,
+      "step": 5424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9314038278249568e-07,
+      "loss": 0.6144,
+      "step": 5425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9239007219670147e-07,
+      "loss": 0.5659,
+      "step": 5426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9164119325972563e-07,
+      "loss": 0.6125,
+      "step": 5427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.908937461945748e-07,
+      "loss": 0.651,
+      "step": 5428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9014773122382924e-07,
+      "loss": 0.5428,
+      "step": 5429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8940314856964459e-07,
+      "loss": 0.5429,
+      "step": 5430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8865999845374794e-07,
+      "loss": 0.6572,
+      "step": 5431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791828109744115e-07,
+      "loss": 0.6464,
+      "step": 5432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8717799672159808e-07,
+      "loss": 0.6609,
+      "step": 5433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8643914554666632e-07,
+      "loss": 0.5986,
+      "step": 5434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8570172779266759e-07,
+      "loss": 0.6002,
+      "step": 5435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8496574367919685e-07,
+      "loss": 0.5783,
+      "step": 5436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.842311934254193e-07,
+      "loss": 0.6401,
+      "step": 5437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.834980772500766e-07,
+      "loss": 0.6116,
+      "step": 5438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.82766395371482e-07,
+      "loss": 0.5551,
+      "step": 5439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8203614800752055e-07,
+      "loss": 0.5703,
+      "step": 5440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8130733537565336e-07,
+      "loss": 0.5674,
+      "step": 5441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.805799576929107e-07,
+      "loss": 0.5892,
+      "step": 5442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7985401517589597e-07,
+      "loss": 0.5216,
+      "step": 5443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7912950804078787e-07,
+      "loss": 0.6179,
+      "step": 5444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7840643650333555e-07,
+      "loss": 0.595,
+      "step": 5445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7768480077886062e-07,
+      "loss": 0.2701,
+      "step": 5446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.769646010822579e-07,
+      "loss": 0.6191,
+      "step": 5447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7624583762799408e-07,
+      "loss": 0.5905,
+      "step": 5448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7552851063010856e-07,
+      "loss": 0.6133,
+      "step": 5449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7481262030221325e-07,
+      "loss": 0.6538,
+      "step": 5450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7409816685749036e-07,
+      "loss": 0.5238,
+      "step": 5451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.733851505086964e-07,
+      "loss": 0.5881,
+      "step": 5452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7267357146815988e-07,
+      "loss": 0.5292,
+      "step": 5453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7196342994777903e-07,
+      "loss": 0.6039,
+      "step": 5454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7125472615902638e-07,
+      "loss": 0.5996,
+      "step": 5455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7054746031294588e-07,
+      "loss": 0.2515,
+      "step": 5456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.698416326201513e-07,
+      "loss": 0.5873,
+      "step": 5457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.691372432908317e-07,
+      "loss": 0.585,
+      "step": 5458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6843429253474374e-07,
+      "loss": 0.6817,
+      "step": 5459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6773278056122e-07,
+      "loss": 0.5122,
+      "step": 5460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6703270757916e-07,
+      "loss": 0.6257,
+      "step": 5461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6633407379703925e-07,
+      "loss": 0.6202,
+      "step": 5462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6563687942290185e-07,
+      "loss": 0.6025,
+      "step": 5463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6494112466436285e-07,
+      "loss": 0.6523,
+      "step": 5464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.642468097286104e-07,
+      "loss": 0.636,
+      "step": 5465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6355393482240412e-07,
+      "loss": 0.5723,
+      "step": 5466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286250015207283e-07,
+      "loss": 0.548,
+      "step": 5467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621725059235174e-07,
+      "loss": 0.2511,
+      "step": 5468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6148395234221015e-07,
+      "loss": 0.5672,
+      "step": 5469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6079683961319537e-07,
+      "loss": 0.5449,
+      "step": 5470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601111679410855e-07,
+      "loss": 0.6683,
+      "step": 5471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5942693753006498e-07,
+      "loss": 0.5793,
+      "step": 5472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5874414858389088e-07,
+      "loss": 0.5755,
+      "step": 5473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.580628013058888e-07,
+      "loss": 0.6443,
+      "step": 5474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.573828958989565e-07,
+      "loss": 0.5719,
+      "step": 5475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5670443256556033e-07,
+      "loss": 0.6529,
+      "step": 5476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5602741150773983e-07,
+      "loss": 0.596,
+      "step": 5477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.553518329271031e-07,
+      "loss": 0.5592,
+      "step": 5478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5467769702483036e-07,
+      "loss": 0.2527,
+      "step": 5479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.540050040016694e-07,
+      "loss": 0.5697,
+      "step": 5480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5333375405794216e-07,
+      "loss": 0.5439,
+      "step": 5481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5266394739353764e-07,
+      "loss": 0.5145,
+      "step": 5482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5199558420791682e-07,
+      "loss": 0.627,
+      "step": 5483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5132866470010943e-07,
+      "loss": 0.2633,
+      "step": 5484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.506631890687177e-07,
+      "loss": 0.5326,
+      "step": 5485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.499991575119103e-07,
+      "loss": 0.6701,
+      "step": 5486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.493365702274291e-07,
+      "loss": 0.5903,
+      "step": 5487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4867542741258621e-07,
+      "loss": 0.5784,
+      "step": 5488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4801572926425868e-07,
+      "loss": 0.5385,
+      "step": 5489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.473574759788987e-07,
+      "loss": 0.5285,
+      "step": 5490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4670066775252622e-07,
+      "loss": 0.5785,
+      "step": 5491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4604530478073142e-07,
+      "loss": 0.5689,
+      "step": 5492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4539138725867154e-07,
+      "loss": 0.5313,
+      "step": 5493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4473891538107686e-07,
+      "loss": 0.6268,
+      "step": 5494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.440878893422465e-07,
+      "loss": 0.5737,
+      "step": 5495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.434383093360464e-07,
+      "loss": 0.57,
+      "step": 5496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4279017555591524e-07,
+      "loss": 0.5796,
+      "step": 5497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4214348819485802e-07,
+      "loss": 0.563,
+      "step": 5498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.414982474454524e-07,
+      "loss": 0.5685,
+      "step": 5499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4085445349984185e-07,
+      "loss": 0.6049,
+      "step": 5500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4021210654974082e-07,
+      "loss": 0.5621,
+      "step": 5501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3957120678643354e-07,
+      "loss": 0.6201,
+      "step": 5502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3893175440077067e-07,
+      "loss": 0.6265,
+      "step": 5503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3829374958317432e-07,
+      "loss": 0.628,
+      "step": 5504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.376571925236353e-07,
+      "loss": 0.6349,
+      "step": 5505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3702208341171196e-07,
+      "loss": 0.5341,
+      "step": 5506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3638842243653194e-07,
+      "loss": 0.5657,
+      "step": 5507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.357562097867926e-07,
+      "loss": 0.618,
+      "step": 5508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.351254456507589e-07,
+      "loss": 0.5763,
+      "step": 5509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3449613021626563e-07,
+      "loss": 0.5488,
+      "step": 5510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3386826367071448e-07,
+      "loss": 0.5938,
+      "step": 5511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.332418462010765e-07,
+      "loss": 0.6495,
+      "step": 5512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3261687799389245e-07,
+      "loss": 0.6212,
+      "step": 5513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3199335923526956e-07,
+      "loss": 0.601,
+      "step": 5514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.313712901108838e-07,
+      "loss": 0.3181,
+      "step": 5515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3075067080598136e-07,
+      "loss": 0.596,
+      "step": 5516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.301315015053739e-07,
+      "loss": 0.6334,
+      "step": 5517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2951378239344337e-07,
+      "loss": 0.5996,
+      "step": 5518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2889751365413872e-07,
+      "loss": 0.5954,
+      "step": 5519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2828269547097815e-07,
+      "loss": 0.6424,
+      "step": 5520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2766932802704635e-07,
+      "loss": 0.6557,
+      "step": 5521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2705741150499774e-07,
+      "loss": 0.5587,
+      "step": 5522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.264469460870521e-07,
+      "loss": 0.6344,
+      "step": 5523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2583793195500072e-07,
+      "loss": 0.5572,
+      "step": 5524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.252303692901996e-07,
+      "loss": 0.6121,
+      "step": 5525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2462425827357348e-07,
+      "loss": 0.59,
+      "step": 5526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2401959908561634e-07,
+      "loss": 0.5688,
+      "step": 5527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2341639190638633e-07,
+      "loss": 0.2328,
+      "step": 5528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2281463691551255e-07,
+      "loss": 0.4908,
+      "step": 5529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.222143342921911e-07,
+      "loss": 0.5978,
+      "step": 5530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2161548421518454e-07,
+      "loss": 0.5415,
+      "step": 5531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.210180868628219e-07,
+      "loss": 0.5561,
+      "step": 5532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2042214241300255e-07,
+      "loss": 0.6116,
+      "step": 5533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1982765104319117e-07,
+      "loss": 0.6359,
+      "step": 5534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.192346129304206e-07,
+      "loss": 0.5826,
+      "step": 5535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1864302825128903e-07,
+      "loss": 0.5879,
+      "step": 5536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1805289718196499e-07,
+      "loss": 0.5986,
+      "step": 5537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.174642198981818e-07,
+      "loss": 0.6003,
+      "step": 5538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1687699657524032e-07,
+      "loss": 0.6055,
+      "step": 5539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1629122738800902e-07,
+      "loss": 0.5826,
+      "step": 5540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1570691251092281e-07,
+      "loss": 0.5501,
+      "step": 5541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1512405211798361e-07,
+      "loss": 0.6309,
+      "step": 5542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.145426463827598e-07,
+      "loss": 0.6389,
+      "step": 5543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1396269547838845e-07,
+      "loss": 0.5679,
+      "step": 5544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1338419957757085e-07,
+      "loss": 0.6219,
+      "step": 5545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1280715885257642e-07,
+      "loss": 0.62,
+      "step": 5546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1223157347524105e-07,
+      "loss": 0.6051,
+      "step": 5547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1165744361696817e-07,
+      "loss": 0.57,
+      "step": 5548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1108476944872437e-07,
+      "loss": 0.5677,
+      "step": 5549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1051355114104711e-07,
+      "loss": 0.5557,
+      "step": 5550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0994378886403812e-07,
+      "loss": 0.5806,
+      "step": 5551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0937548278736554e-07,
+      "loss": 0.2525,
+      "step": 5552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0880863308026457e-07,
+      "loss": 0.6213,
+      "step": 5553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0824323991153519e-07,
+      "loss": 0.5811,
+      "step": 5554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0767930344954603e-07,
+      "loss": 0.5909,
+      "step": 5555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0711682386222943e-07,
+      "loss": 0.5694,
+      "step": 5556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0655580131708642e-07,
+      "loss": 0.5612,
+      "step": 5557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.059962359811817e-07,
+      "loss": 0.5725,
+      "step": 5558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0543812802114805e-07,
+      "loss": 0.6448,
+      "step": 5559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0488147760318202e-07,
+      "loss": 0.6543,
+      "step": 5560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0432628489304874e-07,
+      "loss": 0.5837,
+      "step": 5561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0377255005607711e-07,
+      "loss": 0.6086,
+      "step": 5562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.032202732571641e-07,
+      "loss": 0.6313,
+      "step": 5563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0266945466076927e-07,
+      "loss": 0.6241,
+      "step": 5564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0212009443092141e-07,
+      "loss": 0.613,
+      "step": 5565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0157219273121244e-07,
+      "loss": 0.6148,
+      "step": 5566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0102574972480129e-07,
+      "loss": 0.5957,
+      "step": 5567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0048076557441222e-07,
+      "loss": 0.6278,
+      "step": 5568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.993724044233489e-08,
+      "loss": 0.5862,
+      "step": 5569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.939517449042479e-08,
+      "loss": 0.5315,
+      "step": 5570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.885456788010228e-08,
+      "loss": 0.5286,
+      "step": 5571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.831542077235356e-08,
+      "loss": 0.5877,
+      "step": 5572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.777773332773077e-08,
+      "loss": 0.2853,
+      "step": 5573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.724150570635083e-08,
+      "loss": 0.5937,
+      "step": 5574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.670673806789543e-08,
+      "loss": 0.5289,
+      "step": 5575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.617343057161166e-08,
+      "loss": 0.6049,
+      "step": 5576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.564158337631302e-08,
+      "loss": 0.5761,
+      "step": 5577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.511119664037671e-08,
+      "loss": 0.5825,
+      "step": 5578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.458227052174529e-08,
+      "loss": 0.27,
+      "step": 5579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.405480517792775e-08,
+      "loss": 0.5343,
+      "step": 5580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.352880076599568e-08,
+      "loss": 0.5138,
+      "step": 5581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.300425744258824e-08,
+      "loss": 0.5687,
+      "step": 5582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.248117536390655e-08,
+      "loss": 0.6329,
+      "step": 5583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.195955468571938e-08,
+      "loss": 0.5295,
+      "step": 5584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.143939556335856e-08,
+      "loss": 0.6149,
+      "step": 5585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.092069815172189e-08,
+      "loss": 0.6288,
+      "step": 5586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.040346260527134e-08,
+      "loss": 0.5546,
+      "step": 5587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.988768907803202e-08,
+      "loss": 0.6334,
+      "step": 5588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.937337772359667e-08,
+      "loss": 0.5671,
+      "step": 5589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.886052869512051e-08,
+      "loss": 0.6068,
+      "step": 5590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.834914214532419e-08,
+      "loss": 0.5538,
+      "step": 5591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.7839218226492e-08,
+      "loss": 0.5485,
+      "step": 5592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.7330757090473e-08,
+      "loss": 0.6093,
+      "step": 5593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.682375888868167e-08,
+      "loss": 0.5774,
+      "step": 5594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.631822377209554e-08,
+      "loss": 0.5739,
+      "step": 5595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.581415189125697e-08,
+      "loss": 0.5209,
+      "step": 5596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.531154339627257e-08,
+      "loss": 0.6059,
+      "step": 5597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.481039843681316e-08,
+      "loss": 0.6505,
+      "step": 5598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.431071716211381e-08,
+      "loss": 0.6248,
+      "step": 5599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.381249972097328e-08,
+      "loss": 0.6426,
+      "step": 5600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.331574626175509e-08,
+      "loss": 0.6913,
+      "step": 5601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.28204569323865e-08,
+      "loss": 0.5753,
+      "step": 5602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.232663188035895e-08,
+      "loss": 0.5798,
+      "step": 5603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.183427125272703e-08,
+      "loss": 0.6481,
+      "step": 5604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.134337519611068e-08,
+      "loss": 0.6143,
+      "step": 5605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.085394385669187e-08,
+      "loss": 0.5753,
+      "step": 5606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.03659773802179e-08,
+      "loss": 0.6408,
+      "step": 5607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.987947591199974e-08,
+      "loss": 0.5852,
+      "step": 5608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.939443959691151e-08,
+      "loss": 0.572,
+      "step": 5609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.891086857939045e-08,
+      "loss": 0.6002,
+      "step": 5610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.842876300343916e-08,
+      "loss": 0.5675,
+      "step": 5611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.794812301262278e-08,
+      "loss": 0.5352,
+      "step": 5612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.746894875007016e-08,
+      "loss": 0.6063,
+      "step": 5613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.699124035847328e-08,
+      "loss": 0.6172,
+      "step": 5614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.651499798008832e-08,
+      "loss": 0.5934,
+      "step": 5615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.604022175673465e-08,
+      "loss": 0.5713,
+      "step": 5616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.55669118297947e-08,
+      "loss": 0.6131,
+      "step": 5617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.509506834021463e-08,
+      "loss": 0.642,
+      "step": 5618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.462469142850425e-08,
+      "loss": 0.5574,
+      "step": 5619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.415578123473488e-08,
+      "loss": 0.626,
+      "step": 5620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.368833789854368e-08,
+      "loss": 0.6384,
+      "step": 5621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.322236155912931e-08,
+      "loss": 0.6276,
+      "step": 5622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.275785235525356e-08,
+      "loss": 0.6156,
+      "step": 5623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.229481042524189e-08,
+      "loss": 0.599,
+      "step": 5624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.183323590698344e-08,
+      "loss": 0.6525,
+      "step": 5625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.137312893792881e-08,
+      "loss": 0.6074,
+      "step": 5626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.091448965509229e-08,
+      "loss": 0.654,
+      "step": 5627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.04573181950513e-08,
+      "loss": 0.5797,
+      "step": 5628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.000161469394639e-08,
+      "loss": 0.5463,
+      "step": 5629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.954737928748067e-08,
+      "loss": 0.6174,
+      "step": 5630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.909461211091928e-08,
+      "loss": 0.597,
+      "step": 5631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.864331329909102e-08,
+      "loss": 0.5731,
+      "step": 5632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.819348298638839e-08,
+      "loss": 0.5876,
+      "step": 5633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.774512130676425e-08,
+      "loss": 0.5978,
+      "step": 5634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.729822839373567e-08,
+      "loss": 0.5309,
+      "step": 5635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.685280438038233e-08,
+      "loss": 0.6091,
+      "step": 5636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.640884939934589e-08,
+      "loss": 0.5506,
+      "step": 5637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.596636358283115e-08,
+      "loss": 0.5901,
+      "step": 5638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.552534706260438e-08,
+      "loss": 0.5588,
+      "step": 5639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.508579996999609e-08,
+      "loss": 0.6661,
+      "step": 5640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.464772243589713e-08,
+      "loss": 0.5606,
+      "step": 5641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.421111459076312e-08,
+      "loss": 0.6448,
+      "step": 5642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.377597656460898e-08,
+      "loss": 0.6684,
+      "step": 5643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.33423084870144e-08,
+      "loss": 0.5745,
+      "step": 5644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.291011048712048e-08,
+      "loss": 0.618,
+      "step": 5645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.247938269363096e-08,
+      "loss": 0.5748,
+      "step": 5646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.205012523481158e-08,
+      "loss": 0.5695,
+      "step": 5647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.162233823848951e-08,
+      "loss": 0.6555,
+      "step": 5648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.119602183205453e-08,
+      "loss": 0.5997,
+      "step": 5649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.077117614245842e-08,
+      "loss": 0.5455,
+      "step": 5650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.034780129621664e-08,
+      "loss": 0.5623,
+      "step": 5651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.992589741940336e-08,
+      "loss": 0.5903,
+      "step": 5652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.95054646376575e-08,
+      "loss": 0.6394,
+      "step": 5653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.908650307617836e-08,
+      "loss": 0.5614,
+      "step": 5654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.866901285972837e-08,
+      "loss": 0.597,
+      "step": 5655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.82529941126303e-08,
+      "loss": 0.606,
+      "step": 5656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.783844695877061e-08,
+      "loss": 0.5637,
+      "step": 5657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.742537152159555e-08,
+      "loss": 0.5766,
+      "step": 5658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.701376792411395e-08,
+      "loss": 0.5838,
+      "step": 5659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.660363628889776e-08,
+      "loss": 0.5905,
+      "step": 5660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.619497673807761e-08,
+      "loss": 0.6119,
+      "step": 5661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5787789393348944e-08,
+      "loss": 0.3123,
+      "step": 5662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5382074375966436e-08,
+      "loss": 0.6318,
+      "step": 5663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.497783180674732e-08,
+      "loss": 0.6025,
+      "step": 5664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.457506180607031e-08,
+      "loss": 0.6132,
+      "step": 5665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.417376449387501e-08,
+      "loss": 0.5999,
+      "step": 5666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.377393998966307e-08,
+      "loss": 0.5703,
+      "step": 5667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3375588412498125e-08,
+      "loss": 0.5618,
+      "step": 5668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.297870988100473e-08,
+      "loss": 0.631,
+      "step": 5669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.258330451336724e-08,
+      "loss": 0.5818,
+      "step": 5670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.218937242733313e-08,
+      "loss": 0.5988,
+      "step": 5671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1796913740211894e-08,
+      "loss": 0.5926,
+      "step": 5672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.140592856887172e-08,
+      "loss": 0.6122,
+      "step": 5673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.101641702974336e-08,
+      "loss": 0.6317,
+      "step": 5674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.062837923881903e-08,
+      "loss": 0.5967,
+      "step": 5675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.024181531165184e-08,
+      "loss": 0.5589,
+      "step": 5676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.985672536335584e-08,
+      "loss": 0.677,
+      "step": 5677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9473109508605957e-08,
+      "loss": 0.633,
+      "step": 5678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9090967861639136e-08,
+      "loss": 0.6246,
+      "step": 5679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.871030053625159e-08,
+      "loss": 0.6358,
+      "step": 5680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8331107645802624e-08,
+      "loss": 0.5476,
+      "step": 5681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7953389303210254e-08,
+      "loss": 0.6139,
+      "step": 5682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7577145620955056e-08,
+      "loss": 0.6254,
+      "step": 5683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.720237671107741e-08,
+      "loss": 0.2794,
+      "step": 5684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.682908268518027e-08,
+      "loss": 0.5477,
+      "step": 5685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6457263654424714e-08,
+      "loss": 0.5464,
+      "step": 5686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.608691972953494e-08,
+      "loss": 0.5351,
+      "step": 5687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5718051020794407e-08,
+      "loss": 0.4975,
+      "step": 5688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.535065763804802e-08,
+      "loss": 0.6237,
+      "step": 5689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4984739690701054e-08,
+      "loss": 0.6073,
+      "step": 5690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.462029728771966e-08,
+      "loss": 0.6533,
+      "step": 5691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.425733053763037e-08,
+      "loss": 0.2478,
+      "step": 5692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.389583954852006e-08,
+      "loss": 0.579,
+      "step": 5693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.353582442803705e-08,
+      "loss": 0.5378,
+      "step": 5694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.317728528338894e-08,
+      "loss": 0.5786,
+      "step": 5695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2820222221344744e-08,
+      "loss": 0.5832,
+      "step": 5696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.246463534823386e-08,
+      "loss": 0.6034,
+      "step": 5697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.211052476994493e-08,
+      "loss": 0.5881,
+      "step": 5698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.175789059192859e-08,
+      "loss": 0.5967,
+      "step": 5699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1406732919194723e-08,
+      "loss": 0.6811,
+      "step": 5700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1057051856314147e-08,
+      "loss": 0.6491,
+      "step": 5701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0708847507417435e-08,
+      "loss": 0.5862,
+      "step": 5702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0362119976196104e-08,
+      "loss": 0.5901,
+      "step": 5703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.001686936590143e-08,
+      "loss": 0.6023,
+      "step": 5704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.967309577934397e-08,
+      "loss": 0.579,
+      "step": 5705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.933079931889627e-08,
+      "loss": 0.5374,
+      "step": 5706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.898998008649013e-08,
+      "loss": 0.6115,
+      "step": 5707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8650638183617695e-08,
+      "loss": 0.5799,
+      "step": 5708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8312773711329796e-08,
+      "loss": 0.584,
+      "step": 5709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.797638677023929e-08,
+      "loss": 0.6784,
+      "step": 5710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7641477460518826e-08,
+      "loss": 0.5884,
+      "step": 5711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.730804588189918e-08,
+      "loss": 0.6645,
+      "step": 5712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6976092133672595e-08,
+      "loss": 0.5365,
+      "step": 5713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.66456163146911e-08,
+      "loss": 0.6677,
+      "step": 5714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.631661852336654e-08,
+      "loss": 0.5887,
+      "step": 5715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5989098857670547e-08,
+      "loss": 0.6449,
+      "step": 5716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5663057415133986e-08,
+      "loss": 0.6223,
+      "step": 5717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.533849429284919e-08,
+      "loss": 0.6314,
+      "step": 5718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5015409587466074e-08,
+      "loss": 0.6221,
+      "step": 5719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.469380339519601e-08,
+      "loss": 0.6595,
+      "step": 5720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.437367581180962e-08,
+      "loss": 0.599,
+      "step": 5721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4055026932636735e-08,
+      "loss": 0.5934,
+      "step": 5722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.373785685256703e-08,
+      "loss": 0.5685,
+      "step": 5723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.342216566604994e-08,
+      "loss": 0.6267,
+      "step": 5724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3107953467095276e-08,
+      "loss": 0.6087,
+      "step": 5725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.279522034927096e-08,
+      "loss": 0.6107,
+      "step": 5726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.248396640570528e-08,
+      "loss": 0.6423,
+      "step": 5727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.217419172908687e-08,
+      "loss": 0.623,
+      "step": 5728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1865896411661376e-08,
+      "loss": 0.6007,
+      "step": 5729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.155908054523649e-08,
+      "loss": 0.6585,
+      "step": 5730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.125374422117744e-08,
+      "loss": 0.5754,
+      "step": 5731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.094988753041095e-08,
+      "loss": 0.6286,
+      "step": 5732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.06475105634213e-08,
+      "loss": 0.5546,
+      "step": 5733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.034661341025258e-08,
+      "loss": 0.6414,
+      "step": 5734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.004719616050866e-08,
+      "loss": 0.5328,
+      "step": 5735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.974925890335212e-08,
+      "loss": 0.5785,
+      "step": 5736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9452801727505308e-08,
+      "loss": 0.6433,
+      "step": 5737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9157824721249837e-08,
+      "loss": 0.5443,
+      "step": 5738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8864327972426e-08,
+      "loss": 0.6127,
+      "step": 5739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8572311568433318e-08,
+      "loss": 0.5262,
+      "step": 5740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8281775596231685e-08,
+      "loss": 0.5771,
+      "step": 5741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7992720142338558e-08,
+      "loss": 0.5666,
+      "step": 5742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7705145292831746e-08,
+      "loss": 0.5771,
+      "step": 5743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7419051133346642e-08,
+      "loss": 0.5646,
+      "step": 5744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.713443774908009e-08,
+      "loss": 0.5342,
+      "step": 5745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.685130522478485e-08,
+      "loss": 0.6296,
+      "step": 5746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6569653644775705e-08,
+      "loss": 0.6325,
+      "step": 5747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6289483092925007e-08,
+      "loss": 0.6182,
+      "step": 5748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6010793652663235e-08,
+      "loss": 0.6056,
+      "step": 5749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.573358540698234e-08,
+      "loss": 0.5335,
+      "step": 5750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5457858438430183e-08,
+      "loss": 0.6149,
+      "step": 5751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.518361282911552e-08,
+      "loss": 0.6022,
+      "step": 5752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.491084866070581e-08,
+      "loss": 0.5534,
+      "step": 5753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.463956601442663e-08,
+      "loss": 0.6554,
+      "step": 5754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4369764971062248e-08,
+      "loss": 0.6501,
+      "step": 5755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.410144561095673e-08,
+      "loss": 0.656,
+      "step": 5756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3834608014012273e-08,
+      "loss": 0.5682,
+      "step": 5757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3569252259689755e-08,
+      "loss": 0.6326,
+      "step": 5758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3305378427009863e-08,
+      "loss": 0.5535,
+      "step": 5759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3042986594549177e-08,
+      "loss": 0.2645,
+      "step": 5760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2782076840446865e-08,
+      "loss": 0.56,
+      "step": 5761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2522649242397443e-08,
+      "loss": 0.561,
+      "step": 5762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2264703877656334e-08,
+      "loss": 0.5892,
+      "step": 5763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2008240823035433e-08,
+      "loss": 0.5986,
+      "step": 5764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1753260154906973e-08,
+      "loss": 0.59,
+      "step": 5765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1499761949201337e-08,
+      "loss": 0.5875,
+      "step": 5766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.124774628140702e-08,
+      "loss": 0.5939,
+      "step": 5767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.099721322657178e-08,
+      "loss": 0.5508,
+      "step": 5768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.074816285930037e-08,
+      "loss": 0.699,
+      "step": 5769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050059525375736e-08,
+      "loss": 0.6318,
+      "step": 5770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0254510483666e-08,
+      "loss": 0.5835,
+      "step": 5771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0009908622307118e-08,
+      "loss": 0.6583,
+      "step": 5772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9766789742520222e-08,
+      "loss": 0.2467,
+      "step": 5773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9525153916702956e-08,
+      "loss": 0.6462,
+      "step": 5774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.928500121681165e-08,
+      "loss": 0.5808,
+      "step": 5775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.904633171436132e-08,
+      "loss": 0.5629,
+      "step": 5776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8809145480424008e-08,
+      "loss": 0.536,
+      "step": 5777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8573442585632097e-08,
+      "loss": 0.2701,
+      "step": 5778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8339223100173886e-08,
+      "loss": 0.5819,
+      "step": 5779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8106487093798585e-08,
+      "loss": 0.551,
+      "step": 5780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7875234635810758e-08,
+      "loss": 0.5904,
+      "step": 5781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.764546579507531e-08,
+      "loss": 0.5718,
+      "step": 5782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.741718064001474e-08,
+      "loss": 0.5996,
+      "step": 5783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7190379238609666e-08,
+      "loss": 0.5726,
+      "step": 5784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6965061658398286e-08,
+      "loss": 0.5617,
+      "step": 5785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6741227966478037e-08,
+      "loss": 0.6309,
+      "step": 5786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6518878229503932e-08,
+      "loss": 0.5289,
+      "step": 5787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6298012513688566e-08,
+      "loss": 0.6006,
+      "step": 5788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607863088480377e-08,
+      "loss": 0.617,
+      "step": 5789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5860733408178398e-08,
+      "loss": 0.6564,
+      "step": 5790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5644320148699433e-08,
+      "loss": 0.6293,
+      "step": 5791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5429391170812546e-08,
+      "loss": 0.5967,
+      "step": 5792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5215946538520987e-08,
+      "loss": 0.6973,
+      "step": 5793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5003986315386132e-08,
+      "loss": 0.6454,
+      "step": 5794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4793510564526936e-08,
+      "loss": 0.5998,
+      "step": 5795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4584519348620485e-08,
+      "loss": 0.6464,
+      "step": 5796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.437701272990255e-08,
+      "loss": 0.5447,
+      "step": 5797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.417099077016537e-08,
+      "loss": 0.5799,
+      "step": 5798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3966453530759872e-08,
+      "loss": 0.5922,
+      "step": 5799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3763401072595107e-08,
+      "loss": 0.636,
+      "step": 5800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3561833456138262e-08,
+      "loss": 0.5942,
+      "step": 5801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3361750741412437e-08,
+      "loss": 0.7077,
+      "step": 5802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3163152988000527e-08,
+      "loss": 0.5314,
+      "step": 5803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2966040255042444e-08,
+      "loss": 0.2603,
+      "step": 5804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2770412601236237e-08,
+      "loss": 0.5855,
+      "step": 5805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2576270084837528e-08,
+      "loss": 0.6306,
+      "step": 5806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2383612763659513e-08,
+      "loss": 0.5891,
+      "step": 5807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.219244069507297e-08,
+      "loss": 0.6403,
+      "step": 5808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2002753936007362e-08,
+      "loss": 0.5733,
+      "step": 5809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.181455254294861e-08,
+      "loss": 0.5321,
+      "step": 5810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1627836571940775e-08,
+      "loss": 0.659,
+      "step": 5811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1442606078585494e-08,
+      "loss": 0.6166,
+      "step": 5812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.125886111804253e-08,
+      "loss": 0.5947,
+      "step": 5813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1076601745029225e-08,
+      "loss": 0.6115,
+      "step": 5814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0895828013819942e-08,
+      "loss": 0.5908,
+      "step": 5815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.071653997824662e-08,
+      "loss": 0.5328,
+      "step": 5816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0538737691699884e-08,
+      "loss": 0.6174,
+      "step": 5817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0362421207126272e-08,
+      "loss": 0.6213,
+      "step": 5818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0187590577031003e-08,
+      "loss": 0.5975,
+      "step": 5819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0014245853476878e-08,
+      "loss": 0.6256,
+      "step": 5820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.84238708808316e-09,
+      "loss": 0.6142,
+      "step": 5821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.672014332028357e-09,
+      "loss": 0.6149,
+      "step": 5822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.503127636046883e-09,
+      "loss": 0.5736,
+      "step": 5823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.335727050430731e-09,
+      "loss": 0.6232,
+      "step": 5824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.169812625030583e-09,
+      "loss": 0.5727,
+      "step": 5825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.005384409253582e-09,
+      "loss": 0.6565,
+      "step": 5826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.842442452063894e-09,
+      "loss": 0.5663,
+      "step": 5827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.680986801984925e-09,
+      "loss": 0.5358,
+      "step": 5828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.521017507094887e-09,
+      "loss": 0.5628,
+      "step": 5829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.362534615031226e-09,
+      "loss": 0.5616,
+      "step": 5830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.205538172988415e-09,
+      "loss": 0.5968,
+      "step": 5831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.050028227718498e-09,
+      "loss": 0.5832,
+      "step": 5832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.896004825529436e-09,
+      "loss": 0.5684,
+      "step": 5833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.743468012288424e-09,
+      "loss": 0.5985,
+      "step": 5834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.59241783341913e-09,
+      "loss": 0.5748,
+      "step": 5835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.442854333901684e-09,
+      "loss": 0.6057,
+      "step": 5836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.294777558274902e-09,
+      "loss": 0.6042,
+      "step": 5837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.148187550634067e-09,
+      "loss": 0.5755,
+      "step": 5838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.003084354632595e-09,
+      "loss": 0.6021,
+      "step": 5839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.859468013479809e-09,
+      "loss": 0.515,
+      "step": 5840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.717338569942611e-09,
+      "loss": 0.586,
+      "step": 5841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.576696066346033e-09,
+      "loss": 0.5465,
+      "step": 5842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.43754054457213e-09,
+      "loss": 0.6104,
+      "step": 5843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.299872046058864e-09,
+      "loss": 0.6241,
+      "step": 5844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.163690611802331e-09,
+      "loss": 0.6989,
+      "step": 5845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.028996282356758e-09,
+      "loss": 0.6331,
+      "step": 5846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.895789097831173e-09,
+      "loss": 0.5688,
+      "step": 5847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.764069097893843e-09,
+      "loss": 0.6134,
+      "step": 5848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.633836321769504e-09,
+      "loss": 0.645,
+      "step": 5849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.505090808239355e-09,
+      "loss": 0.5882,
+      "step": 5850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.377832595643284e-09,
+      "loss": 0.5805,
+      "step": 5851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.252061721875979e-09,
+      "loss": 0.5603,
+      "step": 5852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.127778224391367e-09,
+      "loss": 0.5707,
+      "step": 5853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.004982140199288e-09,
+      "loss": 0.2745,
+      "step": 5854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.883673505867159e-09,
+      "loss": 0.546,
+      "step": 5855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7638523575188614e-09,
+      "loss": 0.5,
+      "step": 5856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.645518730835852e-09,
+      "loss": 0.5686,
+      "step": 5857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.52867266105661e-09,
+      "loss": 0.5068,
+      "step": 5858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.413314182976636e-09,
+      "loss": 0.5568,
+      "step": 5859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.299443330947895e-09,
+      "loss": 0.6592,
+      "step": 5860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.187060138879928e-09,
+      "loss": 0.5873,
+      "step": 5861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.076164640239854e-09,
+      "loss": 0.6045,
+      "step": 5862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9667568680495925e-09,
+      "loss": 0.5775,
+      "step": 5863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.858836854890302e-09,
+      "loss": 0.611,
+      "step": 5864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7524046328996086e-09,
+      "loss": 0.6335,
+      "step": 5865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6474602337710495e-09,
+      "loss": 0.5963,
+      "step": 5866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.544003688755737e-09,
+      "loss": 0.5801,
+      "step": 5867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4420350286629157e-09,
+      "loss": 0.5819,
+      "step": 5868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3415542838560744e-09,
+      "loss": 0.5094,
+      "step": 5869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2425614842584995e-09,
+      "loss": 0.2856,
+      "step": 5870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.145056659348278e-09,
+      "loss": 0.5651,
+      "step": 5871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0490398381616272e-09,
+      "loss": 0.6849,
+      "step": 5872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.954511049291231e-09,
+      "loss": 0.5448,
+      "step": 5873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8614703208862395e-09,
+      "loss": 0.6168,
+      "step": 5874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.769917680653378e-09,
+      "loss": 0.5506,
+      "step": 5875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6798531558552833e-09,
+      "loss": 0.5787,
+      "step": 5876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.591276773313278e-09,
+      "loss": 0.6095,
+      "step": 5877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5041885594034866e-09,
+      "loss": 0.6043,
+      "step": 5878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4185885400596076e-09,
+      "loss": 0.6053,
+      "step": 5879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3344767407729175e-09,
+      "loss": 0.6217,
+      "step": 5880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2518531865906024e-09,
+      "loss": 0.6239,
+      "step": 5881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.170717902117425e-09,
+      "loss": 0.6097,
+      "step": 5882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.091070911513504e-09,
+      "loss": 0.2632,
+      "step": 5883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0129122384976443e-09,
+      "loss": 0.5462,
+      "step": 5884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.936241906344005e-09,
+      "loss": 0.6294,
+      "step": 5885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8610599378848792e-09,
+      "loss": 0.6584,
+      "step": 5886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7873663555073585e-09,
+      "loss": 0.6066,
+      "step": 5887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7151611811577761e-09,
+      "loss": 0.6252,
+      "step": 5888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6444444363367118e-09,
+      "loss": 0.6125,
+      "step": 5889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5752161421039858e-09,
+      "loss": 0.5905,
+      "step": 5890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5074763190742192e-09,
+      "loss": 0.6335,
+      "step": 5891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4412249874201645e-09,
+      "loss": 0.5518,
+      "step": 5892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3764621668693745e-09,
+      "loss": 0.2469,
+      "step": 5893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3131878767091988e-09,
+      "loss": 0.6275,
+      "step": 5894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.251402135780122e-09,
+      "loss": 0.6336,
+      "step": 5895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1911049624829806e-09,
+      "loss": 0.5988,
+      "step": 5896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1322963747723016e-09,
+      "loss": 0.6007,
+      "step": 5897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0749763901607425e-09,
+      "loss": 0.5907,
+      "step": 5898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0191450257179824e-09,
+      "loss": 0.6111,
+      "step": 5899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.648022980701665e-10,
+      "loss": 0.2662,
+      "step": 5900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.119482233987953e-10,
+      "loss": 0.54,
+      "step": 5901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.605828174440556e-10,
+      "loss": 0.6319,
+      "step": 5902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.107060955020451e-10,
+      "loss": 0.5938,
+      "step": 5903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.623180724247725e-10,
+      "loss": 0.5974,
+      "step": 5904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.154187626223775e-10,
+      "loss": 0.6256,
+      "step": 5905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.700081800598002e-10,
+      "loss": 0.576,
+      "step": 5906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.260863382612226e-10,
+      "loss": 0.5862,
+      "step": 5907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.836532503050718e-10,
+      "loss": 0.629,
+      "step": 5908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.427089288273513e-10,
+      "loss": 0.566,
+      "step": 5909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.032533860216404e-10,
+      "loss": 0.4785,
+      "step": 5910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.652866336363193e-10,
+      "loss": 0.5331,
+      "step": 5911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.288086829778992e-10,
+      "loss": 0.6119,
+      "step": 5912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9381954490935737e-10,
+      "loss": 0.5368,
+      "step": 5913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.603192298495817e-10,
+      "loss": 0.5774,
+      "step": 5914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.283077477750363e-10,
+      "loss": 0.553,
+      "step": 5915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.977851082175409e-10,
+      "loss": 0.5823,
+      "step": 5916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6875132026760173e-10,
+      "loss": 0.5451,
+      "step": 5917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.412063925699704e-10,
+      "loss": 0.585,
+      "step": 5918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1515033332752954e-10,
+      "loss": 0.6726,
+      "step": 5919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9058315030018314e-10,
+      "loss": 0.6181,
+      "step": 5920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6750485080263556e-10,
+      "loss": 0.6782,
+      "step": 5921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4591544170772242e-10,
+      "loss": 0.608,
+      "step": 5922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2581492944474526e-10,
+      "loss": 0.6452,
+      "step": 5923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0720331999947153e-10,
+      "loss": 0.619,
+      "step": 5924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.008061891413456e-11,
+      "loss": 0.5198,
+      "step": 5925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.444683128743358e-11,
+      "loss": 0.5945,
+      "step": 5926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.030196177508885e-11,
+      "loss": 0.5327,
+      "step": 5927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.764601458928653e-11,
+      "loss": 0.5948,
+      "step": 5928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.647899349867867e-11,
+      "loss": 0.6298,
+      "step": 5929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6800901828938352e-11,
+      "loss": 0.6445,
+      "step": 5930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.861174246164943e-11,
+      "loss": 0.5626,
+      "step": 5931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1911517835971886e-11,
+      "loss": 0.5683,
+      "step": 5932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.7002299464213885e-12,
+      "loss": 0.5835,
+      "step": 5933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9778803456448347e-12,
+      "loss": 0.6365,
+      "step": 5934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.444701416448042e-13,
+      "loss": 0.5594,
+      "step": 5935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3066,
+      "step": 5936
+    },
+    {
+      "epoch": 1.0,
+      "step": 5936,
+      "total_flos": 3.1821740738451866e+17,
+      "train_loss": 0.042780797413959494,
+      "train_runtime": 16292.0269,
+      "train_samples_per_second": 46.636,
+      "train_steps_per_second": 0.364
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5936,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 3.1821740738451866e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}