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": 2.793296089385475e-08,
+      "loss": 3.4724,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-08,
+      "loss": 3.6148,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.379888268156426e-08,
+      "loss": 3.7749,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-07,
+      "loss": 3.5213,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3966480446927375e-07,
+      "loss": 1.4934,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6759776536312851e-07,
+      "loss": 3.5432,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9553072625698325e-07,
+      "loss": 3.3006,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 3.2916,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5139664804469275e-07,
+      "loss": 3.6043,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-07,
+      "loss": 3.3135,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0726256983240227e-07,
+      "loss": 3.0291,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 3.056,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.631284916201118e-07,
+      "loss": 2.8945,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.910614525139665e-07,
+      "loss": 2.4446,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1899441340782126e-07,
+      "loss": 2.3166,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 1.3902,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7486033519553073e-07,
+      "loss": 2.2109,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.027932960893855e-07,
+      "loss": 2.1309,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.307262569832403e-07,
+      "loss": 1.3602,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 1.9244,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.865921787709498e-07,
+      "loss": 1.8907,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.145251396648045e-07,
+      "loss": 1.8585,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.424581005586592e-07,
+      "loss": 1.9147,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.8394,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.983240223463688e-07,
+      "loss": 1.2114,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.262569832402236e-07,
+      "loss": 1.7009,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.541899441340783e-07,
+      "loss": 1.7512,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 1.7689,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.100558659217877e-07,
+      "loss": 1.8123,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379888268156425e-07,
+      "loss": 1.5773,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.659217877094973e-07,
+      "loss": 1.0594,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 1.5013,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.217877094972068e-07,
+      "loss": 1.5397,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497206703910615e-07,
+      "loss": 1.4376,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.776536312849163e-07,
+      "loss": 1.5784,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.497,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.033519553072626e-06,
+      "loss": 1.5072,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0614525139664806e-06,
+      "loss": 1.4975,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0893854748603353e-06,
+      "loss": 1.5198,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 1.2784,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1452513966480447e-06,
+      "loss": 1.2866,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1731843575418997e-06,
+      "loss": 1.392,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2011173184357544e-06,
+      "loss": 1.2503,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 1.3691,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2569832402234638e-06,
+      "loss": 1.3369,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2849162011173185e-06,
+      "loss": 1.2646,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3128491620111732e-06,
+      "loss": 1.3147,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 1.3156,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3687150837988828e-06,
+      "loss": 1.2892,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3966480446927375e-06,
+      "loss": 1.261,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4245810055865922e-06,
+      "loss": 1.2973,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 1.4161,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4804469273743019e-06,
+      "loss": 1.3388,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5083798882681566e-06,
+      "loss": 1.29,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5363128491620113e-06,
+      "loss": 1.2047,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 1.32,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5921787709497207e-06,
+      "loss": 1.3155,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6201117318435754e-06,
+      "loss": 1.2733,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6480446927374303e-06,
+      "loss": 1.2791,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 1.2631,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7039106145251397e-06,
+      "loss": 1.251,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7318435754189947e-06,
+      "loss": 0.9887,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7597765363128494e-06,
+      "loss": 1.211,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 1.1428,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8156424581005588e-06,
+      "loss": 1.2329,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8435754189944135e-06,
+      "loss": 1.2707,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8715083798882682e-06,
+      "loss": 1.3228,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 1.2216,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.927374301675978e-06,
+      "loss": 1.2041,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9553072625698325e-06,
+      "loss": 1.1646,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9832402234636873e-06,
+      "loss": 1.1154,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.2112,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.039106145251397e-06,
+      "loss": 0.9445,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.067039106145252e-06,
+      "loss": 1.2835,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0949720670391065e-06,
+      "loss": 1.2072,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 1.1984,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.150837988826816e-06,
+      "loss": 1.0488,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1787709497206706e-06,
+      "loss": 1.2245,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2067039106145253e-06,
+      "loss": 1.2161,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 1.2349,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2625698324022348e-06,
+      "loss": 1.1992,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2905027932960895e-06,
+      "loss": 0.8595,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.318435754189944e-06,
+      "loss": 1.184,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 1.1708,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.374301675977654e-06,
+      "loss": 1.0987,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4022346368715087e-06,
+      "loss": 0.87,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4301675977653634e-06,
+      "loss": 1.1829,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 1.1074,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.486033519553073e-06,
+      "loss": 1.1392,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5139664804469276e-06,
+      "loss": 1.1762,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5418994413407823e-06,
+      "loss": 1.1539,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 1.1638,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5977653631284917e-06,
+      "loss": 1.0597,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6256983240223464e-06,
+      "loss": 0.93,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6536312849162015e-06,
+      "loss": 1.1677,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 1.1712,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.709497206703911e-06,
+      "loss": 1.0182,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7374301675977656e-06,
+      "loss": 1.2089,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7653631284916204e-06,
+      "loss": 1.1234,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 0.7962,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8212290502793298e-06,
+      "loss": 1.1751,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8491620111731845e-06,
+      "loss": 1.0022,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.877094972067039e-06,
+      "loss": 1.0735,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 1.1093,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9329608938547486e-06,
+      "loss": 1.0696,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9608938547486037e-06,
+      "loss": 1.1346,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9888268156424584e-06,
+      "loss": 1.1298,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 1.102,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.044692737430168e-06,
+      "loss": 1.1117,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0726256983240226e-06,
+      "loss": 1.1467,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1005586592178773e-06,
+      "loss": 1.0688,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 1.1017,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1564245810055867e-06,
+      "loss": 1.1244,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1843575418994414e-06,
+      "loss": 1.057,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2122905027932965e-06,
+      "loss": 1.0514,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 1.1198,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.268156424581006e-06,
+      "loss": 1.0747,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2960893854748607e-06,
+      "loss": 1.139,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3240223463687154e-06,
+      "loss": 1.1518,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 1.1098,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3798882681564248e-06,
+      "loss": 1.0476,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4078212290502795e-06,
+      "loss": 1.0687,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.435754189944134e-06,
+      "loss": 0.9849,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 1.0658,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4916201117318436e-06,
+      "loss": 1.1124,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5195530726256988e-06,
+      "loss": 1.0805,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.547486033519553e-06,
+      "loss": 0.9558,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 1.1666,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.603351955307263e-06,
+      "loss": 1.0848,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6312849162011176e-06,
+      "loss": 1.0848,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6592178770949723e-06,
+      "loss": 1.1992,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 1.0936,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.715083798882682e-06,
+      "loss": 1.0449,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7430167597765364e-06,
+      "loss": 1.1093,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7709497206703915e-06,
+      "loss": 1.0272,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 1.1063,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.826815642458101e-06,
+      "loss": 1.1045,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.854748603351956e-06,
+      "loss": 1.0922,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88268156424581e-06,
+      "loss": 1.0071,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 1.1656,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.93854748603352e-06,
+      "loss": 1.0565,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9664804469273745e-06,
+      "loss": 1.0761,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.994413407821229e-06,
+      "loss": 1.0255,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 1.0443,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.050279329608939e-06,
+      "loss": 1.0346,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.078212290502794e-06,
+      "loss": 1.0728,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.106145251396648e-06,
+      "loss": 1.0455,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.9913,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1620111731843575e-06,
+      "loss": 1.0221,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.189944134078213e-06,
+      "loss": 1.0357,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.217877094972068e-06,
+      "loss": 0.9742,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 1.0862,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.273743016759777e-06,
+      "loss": 1.0621,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.301675977653632e-06,
+      "loss": 1.0702,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3296089385474866e-06,
+      "loss": 1.0322,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 1.0038,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.385474860335196e-06,
+      "loss": 1.1013,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.413407821229051e-06,
+      "loss": 1.0705,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.441340782122905e-06,
+      "loss": 0.9828,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 1.0815,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.497206703910615e-06,
+      "loss": 0.9218,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5251396648044695e-06,
+      "loss": 1.0533,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.553072625698324e-06,
+      "loss": 1.1235,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 1.0156,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.608938547486034e-06,
+      "loss": 1.0302,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.636871508379888e-06,
+      "loss": 1.0419,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.664804469273743e-06,
+      "loss": 0.974,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 1.045,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7206703910614525e-06,
+      "loss": 0.9544,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.748603351955308e-06,
+      "loss": 1.0089,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.776536312849163e-06,
+      "loss": 1.1057,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.0101,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.832402234636872e-06,
+      "loss": 0.9799,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.860335195530727e-06,
+      "loss": 0.9844,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8882681564245816e-06,
+      "loss": 1.0595,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 1.0677,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944134078212291e-06,
+      "loss": 1.0438,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.972067039106146e-06,
+      "loss": 0.8428,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 1.0513,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99999962776493e-06,
+      "loss": 0.9773,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999998511059827e-06,
+      "loss": 1.1264,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999996649885027e-06,
+      "loss": 1.0104,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999994044241082e-06,
+      "loss": 0.9964,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999990694128769e-06,
+      "loss": 1.053,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999986599549086e-06,
+      "loss": 1.0525,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999817605032505e-06,
+      "loss": 0.779,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999976176992706e-06,
+      "loss": 0.7617,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999969849019113e-06,
+      "loss": 0.9807,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999962776584357e-06,
+      "loss": 1.0079,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999954959690543e-06,
+      "loss": 1.0135,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999946398340001e-06,
+      "loss": 0.9406,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999937092535278e-06,
+      "loss": 1.0482,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999927042279147e-06,
+      "loss": 1.0461,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999916247574599e-06,
+      "loss": 1.0279,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999047084248505e-06,
+      "loss": 0.9981,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999892424833337e-06,
+      "loss": 1.0503,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999879396803715e-06,
+      "loss": 1.0422,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999865624339867e-06,
+      "loss": 1.1114,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999851107445892e-06,
+      "loss": 1.0087,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999835846126113e-06,
+      "loss": 1.0775,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998198403850755e-06,
+      "loss": 0.9393,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999803090227546e-06,
+      "loss": 0.9629,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999785595658512e-06,
+      "loss": 1.0241,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999767356683183e-06,
+      "loss": 0.7411,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99974837330699e-06,
+      "loss": 1.0693,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999728645535587e-06,
+      "loss": 0.9595,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999708173374847e-06,
+      "loss": 1.0516,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99968695683087e-06,
+      "loss": 0.9955,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999664995909971e-06,
+      "loss": 0.9825,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999642290618689e-06,
+      "loss": 0.9343,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999618840963788e-06,
+      "loss": 1.0269,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9995946469522495e-06,
+      "loss": 1.0089,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999569708591278e-06,
+      "loss": 1.0171,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999544025888301e-06,
+      "loss": 1.0502,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9995175988509656e-06,
+      "loss": 0.9343,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999490427487142e-06,
+      "loss": 1.0295,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99946251180492e-06,
+      "loss": 1.0569,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999433851812614e-06,
+      "loss": 0.9898,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999404447518759e-06,
+      "loss": 0.9959,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99937429893211e-06,
+      "loss": 1.0635,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999343406061646e-06,
+      "loss": 0.9595,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999311768916566e-06,
+      "loss": 1.0121,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999279387506291e-06,
+      "loss": 1.0977,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999246261840464e-06,
+      "loss": 0.96,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999212391928948e-06,
+      "loss": 1.0575,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999177777781831e-06,
+      "loss": 0.9759,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999142419409422e-06,
+      "loss": 0.9813,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999106316822247e-06,
+      "loss": 0.9691,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999069470031058e-06,
+      "loss": 0.9574,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999031879046828e-06,
+      "loss": 1.0683,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9989935438807515e-06,
+      "loss": 0.9631,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998954464544243e-06,
+      "loss": 1.0104,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998914641048942e-06,
+      "loss": 1.0135,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998874073406705e-06,
+      "loss": 0.9775,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998832761629614e-06,
+      "loss": 0.9612,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99879070572997e-06,
+      "loss": 0.9852,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998747905720299e-06,
+      "loss": 1.0079,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9987043616133435e-06,
+      "loss": 1.0289,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998660073422073e-06,
+      "loss": 0.9866,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998615041159673e-06,
+      "loss": 1.0243,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998569264839557e-06,
+      "loss": 1.0444,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998522744475355e-06,
+      "loss": 1.0241,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99847548008092e-06,
+      "loss": 0.9639,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998427471670326e-06,
+      "loss": 1.0573,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998378719257871e-06,
+      "loss": 0.8963,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998329222858073e-06,
+      "loss": 0.9891,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982789824856695e-06,
+      "loss": 0.9664,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998227998155623e-06,
+      "loss": 0.9938,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998176269883115e-06,
+      "loss": 0.9534,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998123797683551e-06,
+      "loss": 0.9989,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998070581572555e-06,
+      "loss": 0.999,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998016621565975e-06,
+      "loss": 0.9283,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99796191767988e-06,
+      "loss": 1.0075,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99790646993056e-06,
+      "loss": 1.0066,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997850278334526e-06,
+      "loss": 0.9431,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997793342908513e-06,
+      "loss": 0.9558,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997735663669472e-06,
+      "loss": 1.0287,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997677240634583e-06,
+      "loss": 1.0317,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997618073821241e-06,
+      "loss": 1.0222,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997558163247067e-06,
+      "loss": 0.9022,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997497508929901e-06,
+      "loss": 0.9818,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997436110887805e-06,
+      "loss": 0.9587,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997373969139062e-06,
+      "loss": 0.9931,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973110837021785e-06,
+      "loss": 0.9274,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997247454595881e-06,
+      "loss": 1.0015,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997183081839115e-06,
+      "loss": 1.0831,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997117965451054e-06,
+      "loss": 0.9412,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997052105451084e-06,
+      "loss": 0.9937,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996985501858822e-06,
+      "loss": 1.0239,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996918154694099e-06,
+      "loss": 0.9979,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996850063976971e-06,
+      "loss": 0.9749,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9967812297277145e-06,
+      "loss": 1.0533,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996711651966827e-06,
+      "loss": 0.9199,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996641330715029e-06,
+      "loss": 0.8893,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996570265993261e-06,
+      "loss": 0.9291,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9964984578226835e-06,
+      "loss": 0.9044,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996425906224683e-06,
+      "loss": 1.0109,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9963526112208624e-06,
+      "loss": 0.9918,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99627857283305e-06,
+      "loss": 0.7708,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996203791083291e-06,
+      "loss": 0.9884,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996128265993856e-06,
+      "loss": 0.8928,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996051997587236e-06,
+      "loss": 0.9441,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995974985886141e-06,
+      "loss": 0.9329,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995897230913507e-06,
+      "loss": 0.7478,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9958187326924844e-06,
+      "loss": 1.0003,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995739491246453e-06,
+      "loss": 0.6759,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995659506599008e-06,
+      "loss": 0.9169,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995578778773969e-06,
+      "loss": 1.0488,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995497307795374e-06,
+      "loss": 1.0234,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995415093687485e-06,
+      "loss": 1.0343,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995332136474786e-06,
+      "loss": 1.0026,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9952484361819765e-06,
+      "loss": 0.9807,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995163992833986e-06,
+      "loss": 0.9621,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995078806455959e-06,
+      "loss": 0.9072,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994992877073262e-06,
+      "loss": 0.9904,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994906204711485e-06,
+      "loss": 0.9034,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994818789396437e-06,
+      "loss": 0.9984,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99473063115415e-06,
+      "loss": 1.0065,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9946417300108776e-06,
+      "loss": 0.9911,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99455208599309e-06,
+      "loss": 0.9261,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9944616991274855e-06,
+      "loss": 0.9476,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994370569440979e-06,
+      "loss": 0.7906,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994278696960708e-06,
+      "loss": 0.9146,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9941860817140295e-06,
+      "loss": 1.0366,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994092723728526e-06,
+      "loss": 1.0033,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993998623031997e-06,
+      "loss": 0.915,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9939037796524636e-06,
+      "loss": 0.9074,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9938081936181705e-06,
+      "loss": 0.6516,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993711864957582e-06,
+      "loss": 0.9219,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993614793699382e-06,
+      "loss": 0.9545,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993516979872479e-06,
+      "loss": 0.9581,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9934184235060005e-06,
+      "loss": 0.9234,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993319124629294e-06,
+      "loss": 0.995,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993219083271932e-06,
+      "loss": 0.6638,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993118299463703e-06,
+      "loss": 0.9139,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9930167732346206e-06,
+      "loss": 0.9413,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992914504614917e-06,
+      "loss": 1.0288,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992811493635049e-06,
+      "loss": 0.9604,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99270774032569e-06,
+      "loss": 0.9579,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9926032447177365e-06,
+      "loss": 0.9578,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992498006842307e-06,
+      "loss": 0.9453,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99239202673074e-06,
+      "loss": 0.9672,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992285304414594e-06,
+      "loss": 1.0038,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99217783992565e-06,
+      "loss": 0.9317,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992069633295912e-06,
+      "loss": 0.9958,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991960684557598e-06,
+      "loss": 0.902,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991850993743156e-06,
+      "loss": 0.6677,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991740560885249e-06,
+      "loss": 0.981,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991629386016761e-06,
+      "loss": 1.0199,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991517469170801e-06,
+      "loss": 0.9812,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991404810380695e-06,
+      "loss": 0.9809,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9912914096799934e-06,
+      "loss": 0.9814,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991177267102463e-06,
+      "loss": 1.0407,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991062382682095e-06,
+      "loss": 1.005,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990946756453101e-06,
+      "loss": 0.9468,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990830388449913e-06,
+      "loss": 0.8302,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990713278707184e-06,
+      "loss": 0.9298,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990595427259788e-06,
+      "loss": 0.9909,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99047683414282e-06,
+      "loss": 0.9313,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990357499391595e-06,
+      "loss": 0.8799,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990237423041649e-06,
+      "loss": 1.0032,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99011660512874e-06,
+      "loss": 0.9385,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989995045688847e-06,
+      "loss": 0.964,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989872744758167e-06,
+      "loss": 0.8961,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989749702373121e-06,
+      "loss": 0.9757,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98962591857035e-06,
+      "loss": 0.9803,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989501393386714e-06,
+      "loss": 1.0349,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989376126859296e-06,
+      "loss": 0.9655,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9892501190254e-06,
+      "loss": 0.9138,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989123369922547e-06,
+      "loss": 0.9746,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988995879588483e-06,
+      "loss": 0.8996,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988867648061172e-06,
+      "loss": 0.8254,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988738675378802e-06,
+      "loss": 0.9359,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988608961579777e-06,
+      "loss": 0.935,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988478506702726e-06,
+      "loss": 0.9253,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9883473107864955e-06,
+      "loss": 0.9359,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988215373870156e-06,
+      "loss": 0.9268,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988082695992994e-06,
+      "loss": 0.9757,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987949277194522e-06,
+      "loss": 0.9075,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98781511751447e-06,
+      "loss": 0.9173,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987680216992788e-06,
+      "loss": 1.0166,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987544575669647e-06,
+      "loss": 0.9637,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987408193585442e-06,
+      "loss": 0.9849,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987271070780786e-06,
+      "loss": 0.9455,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98713320729651e-06,
+      "loss": 0.9745,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986994603173669e-06,
+      "loss": 0.888,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986855258453538e-06,
+      "loss": 0.7273,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9867151731776126e-06,
+      "loss": 0.8772,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986574347387608e-06,
+      "loss": 0.964,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986432781125461e-06,
+      "loss": 1.0259,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986290474433328e-06,
+      "loss": 0.9538,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986147427353585e-06,
+      "loss": 0.9656,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986003639928831e-06,
+      "loss": 0.9199,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985859112201884e-06,
+      "loss": 0.9583,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9857138442157836e-06,
+      "loss": 0.9688,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985567836013787e-06,
+      "loss": 0.9437,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985421087639376e-06,
+      "loss": 0.9829,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985273599136248e-06,
+      "loss": 0.8504,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9851253705483245e-06,
+      "loss": 0.8677,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9849764019197465e-06,
+      "loss": 0.9296,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9848266932948745e-06,
+      "loss": 0.9781,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9846762447182896e-06,
+      "loss": 0.9394,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984525056234795e-06,
+      "loss": 0.8761,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984373127889411e-06,
+      "loss": 0.9508,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984220459727382e-06,
+      "loss": 0.8961,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98406705179417e-06,
+      "loss": 0.9304,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9839129041354576e-06,
+      "loss": 0.9175,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9837580167971476e-06,
+      "loss": 0.9199,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983602389825365e-06,
+      "loss": 0.992,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983446023266453e-06,
+      "loss": 1.001,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983288917166975e-06,
+      "loss": 0.6084,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983131071573717e-06,
+      "loss": 0.908,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982972486533682e-06,
+      "loss": 0.9426,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982813162094095e-06,
+      "loss": 0.8783,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982653098302402e-06,
+      "loss": 1.0072,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982492295206267e-06,
+      "loss": 0.895,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982330752853576e-06,
+      "loss": 0.8518,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982168471292433e-06,
+      "loss": 0.9579,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982005450571165e-06,
+      "loss": 0.9863,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981841690738317e-06,
+      "loss": 1.008,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981677191842655e-06,
+      "loss": 0.9536,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9815119539331645e-06,
+      "loss": 0.953,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981345977059051e-06,
+      "loss": 1.0166,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981179261269741e-06,
+      "loss": 0.9035,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981011806614881e-06,
+      "loss": 0.9321,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980843613144336e-06,
+      "loss": 0.8814,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980674680908192e-06,
+      "loss": 0.8619,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980505009956756e-06,
+      "loss": 0.9856,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980334600340552e-06,
+      "loss": 0.9718,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980163452110328e-06,
+      "loss": 0.9152,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97999156531705e-06,
+      "loss": 0.919,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979818940011902e-06,
+      "loss": 0.9657,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979645576246292e-06,
+      "loss": 0.9834,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979471474071843e-06,
+      "loss": 0.9452,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979296633540403e-06,
+      "loss": 0.909,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979121054704036e-06,
+      "loss": 0.9382,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978944737615028e-06,
+      "loss": 0.9826,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978767682325883e-06,
+      "loss": 0.9104,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978589888889328e-06,
+      "loss": 0.9875,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978411357358306e-06,
+      "loss": 0.9952,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978232087785982e-06,
+      "loss": 0.9459,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978052080225741e-06,
+      "loss": 0.9552,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977871334731185e-06,
+      "loss": 0.9905,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9776898513561404e-06,
+      "loss": 0.9818,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97750763015465e-06,
+      "loss": 0.8416,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977324671180976e-06,
+      "loss": 0.9155,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977140974489604e-06,
+      "loss": 0.9197,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9769565401352325e-06,
+      "loss": 0.9228,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976771368172788e-06,
+      "loss": 1.016,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9765854586574105e-06,
+      "loss": 0.8823,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976398811644462e-06,
+      "loss": 0.9224,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976211427189523e-06,
+      "loss": 0.8845,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976023305348395e-06,
+      "loss": 0.9301,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975834446177099e-06,
+      "loss": 0.8353,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975644849731874e-06,
+      "loss": 0.9011,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975454516069181e-06,
+      "loss": 0.8569,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975263445245697e-06,
+      "loss": 0.6091,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975071637318323e-06,
+      "loss": 0.9534,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9748790923441744e-06,
+      "loss": 0.8709,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9746858103805905e-06,
+      "loss": 0.9386,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974491791485129e-06,
+      "loss": 0.9586,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974297035715565e-06,
+      "loss": 0.9669,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974101543129895e-06,
+      "loss": 0.9236,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.973905313786334e-06,
+      "loss": 0.8751,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9737083477433165e-06,
+      "loss": 0.9612,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9735106450594984e-06,
+      "loss": 0.9402,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973312205793751e-06,
+      "loss": 0.9103,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9731130300051685e-06,
+      "loss": 0.8607,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9729131177530634e-06,
+      "loss": 0.6443,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972712469096965e-06,
+      "loss": 0.9337,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972511084096627e-06,
+      "loss": 0.851,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9723089628120165e-06,
+      "loss": 0.9695,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972106105303326e-06,
+      "loss": 0.868,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971902511630962e-06,
+      "loss": 0.9743,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971698181855552e-06,
+      "loss": 0.9236,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9714931160379435e-06,
+      "loss": 0.9193,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971287314239203e-06,
+      "loss": 0.9,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971080776520615e-06,
+      "loss": 0.8857,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970873502943685e-06,
+      "loss": 0.9085,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970665493570136e-06,
+      "loss": 0.9037,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970456748461911e-06,
+      "loss": 0.8825,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970247267681172e-06,
+      "loss": 0.9115,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970037051290298e-06,
+      "loss": 0.9214,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969826099351892e-06,
+      "loss": 0.9349,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969614411928771e-06,
+      "loss": 0.9979,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969401989083973e-06,
+      "loss": 0.9124,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969188830880756e-06,
+      "loss": 0.8641,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968974937382595e-06,
+      "loss": 0.9603,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9687603086531846e-06,
+      "loss": 0.8438,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968544944756441e-06,
+      "loss": 0.9452,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9683288457564934e-06,
+      "loss": 0.9837,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9681120117176965e-06,
+      "loss": 0.8724,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967894442704619e-06,
+      "loss": 0.9527,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9676761387820515e-06,
+      "loss": 0.7914,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967457100015003e-06,
+      "loss": 0.9035,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967237326468698e-06,
+      "loss": 0.9143,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967016818208585e-06,
+      "loss": 0.9255,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9667955753003276e-06,
+      "loss": 0.8771,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96657359780981e-06,
+      "loss": 0.8913,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966350885803133e-06,
+      "loss": 0.6446,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9661274393466185e-06,
+      "loss": 0.8816,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965903258506806e-06,
+      "loss": 0.926,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965678343350455e-06,
+      "loss": 0.9774,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96545269394454e-06,
+      "loss": 0.9668,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965226310356261e-06,
+      "loss": 0.885,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964999192653027e-06,
+      "loss": 0.9104,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964771340902475e-06,
+      "loss": 0.8495,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964542755172455e-06,
+      "loss": 0.8884,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964313435531036e-06,
+      "loss": 0.8915,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9640833820465086e-06,
+      "loss": 0.9747,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96385259478738e-06,
+      "loss": 0.8904,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963621073822373e-06,
+      "loss": 0.9499,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963388819220436e-06,
+      "loss": 0.8782,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9631558310507285e-06,
+      "loss": 0.9603,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962922109382633e-06,
+      "loss": 0.944,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9626876542857485e-06,
+      "loss": 0.9013,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962452465829893e-06,
+      "loss": 0.8644,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9622165440851035e-06,
+      "loss": 0.9455,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961979889121633e-06,
+      "loss": 0.9565,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961742501009956e-06,
+      "loss": 0.8856,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961504379820764e-06,
+      "loss": 0.9399,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961265525624965e-06,
+      "loss": 0.933,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961025938493688e-06,
+      "loss": 0.9728,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960785618498281e-06,
+      "loss": 0.8384,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960544565710305e-06,
+      "loss": 0.9285,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960302780201545e-06,
+      "loss": 0.8819,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960060262044001e-06,
+      "loss": 0.8973,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9598170113098916e-06,
+      "loss": 0.7928,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959573028071655e-06,
+      "loss": 0.8008,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959328312401945e-06,
+      "loss": 0.6417,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959082864373637e-06,
+      "loss": 0.9015,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958836684059821e-06,
+      "loss": 0.9534,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9585897715338075e-06,
+      "loss": 0.8511,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958342126869123e-06,
+      "loss": 0.8176,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9580937501395135e-06,
+      "loss": 0.906,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957844641418943e-06,
+      "loss": 0.8599,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957594800781594e-06,
+      "loss": 0.9122,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957344228301864e-06,
+      "loss": 0.9566,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9570929240543716e-06,
+      "loss": 0.9042,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956840888113953e-06,
+      "loss": 0.8815,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95658812055566e-06,
+      "loss": 0.7783,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956334621454764e-06,
+      "loss": 0.9304,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956080390886755e-06,
+      "loss": 0.9042,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955825428927339e-06,
+      "loss": 0.8469,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95556973565244e-06,
+      "loss": 0.9669,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955313311138202e-06,
+      "loss": 0.8506,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955056155460985e-06,
+      "loss": 0.6198,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954798268697365e-06,
+      "loss": 0.8633,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954539650924139e-06,
+      "loss": 0.9521,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954280302218321e-06,
+      "loss": 0.8273,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954020222657141e-06,
+      "loss": 0.9284,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953759412318047e-06,
+      "loss": 0.5951,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9534978712787065e-06,
+      "loss": 0.972,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953235599617002e-06,
+      "loss": 0.8768,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9529725974110366e-06,
+      "loss": 0.8431,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952708864739128e-06,
+      "loss": 0.9355,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952444401679812e-06,
+      "loss": 0.8991,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952179208311843e-06,
+      "loss": 0.93,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951913284714195e-06,
+      "loss": 0.9222,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951646630966053e-06,
+      "loss": 0.9134,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951379247146825e-06,
+      "loss": 0.8189,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951111133336135e-06,
+      "loss": 0.886,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950842289613823e-06,
+      "loss": 0.8839,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950572716059949e-06,
+      "loss": 0.9467,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950302412754788e-06,
+      "loss": 0.8857,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950031379778833e-06,
+      "loss": 0.8957,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.949759617212794e-06,
+      "loss": 0.9394,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9494871251376e-06,
+      "loss": 0.8816,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.949213903634395e-06,
+      "loss": 0.849,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948939952784539e-06,
+      "loss": 0.8965,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948665272669616e-06,
+      "loss": 0.8712,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948389863371419e-06,
+      "loss": 0.9681,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948113724971961e-06,
+      "loss": 0.8111,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947836857553476e-06,
+      "loss": 0.8434,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947559261198409e-06,
+      "loss": 0.8333,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947280935989427e-06,
+      "loss": 0.6118,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94700188200941e-06,
+      "loss": 0.8615,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946722099341457e-06,
+      "loss": 0.6255,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946441588068885e-06,
+      "loss": 0.9344,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946160348275227e-06,
+      "loss": 0.9741,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945878380044233e-06,
+      "loss": 0.9448,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945595683459868e-06,
+      "loss": 0.8666,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945312258606317e-06,
+      "loss": 0.9124,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.945028105567982e-06,
+      "loss": 0.9605,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9447432244294766e-06,
+      "loss": 0.9426,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.944457615275639e-06,
+      "loss": 0.9006,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.944171278191516e-06,
+      "loss": 0.8439,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94388421326238e-06,
+      "loss": 0.9734,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943596420573712e-06,
+      "loss": 0.9171,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943307900211215e-06,
+      "loss": 0.894,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943018652260806e-06,
+      "loss": 0.8973,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94272867680862e-06,
+      "loss": 0.904,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942437973941009e-06,
+      "loss": 0.8654,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942146543744539e-06,
+      "loss": 0.9223,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941854386305996e-06,
+      "loss": 0.916,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94156150171238e-06,
+      "loss": 0.8212,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941267890050909e-06,
+      "loss": 0.9146,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940973551409018e-06,
+      "loss": 0.8658,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940678485874356e-06,
+      "loss": 0.9081,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94038269353479e-06,
+      "loss": 0.9525,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940086174478405e-06,
+      "loss": 0.838,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939788928793499e-06,
+      "loss": 0.9029,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939490956568589e-06,
+      "loss": 0.9203,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939192257892409e-06,
+      "loss": 0.8871,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938892832853905e-06,
+      "loss": 0.8986,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938592681542244e-06,
+      "loss": 0.8471,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938291804046807e-06,
+      "loss": 0.9166,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937990200457192e-06,
+      "loss": 0.8726,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937687870863213e-06,
+      "loss": 0.8821,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937384815354901e-06,
+      "loss": 0.8521,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9370810340225e-06,
+      "loss": 0.9391,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936776526956474e-06,
+      "loss": 0.9563,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936471294247502e-06,
+      "loss": 0.9373,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936165335986477e-06,
+      "loss": 0.9013,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9358586522645114e-06,
+      "loss": 0.8824,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935551243172931e-06,
+      "loss": 0.662,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9352431088032785e-06,
+      "loss": 0.8501,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934934249247313e-06,
+      "loss": 0.9906,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93462466459701e-06,
+      "loss": 0.8335,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934314354944558e-06,
+      "loss": 0.7932,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934003320382366e-06,
+      "loss": 0.9655,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9336915610030545e-06,
+      "loss": 0.917,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9333790768994626e-06,
+      "loss": 0.8458,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.933065868164643e-06,
+      "loss": 0.8726,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932751934891867e-06,
+      "loss": 0.9718,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932437277174621e-06,
+      "loss": 0.7875,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932121895106604e-06,
+      "loss": 0.9589,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9318057887817345e-06,
+      "loss": 0.9777,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931488958294145e-06,
+      "loss": 0.8891,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931171403738183e-06,
+      "loss": 0.8348,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930853125208414e-06,
+      "loss": 0.891,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930534122799615e-06,
+      "loss": 0.8284,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930214396606784e-06,
+      "loss": 0.8688,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92989394672513e-06,
+      "loss": 0.8731,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92957277325008e-06,
+      "loss": 0.8541,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929250876277274e-06,
+      "loss": 0.849,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928928255902571e-06,
+      "loss": 0.8479,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928604912222043e-06,
+      "loss": 0.9382,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928280845331977e-06,
+      "loss": 0.8836,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927956055328877e-06,
+      "loss": 0.6684,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927630542309462e-06,
+      "loss": 0.9126,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9273043063706646e-06,
+      "loss": 0.9076,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.926977347609635e-06,
+      "loss": 0.8798,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926649666123737e-06,
+      "loss": 0.9143,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926321262010551e-06,
+      "loss": 0.898,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925992135367871e-06,
+      "loss": 0.8943,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925662286293708e-06,
+      "loss": 0.8145,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925331714886286e-06,
+      "loss": 0.7968,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925000421244045e-06,
+      "loss": 0.9193,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.924668405465642e-06,
+      "loss": 0.7885,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.924335667649946e-06,
+      "loss": 0.913,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.924002207896043e-06,
+      "loss": 0.9004,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923668026303232e-06,
+      "loss": 0.8932,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.92333312297103e-06,
+      "loss": 0.9114,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922997497999166e-06,
+      "loss": 0.8636,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922661151487586e-06,
+      "loss": 0.9138,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9223240835364485e-06,
+      "loss": 0.8715,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.92198629424613e-06,
+      "loss": 0.65,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921647783717221e-06,
+      "loss": 0.9429,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9213085520505224e-06,
+      "loss": 0.8565,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920968599347056e-06,
+      "loss": 0.8628,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920627925708055e-06,
+      "loss": 0.9867,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920286531234968e-06,
+      "loss": 0.895,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9199444160294576e-06,
+      "loss": 0.8735,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919601580193403e-06,
+      "loss": 0.8352,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919258023828895e-06,
+      "loss": 0.8455,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918913747038242e-06,
+      "loss": 0.9266,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9185687499239645e-06,
+      "loss": 0.8978,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918223032588798e-06,
+      "loss": 0.839,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917876595135695e-06,
+      "loss": 0.872,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917529437667819e-06,
+      "loss": 0.9267,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9171815602885495e-06,
+      "loss": 0.8644,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916832963101481e-06,
+      "loss": 0.8647,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91648364621042e-06,
+      "loss": 0.9038,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916133609719391e-06,
+      "loss": 0.9565,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915782853732628e-06,
+      "loss": 0.8482,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915431378354585e-06,
+      "loss": 0.8968,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915079183689925e-06,
+      "loss": 0.7482,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914726269843528e-06,
+      "loss": 0.8781,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914372636920487e-06,
+      "loss": 0.901,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914018285026111e-06,
+      "loss": 0.8392,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91366321426592e-06,
+      "loss": 0.8339,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913307424745652e-06,
+      "loss": 0.896,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912950916571256e-06,
+      "loss": 0.8778,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912593689848894e-06,
+      "loss": 0.8718,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912235744684946e-06,
+      "loss": 0.8519,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911877081186003e-06,
+      "loss": 0.868,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911517699458872e-06,
+      "loss": 0.8289,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91115759961057e-06,
+      "loss": 0.8519,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910796781748333e-06,
+      "loss": 0.845,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910435245979607e-06,
+      "loss": 0.8481,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910072992412053e-06,
+      "loss": 0.8501,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9097100211535455e-06,
+      "loss": 0.8534,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.909346332312174e-06,
+      "loss": 0.853,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90898192599624e-06,
+      "loss": 0.8457,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90861680231426e-06,
+      "loss": 0.8613,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9082509613749615e-06,
+      "loss": 0.9165,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907884403287291e-06,
+      "loss": 0.8843,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907517128160402e-06,
+      "loss": 0.8456,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907149136103667e-06,
+      "loss": 0.9244,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.906780427226667e-06,
+      "loss": 0.6555,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.906411001639202e-06,
+      "loss": 0.8811,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90604085945128e-06,
+      "loss": 0.9454,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905670000773126e-06,
+      "loss": 0.8976,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905298425715178e-06,
+      "loss": 0.912,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904926134388086e-06,
+      "loss": 0.8918,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9045531269027135e-06,
+      "loss": 0.8518,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904179403370138e-06,
+      "loss": 0.8968,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9038049639016495e-06,
+      "loss": 0.8351,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9034298086087525e-06,
+      "loss": 0.8938,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9030539376031635e-06,
+      "loss": 0.8593,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902677350996812e-06,
+      "loss": 0.8232,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902300048901841e-06,
+      "loss": 0.8774,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9019220314306066e-06,
+      "loss": 0.8557,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901543298695678e-06,
+      "loss": 0.8485,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901163850809838e-06,
+      "loss": 0.8892,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900783687886081e-06,
+      "loss": 0.8445,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900402810037614e-06,
+      "loss": 0.8012,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900021217377859e-06,
+      "loss": 0.8352,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89963891002045e-06,
+      "loss": 0.8306,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899255888079232e-06,
+      "loss": 0.9464,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898872151668267e-06,
+      "loss": 0.7965,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8984877009018254e-06,
+      "loss": 0.9382,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8981025358943916e-06,
+      "loss": 0.6208,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8977166567606635e-06,
+      "loss": 0.9226,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8973300636155515e-06,
+      "loss": 0.9331,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896942756574179e-06,
+      "loss": 0.9104,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896554735751881e-06,
+      "loss": 0.8389,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896166001264205e-06,
+      "loss": 0.9665,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.895776553226912e-06,
+      "loss": 0.9126,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.895386391755975e-06,
+      "loss": 0.8862,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894995516967579e-06,
+      "loss": 0.8776,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894603928978123e-06,
+      "loss": 0.8722,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894211627904215e-06,
+      "loss": 0.9066,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8938186138626795e-06,
+      "loss": 0.8303,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893424886970551e-06,
+      "loss": 0.8942,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8930304473450766e-06,
+      "loss": 0.8668,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892635295103715e-06,
+      "loss": 0.9715,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892239430364138e-06,
+      "loss": 0.8253,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891842853244231e-06,
+      "loss": 0.5552,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891445563862089e-06,
+      "loss": 0.8496,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891047562336018e-06,
+      "loss": 0.9073,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89064884878454e-06,
+      "loss": 0.8835,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8902494233263885e-06,
+      "loss": 0.8773,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889849286080505e-06,
+      "loss": 0.8774,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889448437166047e-06,
+      "loss": 0.8633,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889046876702382e-06,
+      "loss": 0.9261,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88864460480909e-06,
+      "loss": 0.8357,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888241621605963e-06,
+      "loss": 0.7645,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887837927213005e-06,
+      "loss": 0.896,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8874335217504305e-06,
+      "loss": 0.9539,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887028405338667e-06,
+      "loss": 0.9743,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886622578098353e-06,
+      "loss": 0.9224,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8862160401503395e-06,
+      "loss": 0.8274,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885808791615688e-06,
+      "loss": 0.8406,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885400832615673e-06,
+      "loss": 0.8209,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88499216327178e-06,
+      "loss": 0.891,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884582783705705e-06,
+      "loss": 0.9332,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8841726940393566e-06,
+      "loss": 0.918,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.883761894394855e-06,
+      "loss": 0.921,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8833503848945305e-06,
+      "loss": 0.8887,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.882938165660928e-06,
+      "loss": 0.8648,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882525236816798e-06,
+      "loss": 0.8297,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882111598485108e-06,
+      "loss": 0.9023,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881697250789035e-06,
+      "loss": 0.8324,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881282193851965e-06,
+      "loss": 0.8852,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880866427797498e-06,
+      "loss": 0.8569,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880449952749444e-06,
+      "loss": 0.8529,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880032768831825e-06,
+      "loss": 0.8342,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879614876168873e-06,
+      "loss": 0.8686,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8791962748850305e-06,
+      "loss": 0.8925,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878776965104953e-06,
+      "loss": 0.8946,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878356946953505e-06,
+      "loss": 0.8173,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877936220555765e-06,
+      "loss": 0.926,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877514786037017e-06,
+      "loss": 0.8387,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8770926435227624e-06,
+      "loss": 0.8792,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876669793138709e-06,
+      "loss": 0.9152,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876246235010775e-06,
+      "loss": 0.8952,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875821969265093e-06,
+      "loss": 0.8648,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875396996028004e-06,
+      "loss": 0.9651,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874971315426059e-06,
+      "loss": 0.8544,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874544927586022e-06,
+      "loss": 0.6234,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874117832634865e-06,
+      "loss": 0.8648,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873690030699771e-06,
+      "loss": 0.8556,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873261521908137e-06,
+      "loss": 0.8071,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872832306387566e-06,
+      "loss": 0.8787,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872402384265873e-06,
+      "loss": 0.9516,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871971755671084e-06,
+      "loss": 0.8505,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871540420731435e-06,
+      "loss": 0.9473,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871108379575374e-06,
+      "loss": 0.8402,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870675632331555e-06,
+      "loss": 0.7808,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870242179128847e-06,
+      "loss": 0.896,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8698080200963265e-06,
+      "loss": 0.9126,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86937315536328e-06,
+      "loss": 0.895,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868937585059207e-06,
+      "loss": 0.9304,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8685013093138135e-06,
+      "loss": 0.8812,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868064328257018e-06,
+      "loss": 0.9369,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867626642018948e-06,
+      "loss": 0.8737,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867188250729941e-06,
+      "loss": 0.5701,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866749154520544e-06,
+      "loss": 0.8383,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866309353521516e-06,
+      "loss": 0.879,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.865868847863825e-06,
+      "loss": 0.8731,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8654276376786456e-06,
+      "loss": 0.8266,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.864985723097368e-06,
+      "loss": 0.9988,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.864543104251587e-06,
+      "loss": 0.8883,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86409978127311e-06,
+      "loss": 0.8626,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.863655754293953e-06,
+      "loss": 0.947,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8632110234463425e-06,
+      "loss": 0.8379,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.862765588862714e-06,
+      "loss": 0.9094,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.862319450675712e-06,
+      "loss": 0.6124,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.861872609018191e-06,
+      "loss": 0.942,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8614250640232155e-06,
+      "loss": 0.9161,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86097681582406e-06,
+      "loss": 0.8796,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.860527864554205e-06,
+      "loss": 0.8739,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.860078210347345e-06,
+      "loss": 0.9017,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859627853337381e-06,
+      "loss": 0.8028,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859176793658424e-06,
+      "loss": 0.8933,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.858725031444794e-06,
+      "loss": 0.9457,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.858272566831022e-06,
+      "loss": 0.9121,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.857819399951843e-06,
+      "loss": 0.8266,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.857365530942208e-06,
+      "loss": 0.7876,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.856910959937271e-06,
+      "loss": 0.8281,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.856455687072401e-06,
+      "loss": 0.8543,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.85599971248317e-06,
+      "loss": 0.8467,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855543036305363e-06,
+      "loss": 0.8581,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855085658674973e-06,
+      "loss": 0.9105,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854627579728201e-06,
+      "loss": 0.8958,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854168799601459e-06,
+      "loss": 0.9019,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.853709318431363e-06,
+      "loss": 0.8352,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.853249136354744e-06,
+      "loss": 0.8516,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.852788253508638e-06,
+      "loss": 0.8826,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.852326670030289e-06,
+      "loss": 0.8286,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.851864386057152e-06,
+      "loss": 0.8581,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.851401401726891e-06,
+      "loss": 0.9063,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850937717177375e-06,
+      "loss": 0.8151,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850473332546686e-06,
+      "loss": 0.8593,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850008247973109e-06,
+      "loss": 0.7769,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8495424635951434e-06,
+      "loss": 0.9433,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.849075979551494e-06,
+      "loss": 0.8487,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848608795981072e-06,
+      "loss": 0.908,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848140913023002e-06,
+      "loss": 0.9146,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847672330816612e-06,
+      "loss": 0.8375,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847203049501441e-06,
+      "loss": 0.8481,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.846733069217235e-06,
+      "loss": 0.7893,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8462623901039485e-06,
+      "loss": 0.8769,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.845791012301744e-06,
+      "loss": 0.8664,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8453189359509925e-06,
+      "loss": 0.9133,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.844846161192273e-06,
+      "loss": 0.9492,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8443726881663714e-06,
+      "loss": 0.879,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8438985170142834e-06,
+      "loss": 0.8534,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84342364787721e-06,
+      "loss": 0.9283,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.842948080896563e-06,
+      "loss": 0.8404,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84247181621396e-06,
+      "loss": 0.8401,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.841994853971226e-06,
+      "loss": 0.8037,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8415171943103965e-06,
+      "loss": 0.8447,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.841038837373711e-06,
+      "loss": 0.9003,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84055978330362e-06,
+      "loss": 0.8627,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.840080032242779e-06,
+      "loss": 0.8411,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.839599584334053e-06,
+      "loss": 0.8952,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.839118439720513e-06,
+      "loss": 0.8239,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.838636598545439e-06,
+      "loss": 0.8874,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.838154060952316e-06,
+      "loss": 0.593,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.837670827084839e-06,
+      "loss": 0.9196,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.837186897086909e-06,
+      "loss": 0.8452,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836702271102636e-06,
+      "loss": 0.8866,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836216949276333e-06,
+      "loss": 0.8987,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835730931752525e-06,
+      "loss": 0.886,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835244218675942e-06,
+      "loss": 0.9188,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834756810191521e-06,
+      "loss": 0.5962,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834268706444406e-06,
+      "loss": 0.8565,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.833779907579949e-06,
+      "loss": 0.887,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.833290413743709e-06,
+      "loss": 0.9011,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.832800225081451e-06,
+      "loss": 0.5936,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.832309341739147e-06,
+      "loss": 0.8703,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.831817763862977e-06,
+      "loss": 0.7666,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.831325491599325e-06,
+      "loss": 0.7794,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830832525094786e-06,
+      "loss": 0.7958,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830338864496159e-06,
+      "loss": 0.8892,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.82984450995045e-06,
+      "loss": 0.8676,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8293494616048716e-06,
+      "loss": 0.915,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.828853719606843e-06,
+      "loss": 0.8649,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.828357284103992e-06,
+      "loss": 0.8928,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.827860155244149e-06,
+      "loss": 0.8928,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.827362333175355e-06,
+      "loss": 0.867,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8268638180458534e-06,
+      "loss": 0.8959,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.826364610004097e-06,
+      "loss": 0.8689,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.825864709198746e-06,
+      "loss": 0.9006,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.825364115778661e-06,
+      "loss": 0.8594,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.824862829892916e-06,
+      "loss": 0.9049,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.824360851690786e-06,
+      "loss": 0.9028,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.823858181321754e-06,
+      "loss": 0.8798,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.823354818935512e-06,
+      "loss": 0.9003,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.822850764681952e-06,
+      "loss": 0.8618,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.822346018711177e-06,
+      "loss": 0.8247,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.821840581173494e-06,
+      "loss": 0.84,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.821334452219417e-06,
+      "loss": 0.8717,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.820827631999664e-06,
+      "loss": 0.8376,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.820320120665159e-06,
+      "loss": 0.9137,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.819811918367036e-06,
+      "loss": 0.5496,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.819303025256629e-06,
+      "loss": 0.7805,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.818793441485482e-06,
+      "loss": 0.878,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.818283167205341e-06,
+      "loss": 0.9004,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.817772202568161e-06,
+      "loss": 0.902,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.817260547726102e-06,
+      "loss": 0.8876,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.816748202831526e-06,
+      "loss": 0.8218,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8162351680370046e-06,
+      "loss": 0.8833,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.815721443495315e-06,
+      "loss": 0.7957,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.815207029359435e-06,
+      "loss": 0.807,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814691925782554e-06,
+      "loss": 0.8914,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814176132918062e-06,
+      "loss": 0.6197,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.813659650919557e-06,
+      "loss": 0.9195,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.81314247994084e-06,
+      "loss": 0.9127,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.812624620135919e-06,
+      "loss": 0.597,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.812106071659007e-06,
+      "loss": 0.8755,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.81158683466452e-06,
+      "loss": 0.798,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811066909307082e-06,
+      "loss": 0.8168,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8105462957415204e-06,
+      "loss": 0.7959,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8100249941228676e-06,
+      "loss": 0.802,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.809503004606359e-06,
+      "loss": 0.8784,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8089803273474406e-06,
+      "loss": 0.878,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.808456962501757e-06,
+      "loss": 0.902,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8079329102251595e-06,
+      "loss": 0.9288,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.807408170673706e-06,
+      "loss": 0.7273,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.806882744003658e-06,
+      "loss": 0.902,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8063566303714806e-06,
+      "loss": 0.9175,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.805829829933843e-06,
+      "loss": 0.8375,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.805302342847622e-06,
+      "loss": 0.8343,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.804774169269896e-06,
+      "loss": 0.5878,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.804245309357949e-06,
+      "loss": 0.8686,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.803715763269269e-06,
+      "loss": 0.7779,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.803185531161548e-06,
+      "loss": 0.9165,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.802654613192683e-06,
+      "loss": 0.8442,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.802123009520777e-06,
+      "loss": 0.9037,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.801590720304132e-06,
+      "loss": 0.8475,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.80105774570126e-06,
+      "loss": 0.8759,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.800524085870874e-06,
+      "loss": 0.5512,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.79998974097189e-06,
+      "loss": 0.8618,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.79945471116343e-06,
+      "loss": 0.8514,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.798918996604821e-06,
+      "loss": 0.8518,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.798382597455591e-06,
+      "loss": 0.7735,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.797845513875474e-06,
+      "loss": 0.8719,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.797307746024407e-06,
+      "loss": 0.7788,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.79676929406253e-06,
+      "loss": 0.7885,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.796230158150188e-06,
+      "loss": 0.8859,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795690338447929e-06,
+      "loss": 0.8771,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795149835116506e-06,
+      "loss": 0.8858,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.794608648316874e-06,
+      "loss": 0.7447,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.794066778210191e-06,
+      "loss": 0.877,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.79352422495782e-06,
+      "loss": 0.8728,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.792980988721327e-06,
+      "loss": 0.8823,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.792437069662482e-06,
+      "loss": 0.899,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7918924679432564e-06,
+      "loss": 0.9371,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.791347183725827e-06,
+      "loss": 0.8637,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.790801217172571e-06,
+      "loss": 0.8347,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.790254568446073e-06,
+      "loss": 0.8946,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789707237709117e-06,
+      "loss": 0.8939,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789159225124693e-06,
+      "loss": 0.9287,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.788610530855991e-06,
+      "loss": 0.8968,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.788061155066407e-06,
+      "loss": 0.8606,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.787511097919538e-06,
+      "loss": 0.895,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.786960359579185e-06,
+      "loss": 0.7659,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7864089402093495e-06,
+      "loss": 0.8276,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.785856839974239e-06,
+      "loss": 0.5886,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.785304059038262e-06,
+      "loss": 0.8216,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78475059756603e-06,
+      "loss": 0.9041,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.784196455722358e-06,
+      "loss": 0.8442,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783641633672263e-06,
+      "loss": 0.91,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783086131580963e-06,
+      "loss": 0.8279,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78252994961388e-06,
+      "loss": 0.896,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7819730879366385e-06,
+      "loss": 0.8313,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.781415546715067e-06,
+      "loss": 0.7789,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.780857326115192e-06,
+      "loss": 0.8041,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.780298426303247e-06,
+      "loss": 0.8636,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.779738847445664e-06,
+      "loss": 0.7404,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.77917858970908e-06,
+      "loss": 0.8695,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778617653260334e-06,
+      "loss": 0.8154,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778056038266464e-06,
+      "loss": 0.917,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.777493744894713e-06,
+      "loss": 0.8098,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.776930773312526e-06,
+      "loss": 0.6467,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.776367123687548e-06,
+      "loss": 0.8375,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.775802796187628e-06,
+      "loss": 0.8223,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.775237790980816e-06,
+      "loss": 0.5533,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.774672108235364e-06,
+      "loss": 0.8629,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7741057481197246e-06,
+      "loss": 0.8543,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.773538710802553e-06,
+      "loss": 0.8699,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.772970996452708e-06,
+      "loss": 0.9237,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.772402605239247e-06,
+      "loss": 0.8285,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.771833537331429e-06,
+      "loss": 0.9097,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7712637928987175e-06,
+      "loss": 0.922,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770693372110774e-06,
+      "loss": 0.8473,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770122275137465e-06,
+      "loss": 0.8575,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7695505021488545e-06,
+      "loss": 0.8035,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.76897805331521e-06,
+      "loss": 0.9218,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.768404928807e-06,
+      "loss": 0.9376,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.767831128794895e-06,
+      "loss": 0.9047,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.767256653449765e-06,
+      "loss": 0.8488,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766681502942681e-06,
+      "loss": 0.8594,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766105677444919e-06,
+      "loss": 0.7955,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.76552917712795e-06,
+      "loss": 0.9228,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.76495200216345e-06,
+      "loss": 0.737,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.764374152723293e-06,
+      "loss": 0.876,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7637956289795596e-06,
+      "loss": 0.8352,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.763216431104523e-06,
+      "loss": 0.9681,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.762636559270665e-06,
+      "loss": 0.8486,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.762056013650663e-06,
+      "loss": 0.8342,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.761474794417396e-06,
+      "loss": 0.953,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.760892901743944e-06,
+      "loss": 0.8639,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7603103358035885e-06,
+      "loss": 0.8323,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.759727096769811e-06,
+      "loss": 0.5642,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.759143184816293e-06,
+      "loss": 0.5937,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7585586001169145e-06,
+      "loss": 0.8015,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757973342845761e-06,
+      "loss": 0.7884,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757387413177112e-06,
+      "loss": 0.8915,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756800811285454e-06,
+      "loss": 0.8458,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756213537345466e-06,
+      "loss": 0.8307,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.755625591532034e-06,
+      "loss": 0.8372,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7550369740202406e-06,
+      "loss": 0.8403,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.754447684985368e-06,
+      "loss": 0.8506,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.753857724602901e-06,
+      "loss": 0.8047,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.753267093048521e-06,
+      "loss": 0.827,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.752675790498114e-06,
+      "loss": 0.8699,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7520838171277595e-06,
+      "loss": 0.7678,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7514911731137415e-06,
+      "loss": 0.81,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750897858632542e-06,
+      "loss": 0.8628,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750303873860844e-06,
+      "loss": 0.9014,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.749709218975528e-06,
+      "loss": 0.8735,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.749113894153676e-06,
+      "loss": 0.8709,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.748517899572567e-06,
+      "loss": 0.8041,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7479212354096835e-06,
+      "loss": 0.8577,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7473239018427025e-06,
+      "loss": 0.8948,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.746725899049504e-06,
+      "loss": 0.9102,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.746127227208166e-06,
+      "loss": 0.8347,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745527886496966e-06,
+      "loss": 0.7776,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.744927877094381e-06,
+      "loss": 0.813,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.744327199179085e-06,
+      "loss": 0.6208,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.743725852929954e-06,
+      "loss": 0.8233,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7431238385260615e-06,
+      "loss": 0.8052,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.74252115614668e-06,
+      "loss": 0.8583,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.741917805971281e-06,
+      "loss": 0.8252,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.741313788179536e-06,
+      "loss": 0.8371,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7407091029513116e-06,
+      "loss": 0.8001,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7401037504666796e-06,
+      "loss": 0.8339,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739497730905904e-06,
+      "loss": 0.944,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7388910444494515e-06,
+      "loss": 0.8956,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7382836912779855e-06,
+      "loss": 0.8509,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737675671572368e-06,
+      "loss": 0.8714,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737066985513662e-06,
+      "loss": 0.846,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.736457633283124e-06,
+      "loss": 0.8123,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735847615062215e-06,
+      "loss": 0.8239,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735236931032588e-06,
+      "loss": 0.8593,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7346255813761e-06,
+      "loss": 0.8363,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.734013566274802e-06,
+      "loss": 0.8886,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733400885910946e-06,
+      "loss": 0.7992,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732787540466979e-06,
+      "loss": 0.9078,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73217353012555e-06,
+      "loss": 0.908,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731558855069503e-06,
+      "loss": 0.8044,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7309435154818816e-06,
+      "loss": 0.8313,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730327511545926e-06,
+      "loss": 0.8728,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729710843445075e-06,
+      "loss": 0.8242,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729093511362964e-06,
+      "loss": 0.6055,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.728475515483428e-06,
+      "loss": 0.8735,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.727856855990499e-06,
+      "loss": 0.8849,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.727237533068406e-06,
+      "loss": 0.7746,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7266175469015765e-06,
+      "loss": 0.925,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725996897674635e-06,
+      "loss": 0.5587,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725375585572402e-06,
+      "loss": 0.8339,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.724753610779899e-06,
+      "loss": 0.7933,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.72413097348234e-06,
+      "loss": 0.8668,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.723507673865142e-06,
+      "loss": 0.8383,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7228837121139136e-06,
+      "loss": 0.825,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7222590884144646e-06,
+      "loss": 0.9008,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7216338029528005e-06,
+      "loss": 0.8124,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.721007855915123e-06,
+      "loss": 0.8734,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720381247487832e-06,
+      "loss": 0.9534,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719753977857524e-06,
+      "loss": 0.8394,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719126047210993e-06,
+      "loss": 0.8427,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718497455735228e-06,
+      "loss": 0.8039,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717868203617417e-06,
+      "loss": 0.8569,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717238291044944e-06,
+      "loss": 0.869,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.716607718205389e-06,
+      "loss": 0.8413,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715976485286529e-06,
+      "loss": 0.7997,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715344592476337e-06,
+      "loss": 0.8257,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714712039962984e-06,
+      "loss": 0.8376,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714078827934837e-06,
+      "loss": 0.9382,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713444956580458e-06,
+      "loss": 0.8384,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712810426088607e-06,
+      "loss": 0.9207,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7121752366482385e-06,
+      "loss": 0.9152,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711539388448505e-06,
+      "loss": 0.8503,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710902881678755e-06,
+      "loss": 0.877,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710265716528533e-06,
+      "loss": 0.8077,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709627893187577e-06,
+      "loss": 0.904,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708989411845826e-06,
+      "loss": 0.757,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70835027269341e-06,
+      "loss": 0.8857,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707710475920657e-06,
+      "loss": 0.9063,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707070021718092e-06,
+      "loss": 0.8405,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706428910276435e-06,
+      "loss": 0.8634,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7057871417865994e-06,
+      "loss": 0.8569,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705144716439698e-06,
+      "loss": 0.868,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704501634427037e-06,
+      "loss": 0.8571,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703857895940117e-06,
+      "loss": 0.8268,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703213501170637e-06,
+      "loss": 0.8529,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70256845031049e-06,
+      "loss": 0.7865,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701922743551766e-06,
+      "loss": 0.8448,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701276381086745e-06,
+      "loss": 0.8136,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700629363107909e-06,
+      "loss": 0.7274,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699981689807931e-06,
+      "loss": 0.8867,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699333361379681e-06,
+      "loss": 0.8491,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698684378016223e-06,
+      "loss": 0.6313,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6980347399108176e-06,
+      "loss": 0.8347,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697384447256917e-06,
+      "loss": 0.8261,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6967335002481724e-06,
+      "loss": 0.8083,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696081899078427e-06,
+      "loss": 0.842,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695429643941722e-06,
+      "loss": 0.8544,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694776735032288e-06,
+      "loss": 0.9097,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694123172544557e-06,
+      "loss": 0.8255,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.693468956673149e-06,
+      "loss": 0.5979,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.692814087612883e-06,
+      "loss": 0.8803,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69215856555877e-06,
+      "loss": 0.8537,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691502390706018e-06,
+      "loss": 0.586,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690845563250029e-06,
+      "loss": 0.8382,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690188083386397e-06,
+      "loss": 0.9212,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689529951310911e-06,
+      "loss": 0.9138,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688871167219556e-06,
+      "loss": 0.8121,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6882117313085095e-06,
+      "loss": 0.9132,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687551643774144e-06,
+      "loss": 0.8416,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6868909048130254e-06,
+      "loss": 0.8098,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686229514621914e-06,
+      "loss": 0.8386,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685567473397763e-06,
+      "loss": 0.8664,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684904781337722e-06,
+      "loss": 0.8762,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684241438639132e-06,
+      "loss": 0.8894,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683577445499529e-06,
+      "loss": 0.7815,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682912802116641e-06,
+      "loss": 0.804,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682247508688392e-06,
+      "loss": 0.8011,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681581565412897e-06,
+      "loss": 0.8268,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680914972488469e-06,
+      "loss": 0.7452,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680247730113608e-06,
+      "loss": 0.8532,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679579838487013e-06,
+      "loss": 0.8519,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678911297807572e-06,
+      "loss": 0.8403,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678242108274371e-06,
+      "loss": 0.8346,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.677572270086686e-06,
+      "loss": 0.8208,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676901783443985e-06,
+      "loss": 0.8377,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676230648545933e-06,
+      "loss": 0.8761,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.675558865592384e-06,
+      "loss": 0.8044,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.674886434783389e-06,
+      "loss": 0.953,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.67421335631919e-06,
+      "loss": 0.8081,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.673539630400219e-06,
+      "loss": 0.8038,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.672865257227106e-06,
+      "loss": 0.8904,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.67219023700067e-06,
+      "loss": 0.7831,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.671514569921925e-06,
+      "loss": 0.8672,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.670838256192076e-06,
+      "loss": 0.8286,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.670161296012521e-06,
+      "loss": 0.8498,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.66948368958485e-06,
+      "loss": 0.7984,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.668805437110848e-06,
+      "loss": 0.8576,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.66812653879249e-06,
+      "loss": 0.8315,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.667446994831943e-06,
+      "loss": 0.8887,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.666766805431566e-06,
+      "loss": 0.9161,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.666085970793914e-06,
+      "loss": 0.8069,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6654044911217285e-06,
+      "loss": 0.8481,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.664722366617948e-06,
+      "loss": 0.8911,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6640395974857015e-06,
+      "loss": 0.9118,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.663356183928308e-06,
+      "loss": 0.8102,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.66267212614928e-06,
+      "loss": 0.5647,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.661987424352323e-06,
+      "loss": 0.9493,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.661302078741332e-06,
+      "loss": 0.8166,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.660616089520395e-06,
+      "loss": 0.8445,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.659929456893791e-06,
+      "loss": 0.9321,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.659242181065991e-06,
+      "loss": 0.8101,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.658554262241659e-06,
+      "loss": 0.7859,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.657865700625648e-06,
+      "loss": 0.8062,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.657176496423002e-06,
+      "loss": 0.8374,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.656486649838959e-06,
+      "loss": 0.8289,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.655796161078948e-06,
+      "loss": 0.8127,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6551050303485875e-06,
+      "loss": 0.8181,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.654413257853688e-06,
+      "loss": 0.8939,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.653720843800251e-06,
+      "loss": 0.8542,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.653027788394469e-06,
+      "loss": 0.8868,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.652334091842725e-06,
+      "loss": 0.8504,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.651639754351595e-06,
+      "loss": 0.8498,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.650944776127844e-06,
+      "loss": 0.6285,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.650249157378428e-06,
+      "loss": 0.7904,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.649552898310493e-06,
+      "loss": 0.5571,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.648855999131379e-06,
+      "loss": 0.8223,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6481584600486115e-06,
+      "loss": 0.8696,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.647460281269911e-06,
+      "loss": 0.8282,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.646761463003186e-06,
+      "loss": 0.885,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6460620054565385e-06,
+      "loss": 0.7993,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.645361908838255e-06,
+      "loss": 0.851,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.644661173356818e-06,
+      "loss": 0.8699,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.643959799220898e-06,
+      "loss": 0.7724,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.643257786639356e-06,
+      "loss": 0.8469,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6425551358212414e-06,
+      "loss": 0.7862,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.641851846975798e-06,
+      "loss": 0.8155,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.641147920312454e-06,
+      "loss": 0.8669,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.640443356040833e-06,
+      "loss": 0.8575,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.639738154370743e-06,
+      "loss": 0.8571,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.639032315512186e-06,
+      "loss": 0.8324,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.638325839675353e-06,
+      "loss": 0.8258,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.637618727070624e-06,
+      "loss": 0.8693,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.636910977908567e-06,
+      "loss": 0.8802,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.636202592399944e-06,
+      "loss": 0.87,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.635493570755702e-06,
+      "loss": 0.8671,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.634783913186979e-06,
+      "loss": 0.8039,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.634073619905102e-06,
+      "loss": 0.8784,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.63336269112159e-06,
+      "loss": 0.7485,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.632651127048147e-06,
+      "loss": 0.8072,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.63193892789667e-06,
+      "loss": 0.8548,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.631226093879242e-06,
+      "loss": 0.8603,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6305126252081375e-06,
+      "loss": 0.817,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.629798522095818e-06,
+      "loss": 0.9078,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.629083784754936e-06,
+      "loss": 0.9203,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.62836841339833e-06,
+      "loss": 0.8261,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.627652408239031e-06,
+      "loss": 0.8694,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626935769490256e-06,
+      "loss": 0.8897,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626218497365411e-06,
+      "loss": 0.8666,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.625500592078092e-06,
+      "loss": 0.9293,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.624782053842081e-06,
+      "loss": 0.8844,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.624062882871352e-06,
+      "loss": 0.8484,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.623343079380065e-06,
+      "loss": 0.81,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.622622643582569e-06,
+      "loss": 0.8127,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.621901575693399e-06,
+      "loss": 0.8623,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.621179875927283e-06,
+      "loss": 0.8765,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.620457544499134e-06,
+      "loss": 0.7804,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.619734581624053e-06,
+      "loss": 0.8917,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.619010987517331e-06,
+      "loss": 0.9004,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.618286762394444e-06,
+      "loss": 0.8713,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.617561906471059e-06,
+      "loss": 0.853,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.616836419963029e-06,
+      "loss": 0.8746,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.616110303086394e-06,
+      "loss": 0.8761,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.615383556057385e-06,
+      "loss": 0.8962,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.614656179092417e-06,
+      "loss": 0.8206,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.613928172408094e-06,
+      "loss": 0.8364,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6131995362212104e-06,
+      "loss": 0.8535,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.612470270748743e-06,
+      "loss": 0.8279,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.611740376207857e-06,
+      "loss": 0.7904,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.611009852815909e-06,
+      "loss": 0.8042,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.610278700790439e-06,
+      "loss": 0.8508,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.609546920349176e-06,
+      "loss": 0.8029,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.608814511710035e-06,
+      "loss": 0.9035,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6080814750911185e-06,
+      "loss": 0.8734,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.607347810710716e-06,
+      "loss": 0.8424,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.606613518787304e-06,
+      "loss": 0.6755,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.605878599539546e-06,
+      "loss": 0.7908,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.605143053186292e-06,
+      "loss": 0.6274,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.604406879946579e-06,
+      "loss": 0.8022,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6036700800396305e-06,
+      "loss": 0.7967,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.602932653684858e-06,
+      "loss": 0.8158,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.602194601101856e-06,
+      "loss": 0.9171,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.601455922510408e-06,
+      "loss": 0.8282,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.600716618130486e-06,
+      "loss": 0.8576,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5999766881822435e-06,
+      "loss": 0.8661,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5992361328860246e-06,
+      "loss": 0.8083,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.598494952462357e-06,
+      "loss": 0.8779,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.597753147131956e-06,
+      "loss": 0.9033,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.59701071711572e-06,
+      "loss": 0.9685,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.596267662634739e-06,
+      "loss": 0.8104,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.595523983910284e-06,
+      "loss": 0.8373,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5947796811638135e-06,
+      "loss": 0.9319,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.594034754616974e-06,
+      "loss": 0.894,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5932892044915924e-06,
+      "loss": 0.9098,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.592543031009687e-06,
+      "loss": 0.8546,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.591796234393458e-06,
+      "loss": 0.8084,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.591048814865295e-06,
+      "loss": 0.915,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.590300772647767e-06,
+      "loss": 0.8023,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.589552107963635e-06,
+      "loss": 0.9158,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.588802821035841e-06,
+      "loss": 0.8779,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.588052912087515e-06,
+      "loss": 0.6091,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.587302381341969e-06,
+      "loss": 0.7886,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.586551229022703e-06,
+      "loss": 0.9317,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585799455353401e-06,
+      "loss": 0.8147,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585047060557933e-06,
+      "loss": 0.937,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.584294044860352e-06,
+      "loss": 0.8343,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.583540408484899e-06,
+      "loss": 0.7956,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5827861516559955e-06,
+      "loss": 0.8427,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.582031274598252e-06,
+      "loss": 0.826,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.58127577753646e-06,
+      "loss": 0.8504,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.580519660695599e-06,
+      "loss": 0.7141,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.579762924300833e-06,
+      "loss": 0.8593,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.579005568577505e-06,
+      "loss": 0.8594,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5782475937511495e-06,
+      "loss": 0.8562,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.577489000047482e-06,
+      "loss": 0.7748,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.576729787692402e-06,
+      "loss": 0.7458,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.575969956911994e-06,
+      "loss": 0.7442,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.575209507932527e-06,
+      "loss": 0.8027,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5744484409804535e-06,
+      "loss": 0.8709,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.57368675628241e-06,
+      "loss": 0.9293,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.572924454065216e-06,
+      "loss": 0.8146,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.572161534555878e-06,
+      "loss": 0.8737,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5713979979815835e-06,
+      "loss": 0.8215,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.570633844569704e-06,
+      "loss": 0.8214,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.569869074547796e-06,
+      "loss": 0.7619,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.569103688143598e-06,
+      "loss": 0.8609,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.568337685585034e-06,
+      "loss": 0.8478,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.567571067100209e-06,
+      "loss": 0.82,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.566803832917415e-06,
+      "loss": 0.8413,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.566035983265124e-06,
+      "loss": 0.8186,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.565267518371992e-06,
+      "loss": 0.8932,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.564498438466858e-06,
+      "loss": 0.8542,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.563728743778747e-06,
+      "loss": 0.8221,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.562958434536864e-06,
+      "loss": 0.8308,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.562187510970597e-06,
+      "loss": 0.8195,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.561415973309519e-06,
+      "loss": 0.7727,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.560643821783384e-06,
+      "loss": 0.9226,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.559871056622129e-06,
+      "loss": 0.5414,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.559097678055876e-06,
+      "loss": 0.8295,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.558323686314926e-06,
+      "loss": 0.8693,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.557549081629765e-06,
+      "loss": 0.856,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.556773864231062e-06,
+      "loss": 0.8853,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.555998034349667e-06,
+      "loss": 0.8292,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.555221592216612e-06,
+      "loss": 0.8415,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.554444538063113e-06,
+      "loss": 0.7906,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.553666872120568e-06,
+      "loss": 0.832,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.552888594620556e-06,
+      "loss": 0.9198,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.552109705794838e-06,
+      "loss": 0.8107,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.551330205875359e-06,
+      "loss": 0.809,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5505500950942434e-06,
+      "loss": 0.885,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5497693736838e-06,
+      "loss": 0.8249,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.548988041876518e-06,
+      "loss": 0.8942,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.54820609990507e-06,
+      "loss": 0.8623,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.547423548002307e-06,
+      "loss": 0.8059,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.546640386401264e-06,
+      "loss": 0.8552,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.545856615335157e-06,
+      "loss": 0.891,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.545072235037386e-06,
+      "loss": 0.8262,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.544287245741527e-06,
+      "loss": 0.8766,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.543501647681341e-06,
+      "loss": 0.8537,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.542715441090772e-06,
+      "loss": 0.7701,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.54192862620394e-06,
+      "loss": 0.8473,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.54114120325515e-06,
+      "loss": 0.8521,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5403531724788884e-06,
+      "loss": 0.858,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.53956453410982e-06,
+      "loss": 0.8753,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5387752883827935e-06,
+      "loss": 0.7993,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.537985435532835e-06,
+      "loss": 0.5955,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5371949757951535e-06,
+      "loss": 0.8917,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.53640390940514e-06,
+      "loss": 0.8257,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.535612236598364e-06,
+      "loss": 0.855,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.534819957610574e-06,
+      "loss": 0.7904,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5340270726777035e-06,
+      "loss": 0.9211,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.533233582035864e-06,
+      "loss": 0.9033,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.532439485921347e-06,
+      "loss": 0.8288,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5316447845706255e-06,
+      "loss": 0.8674,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.530849478220351e-06,
+      "loss": 0.8108,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.530053567107359e-06,
+      "loss": 0.8293,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5292570514686575e-06,
+      "loss": 0.8631,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.528459931541443e-06,
+      "loss": 0.7392,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.527662207563087e-06,
+      "loss": 0.8315,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.526863879771143e-06,
+      "loss": 0.8691,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.526064948403343e-06,
+      "loss": 0.9041,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.525265413697599e-06,
+      "loss": 0.8593,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.524465275892003e-06,
+      "loss": 0.8074,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5236645352248265e-06,
+      "loss": 0.8966,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.52286319193452e-06,
+      "loss": 0.8357,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.522061246259716e-06,
+      "loss": 0.8298,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5212586984392214e-06,
+      "loss": 0.8391,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.520455548712027e-06,
+      "loss": 0.5474,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5196517973173016e-06,
+      "loss": 0.8396,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.518847444494392e-06,
+      "loss": 0.7485,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.518042490482826e-06,
+      "loss": 0.8717,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.517236935522306e-06,
+      "loss": 0.495,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.516430779852721e-06,
+      "loss": 0.8711,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.515624023714132e-06,
+      "loss": 0.8612,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.514816667346782e-06,
+      "loss": 0.8065,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.514008710991091e-06,
+      "loss": 0.9258,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.513200154887661e-06,
+      "loss": 0.9141,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.512390999277269e-06,
+      "loss": 0.8473,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5115812444008705e-06,
+      "loss": 0.8351,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.510770890499603e-06,
+      "loss": 0.7483,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5099599378147795e-06,
+      "loss": 0.8649,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.509148386587892e-06,
+      "loss": 0.8187,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.508336237060611e-06,
+      "loss": 0.7881,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5075234894747844e-06,
+      "loss": 0.8259,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.506710144072439e-06,
+      "loss": 0.8031,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5058962010957795e-06,
+      "loss": 0.7666,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.505081660787189e-06,
+      "loss": 0.8145,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.504266523389226e-06,
+      "loss": 0.8159,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.50345078914463e-06,
+      "loss": 0.8136,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.502634458296317e-06,
+      "loss": 0.9031,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.50181753108738e-06,
+      "loss": 0.7569,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.501000007761091e-06,
+      "loss": 0.8491,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.500181888560897e-06,
+      "loss": 0.8604,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.499363173730427e-06,
+      "loss": 0.8881,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.498543863513481e-06,
+      "loss": 0.8551,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.497723958154043e-06,
+      "loss": 0.8075,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.496903457896269e-06,
+      "loss": 0.7816,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.496082362984495e-06,
+      "loss": 0.7666,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.495260673663233e-06,
+      "loss": 0.8168,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.494438390177172e-06,
+      "loss": 0.8254,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.493615512771179e-06,
+      "loss": 0.9366,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.492792041690296e-06,
+      "loss": 0.7054,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4919679771797435e-06,
+      "loss": 0.8057,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.491143319484918e-06,
+      "loss": 0.9456,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4903180688513925e-06,
+      "loss": 0.8926,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.489492225524917e-06,
+      "loss": 0.8478,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.488665789751418e-06,
+      "loss": 0.8374,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.487838761776998e-06,
+      "loss": 0.784,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.487011141847937e-06,
+      "loss": 0.8008,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.486182930210688e-06,
+      "loss": 0.7987,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.485354127111884e-06,
+      "loss": 0.8682,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.484524732798333e-06,
+      "loss": 0.8182,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4836947475170175e-06,
+      "loss": 0.8628,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.482864171515099e-06,
+      "loss": 0.7512,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.482033005039912e-06,
+      "loss": 0.9203,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.481201248338967e-06,
+      "loss": 0.8328,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.480368901659953e-06,
+      "loss": 0.8043,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.479535965250732e-06,
+      "loss": 0.924,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.478702439359343e-06,
+      "loss": 0.8874,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4778683242339994e-06,
+      "loss": 0.7923,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.477033620123091e-06,
+      "loss": 0.7595,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.476198327275183e-06,
+      "loss": 0.7964,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.475362445939015e-06,
+      "loss": 0.8683,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.474525976363503e-06,
+      "loss": 0.8827,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.473688918797737e-06,
+      "loss": 0.8644,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.472851273490985e-06,
+      "loss": 0.8688,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4720130406926845e-06,
+      "loss": 0.8836,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.471174220652453e-06,
+      "loss": 0.8438,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.470334813620081e-06,
+      "loss": 0.8083,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.469494819845534e-06,
+      "loss": 0.8704,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.468654239578951e-06,
+      "loss": 0.8594,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.467813073070648e-06,
+      "loss": 0.8424,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.466971320571114e-06,
+      "loss": 0.8215,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.466128982331013e-06,
+      "loss": 0.8134,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.465286058601183e-06,
+      "loss": 0.8126,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.464442549632636e-06,
+      "loss": 0.8315,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.463598455676561e-06,
+      "loss": 0.8753,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4627537769843166e-06,
+      "loss": 0.8728,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4619085138074395e-06,
+      "loss": 0.8313,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.461062666397637e-06,
+      "loss": 0.8473,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.460216235006796e-06,
+      "loss": 0.8859,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.459369219886971e-06,
+      "loss": 0.7633,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4585216212903935e-06,
+      "loss": 0.8313,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4576734394694684e-06,
+      "loss": 0.903,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.456824674676773e-06,
+      "loss": 0.631,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4559753271650615e-06,
+      "loss": 0.8394,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.455125397187258e-06,
+      "loss": 0.8002,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.454274884996462e-06,
+      "loss": 0.8427,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.453423790845945e-06,
+      "loss": 0.8528,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.452572114989155e-06,
+      "loss": 0.8115,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.451719857679707e-06,
+      "loss": 0.8114,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.450867019171396e-06,
+      "loss": 0.8014,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.450013599718187e-06,
+      "loss": 0.5641,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.449159599574217e-06,
+      "loss": 0.8858,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.448305018993798e-06,
+      "loss": 0.7933,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.447449858231414e-06,
+      "loss": 0.899,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4465941175417196e-06,
+      "loss": 0.8849,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.445737797179547e-06,
+      "loss": 0.8974,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.444880897399896e-06,
+      "loss": 0.7888,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.444023418457943e-06,
+      "loss": 0.8641,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4431653606090326e-06,
+      "loss": 0.8695,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.442306724108686e-06,
+      "loss": 0.7809,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.441447509212595e-06,
+      "loss": 0.8538,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.440587716176622e-06,
+      "loss": 0.846,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.439727345256805e-06,
+      "loss": 0.8205,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.438866396709351e-06,
+      "loss": 0.8232,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.43800487079064e-06,
+      "loss": 0.5471,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.437142767757225e-06,
+      "loss": 0.7887,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.43628008786583e-06,
+      "loss": 0.7657,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.435416831373349e-06,
+      "loss": 0.8484,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.434552998536852e-06,
+      "loss": 0.7931,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.433688589613576e-06,
+      "loss": 0.881,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.432823604860933e-06,
+      "loss": 0.7714,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4319580445365055e-06,
+      "loss": 0.8441,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.431091908898045e-06,
+      "loss": 0.7953,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.430225198203478e-06,
+      "loss": 0.8462,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4293579127109e-06,
+      "loss": 0.5673,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.428490052678579e-06,
+      "loss": 0.8387,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.427621618364953e-06,
+      "loss": 0.8634,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.426752610028629e-06,
+      "loss": 0.8641,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.425883027928392e-06,
+      "loss": 0.7913,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.425012872323189e-06,
+      "loss": 0.7666,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4241421434721444e-06,
+      "loss": 0.5208,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.42327084163455e-06,
+      "loss": 0.8369,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.422398967069868e-06,
+      "loss": 0.8806,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.421526520037734e-06,
+      "loss": 0.8844,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.420653500797952e-06,
+      "loss": 0.9233,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.419779909610496e-06,
+      "loss": 0.5409,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.418905746735511e-06,
+      "loss": 0.7807,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.418031012433313e-06,
+      "loss": 0.8519,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4171557069643875e-06,
+      "loss": 0.8232,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.416279830589389e-06,
+      "loss": 0.8376,
+      "step": 1457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.415403383569144e-06,
+      "loss": 0.7895,
+      "step": 1458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.414526366164646e-06,
+      "loss": 0.7629,
+      "step": 1459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.413648778637064e-06,
+      "loss": 0.834,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4127706212477296e-06,
+      "loss": 0.8245,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.411891894258149e-06,
+      "loss": 0.8594,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.411012597929997e-06,
+      "loss": 0.9259,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4101327325251165e-06,
+      "loss": 0.9009,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.409252298305522e-06,
+      "loss": 0.8789,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.408371295533396e-06,
+      "loss": 0.8847,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.407489724471089e-06,
+      "loss": 0.8351,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.406607585381125e-06,
+      "loss": 0.8355,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.405724878526193e-06,
+      "loss": 0.8744,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.404841604169152e-06,
+      "loss": 0.8187,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.403957762573033e-06,
+      "loss": 0.8796,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.40307335400103e-06,
+      "loss": 0.7888,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.402188378716512e-06,
+      "loss": 0.8425,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4013028369830125e-06,
+      "loss": 0.7883,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.400416729064237e-06,
+      "loss": 0.8748,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.399530055224055e-06,
+      "loss": 0.8628,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.39864281572651e-06,
+      "loss": 0.9232,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.397755010835811e-06,
+      "loss": 0.825,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.396866640816334e-06,
+      "loss": 0.8009,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.395977705932627e-06,
+      "loss": 0.7839,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.395088206449404e-06,
+      "loss": 0.8372,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.394198142631545e-06,
+      "loss": 0.8068,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.393307514744103e-06,
+      "loss": 0.5793,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3924163230522955e-06,
+      "loss": 0.7615,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3915245678215085e-06,
+      "loss": 0.8092,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3906322493172955e-06,
+      "loss": 0.8045,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.38973936780538e-06,
+      "loss": 0.7943,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.38884592355165e-06,
+      "loss": 0.8514,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.387951916822163e-06,
+      "loss": 0.7634,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.387057347883143e-06,
+      "loss": 0.7928,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.386162217000983e-06,
+      "loss": 0.8923,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.385266524442241e-06,
+      "loss": 0.8206,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3843702704736446e-06,
+      "loss": 0.8034,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3834734553620875e-06,
+      "loss": 0.8733,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3825760793746295e-06,
+      "loss": 0.7912,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3816781427785e-06,
+      "loss": 0.9754,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.380779645841092e-06,
+      "loss": 0.8938,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.379880588829969e-06,
+      "loss": 0.8554,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.378980972012859e-06,
+      "loss": 0.8569,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.378080795657655e-06,
+      "loss": 0.7755,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.377180060032422e-06,
+      "loss": 0.8309,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3762787654053865e-06,
+      "loss": 0.7408,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.375376912044943e-06,
+      "loss": 0.8215,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.374474500219653e-06,
+      "loss": 0.7051,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3735715301982445e-06,
+      "loss": 0.566,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.372668002249611e-06,
+      "loss": 0.8536,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.371763916642812e-06,
+      "loss": 0.7822,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.370859273647074e-06,
+      "loss": 0.8446,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.369954073531788e-06,
+      "loss": 0.8916,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.369048316566512e-06,
+      "loss": 0.8542,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.368142003020971e-06,
+      "loss": 0.7328,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.367235133165052e-06,
+      "loss": 0.9501,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.366327707268812e-06,
+      "loss": 0.7791,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.36541972560247e-06,
+      "loss": 0.7299,
+      "step": 1514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.364511188436413e-06,
+      "loss": 0.8382,
+      "step": 1515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.363602096041194e-06,
+      "loss": 0.8719,
+      "step": 1516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.362692448687527e-06,
+      "loss": 0.9161,
+      "step": 1517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.361782246646296e-06,
+      "loss": 0.8498,
+      "step": 1518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.360871490188546e-06,
+      "loss": 0.9064,
+      "step": 1519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.359960179585493e-06,
+      "loss": 0.8959,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.359048315108512e-06,
+      "loss": 0.8896,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.358135897029145e-06,
+      "loss": 0.8458,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.357222925619101e-06,
+      "loss": 0.8388,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.356309401150251e-06,
+      "loss": 0.8852,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.355395323894631e-06,
+      "loss": 0.7391,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.354480694124444e-06,
+      "loss": 0.8057,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.353565512112054e-06,
+      "loss": 0.8271,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.352649778129993e-06,
+      "loss": 0.7907,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.351733492450954e-06,
+      "loss": 0.8171,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.350816655347797e-06,
+      "loss": 0.737,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.349899267093545e-06,
+      "loss": 0.8022,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.348981327961384e-06,
+      "loss": 0.8001,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.348062838224667e-06,
+      "loss": 0.7176,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.347143798156909e-06,
+      "loss": 0.7953,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.346224208031789e-06,
+      "loss": 0.8508,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.345304068123149e-06,
+      "loss": 0.585,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.344383378704997e-06,
+      "loss": 0.8024,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3434621400515025e-06,
+      "loss": 0.8573,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.342540352437001e-06,
+      "loss": 0.8235,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.341618016135986e-06,
+      "loss": 0.8966,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.340695131423123e-06,
+      "loss": 0.831,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.339771698573232e-06,
+      "loss": 0.8585,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.338847717861303e-06,
+      "loss": 0.9029,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3379231895624855e-06,
+      "loss": 0.765,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3369981139520925e-06,
+      "loss": 0.8467,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.336072491305601e-06,
+      "loss": 0.8251,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.335146321898651e-06,
+      "loss": 0.7636,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.334219606007043e-06,
+      "loss": 0.819,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.333292343906743e-06,
+      "loss": 0.8173,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.332364535873879e-06,
+      "loss": 0.8672,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.33143618218474e-06,
+      "loss": 0.7648,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.33050728311578e-06,
+      "loss": 0.7504,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.329577838943613e-06,
+      "loss": 0.9024,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.328647849945017e-06,
+      "loss": 0.796,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3277173163969304e-06,
+      "loss": 0.7812,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.326786238576457e-06,
+      "loss": 0.7812,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3258546167608584e-06,
+      "loss": 0.8041,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.324922451227562e-06,
+      "loss": 0.8795,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3239897422541555e-06,
+      "loss": 0.8714,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.323056490118388e-06,
+      "loss": 0.7397,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.322122695098171e-06,
+      "loss": 0.7594,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.321188357471577e-06,
+      "loss": 0.8614,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.320253477516841e-06,
+      "loss": 0.8714,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.31931805551236e-06,
+      "loss": 0.8104,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.31838209173669e-06,
+      "loss": 0.8685,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.317445586468551e-06,
+      "loss": 0.8407,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.316508539986822e-06,
+      "loss": 0.7952,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.315570952570546e-06,
+      "loss": 0.8233,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.314632824498923e-06,
+      "loss": 0.533,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.313694156051318e-06,
+      "loss": 0.8471,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.312754947507255e-06,
+      "loss": 0.8399,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.311815199146418e-06,
+      "loss": 0.769,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.310874911248654e-06,
+      "loss": 0.7962,
+      "step": 1573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.309934084093969e-06,
+      "loss": 0.7706,
+      "step": 1574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3089927179625304e-06,
+      "loss": 0.8153,
+      "step": 1575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.308050813134666e-06,
+      "loss": 0.7025,
+      "step": 1576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3071083698908635e-06,
+      "loss": 0.8306,
+      "step": 1577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.30616538851177e-06,
+      "loss": 0.8049,
+      "step": 1578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3052218692781964e-06,
+      "loss": 0.8519,
+      "step": 1579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.30427781247111e-06,
+      "loss": 0.5434,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.303333218371641e-06,
+      "loss": 0.8167,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.302388087261076e-06,
+      "loss": 0.8505,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.301442419420866e-06,
+      "loss": 0.8725,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.300496215132618e-06,
+      "loss": 0.8382,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2995494746781005e-06,
+      "loss": 0.9157,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.298602198339243e-06,
+      "loss": 0.8794,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.297654386398132e-06,
+      "loss": 0.8054,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.296706039137013e-06,
+      "loss": 0.7948,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2957571568382946e-06,
+      "loss": 0.8381,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.294807739784543e-06,
+      "loss": 0.8218,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.293857788258482e-06,
+      "loss": 0.8713,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.292907302542996e-06,
+      "loss": 0.8588,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.291956282921129e-06,
+      "loss": 0.8431,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.291004729676081e-06,
+      "loss": 0.8767,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.290052643091217e-06,
+      "loss": 0.8377,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.289100023450054e-06,
+      "loss": 0.7431,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.288146871036272e-06,
+      "loss": 0.8555,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.287193186133708e-06,
+      "loss": 0.8463,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.286238969026358e-06,
+      "loss": 0.6159,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2852842199983765e-06,
+      "loss": 0.8582,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.284328939334077e-06,
+      "loss": 0.8341,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.283373127317929e-06,
+      "loss": 0.8277,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2824167842345635e-06,
+      "loss": 0.8244,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2814599103687686e-06,
+      "loss": 0.7663,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.280502506005487e-06,
+      "loss": 0.8231,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.279544571429825e-06,
+      "loss": 0.8588,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.278586106927043e-06,
+      "loss": 0.575,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.277627112782561e-06,
+      "loss": 0.7372,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2766675892819545e-06,
+      "loss": 0.8097,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2757075367109594e-06,
+      "loss": 0.8161,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.274746955355468e-06,
+      "loss": 0.8354,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.273785845501529e-06,
+      "loss": 0.8297,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.272824207435351e-06,
+      "loss": 0.817,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.271862041443297e-06,
+      "loss": 0.83,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.270899347811888e-06,
+      "loss": 0.8312,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.269936126827804e-06,
+      "loss": 0.8147,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.268972378777881e-06,
+      "loss": 0.8222,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.26800810394911e-06,
+      "loss": 0.7963,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.267043302628642e-06,
+      "loss": 0.8182,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.266077975103782e-06,
+      "loss": 0.8441,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.265112121661994e-06,
+      "loss": 0.7884,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2641457425908975e-06,
+      "loss": 0.8657,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.263178838178269e-06,
+      "loss": 0.8136,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.262211408712039e-06,
+      "loss": 0.7467,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2612434544803e-06,
+      "loss": 0.8458,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.260274975771294e-06,
+      "loss": 0.7568,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.259305972873424e-06,
+      "loss": 0.7881,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.258336446075248e-06,
+      "loss": 0.8802,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.257366395665478e-06,
+      "loss": 0.7556,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2563958219329835e-06,
+      "loss": 0.7794,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.25542472516679e-06,
+      "loss": 0.8621,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.25445310565608e-06,
+      "loss": 0.7872,
+      "step": 1632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.253480963690188e-06,
+      "loss": 0.8543,
+      "step": 1633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.252508299558608e-06,
+      "loss": 0.824,
+      "step": 1634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.251535113550987e-06,
+      "loss": 0.6174,
+      "step": 1635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.250561405957128e-06,
+      "loss": 0.8011,
+      "step": 1636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.24958717706699e-06,
+      "loss": 0.7772,
+      "step": 1637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.248612427170686e-06,
+      "loss": 0.8074,
+      "step": 1638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2476371565584856e-06,
+      "loss": 0.6456,
+      "step": 1639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2466613655208125e-06,
+      "loss": 0.5453,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.245685054348245e-06,
+      "loss": 0.825,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.244708223331519e-06,
+      "loss": 0.8172,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.24373087276152e-06,
+      "loss": 0.8247,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.242753002929293e-06,
+      "loss": 0.7806,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.241774614126037e-06,
+      "loss": 0.773,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2407957066431026e-06,
+      "loss": 0.8565,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.239816280771998e-06,
+      "loss": 0.7649,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.238836336804384e-06,
+      "loss": 0.7906,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.237855875032075e-06,
+      "loss": 0.7629,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.236874895747044e-06,
+      "loss": 0.8117,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.235893399241411e-06,
+      "loss": 0.9032,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.234911385807458e-06,
+      "loss": 0.8574,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.233928855737614e-06,
+      "loss": 0.8756,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.232945809324465e-06,
+      "loss": 0.7722,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2319622468607514e-06,
+      "loss": 0.8179,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.230978168639366e-06,
+      "loss": 0.8388,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.229993574953355e-06,
+      "loss": 0.8556,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.229008466095919e-06,
+      "loss": 0.7604,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.228022842360413e-06,
+      "loss": 0.7971,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.227036704040343e-06,
+      "loss": 0.8098,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.226050051429367e-06,
+      "loss": 0.7467,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.225062884821302e-06,
+      "loss": 0.836,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.224075204510112e-06,
+      "loss": 0.8053,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.223087010789917e-06,
+      "loss": 0.8816,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.22209830395499e-06,
+      "loss": 0.7867,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.221109084299756e-06,
+      "loss": 0.5649,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.220119352118792e-06,
+      "loss": 0.8353,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2191291077068285e-06,
+      "loss": 0.8507,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.218138351358748e-06,
+      "loss": 0.8828,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.217147083369588e-06,
+      "loss": 0.8496,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2161553040345336e-06,
+      "loss": 0.8838,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2151630136489275e-06,
+      "loss": 0.7585,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.21417021250826e-06,
+      "loss": 0.5386,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.213176900908175e-06,
+      "loss": 0.8261,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.212183079144472e-06,
+      "loss": 0.834,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.211188747513096e-06,
+      "loss": 0.828,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2101939063101475e-06,
+      "loss": 0.7044,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.209198555831879e-06,
+      "loss": 0.8765,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.208202696374695e-06,
+      "loss": 0.8265,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2072063282351484e-06,
+      "loss": 0.5577,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.206209451709948e-06,
+      "loss": 0.8763,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.205212067095949e-06,
+      "loss": 0.85,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2042141746901644e-06,
+      "loss": 0.8764,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.20321577478975e-06,
+      "loss": 0.8227,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.202216867692021e-06,
+      "loss": 0.7841,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.201217453694439e-06,
+      "loss": 0.8362,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.200217533094618e-06,
+      "loss": 0.7901,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.199217106190321e-06,
+      "loss": 0.7444,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.198216173279464e-06,
+      "loss": 0.7631,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.197214734660113e-06,
+      "loss": 0.8126,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.196212790630484e-06,
+      "loss": 0.8335,
+      "step": 1691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.195210341488944e-06,
+      "loss": 0.7796,
+      "step": 1692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.194207387534011e-06,
+      "loss": 0.8459,
+      "step": 1693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1932039290643534e-06,
+      "loss": 0.8234,
+      "step": 1694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.192199966378787e-06,
+      "loss": 0.8333,
+      "step": 1695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.191195499776281e-06,
+      "loss": 0.8317,
+      "step": 1696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1901905295559534e-06,
+      "loss": 0.7908,
+      "step": 1697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.189185056017073e-06,
+      "loss": 0.8002,
+      "step": 1698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.188179079459057e-06,
+      "loss": 0.7519,
+      "step": 1699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1871726001814735e-06,
+      "loss": 0.8224,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.186165618484041e-06,
+      "loss": 0.7126,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1851581346666235e-06,
+      "loss": 0.8057,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.184150149029241e-06,
+      "loss": 0.5479,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.183141661872059e-06,
+      "loss": 0.8698,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.182132673495392e-06,
+      "loss": 0.8395,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.181123184199703e-06,
+      "loss": 0.797,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1801131942856096e-06,
+      "loss": 0.8301,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.179102704053874e-06,
+      "loss": 0.8298,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.178091713805405e-06,
+      "loss": 0.8576,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1770802238412665e-06,
+      "loss": 0.8266,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.176068234462667e-06,
+      "loss": 0.7541,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.175055745970965e-06,
+      "loss": 0.8881,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1740427586676674e-06,
+      "loss": 0.8014,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.17302927285443e-06,
+      "loss": 0.8527,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1720152888330565e-06,
+      "loss": 0.8243,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1710008069054995e-06,
+      "loss": 0.8179,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.169985827373859e-06,
+      "loss": 0.8196,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.168970350540385e-06,
+      "loss": 0.8456,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.167954376707473e-06,
+      "loss": 0.8224,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.166937906177667e-06,
+      "loss": 0.8247,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1659209392536635e-06,
+      "loss": 0.7734,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.164903476238299e-06,
+      "loss": 0.844,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.163885517434565e-06,
+      "loss": 0.8137,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.162867063145596e-06,
+      "loss": 0.7856,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1618481136746745e-06,
+      "loss": 0.7942,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.160828669325234e-06,
+      "loss": 0.7005,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.15980873040085e-06,
+      "loss": 0.8141,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1587882972052514e-06,
+      "loss": 0.756,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.157767370042308e-06,
+      "loss": 0.829,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.156745949216041e-06,
+      "loss": 0.8265,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1557240350306185e-06,
+      "loss": 0.7921,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1547016277903515e-06,
+      "loss": 0.8708,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1536787277997034e-06,
+      "loss": 0.8059,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.152655335363279e-06,
+      "loss": 0.89,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.151631450785835e-06,
+      "loss": 0.8452,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1506070743722716e-06,
+      "loss": 0.7502,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.149582206427634e-06,
+      "loss": 0.86,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.148556847257117e-06,
+      "loss": 0.7417,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.147530997166059e-06,
+      "loss": 0.7418,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.146504656459949e-06,
+      "loss": 0.8858,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.145477825444415e-06,
+      "loss": 0.8018,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.144450504425238e-06,
+      "loss": 0.8379,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.14342269370834e-06,
+      "loss": 0.8227,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.142394393599792e-06,
+      "loss": 0.8814,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.14136560440581e-06,
+      "loss": 0.5608,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.140336326432753e-06,
+      "loss": 0.7988,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.13930655998713e-06,
+      "loss": 0.7305,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.138276305375591e-06,
+      "loss": 0.8051,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.137245562904936e-06,
+      "loss": 0.9299,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.136214332882106e-06,
+      "loss": 0.8238,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.135182615614189e-06,
+      "loss": 0.8533,
+      "step": 1751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.134150411408419e-06,
+      "loss": 0.8093,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.133117720572173e-06,
+      "loss": 0.5838,
+      "step": 1753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1320845434129745e-06,
+      "loss": 0.6797,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1310508802384904e-06,
+      "loss": 0.8125,
+      "step": 1755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1300167313565366e-06,
+      "loss": 0.8201,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.128982097075066e-06,
+      "loss": 0.8743,
+      "step": 1757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.127946977702184e-06,
+      "loss": 0.7622,
+      "step": 1758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.126911373546133e-06,
+      "loss": 0.8741,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1258752849153075e-06,
+      "loss": 0.8716,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1248387121182396e-06,
+      "loss": 0.7911,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.123801655463608e-06,
+      "loss": 0.7984,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.122764115260238e-06,
+      "loss": 0.8745,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.121726091817095e-06,
+      "loss": 0.8106,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.120687585443292e-06,
+      "loss": 0.5345,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.119648596448081e-06,
+      "loss": 0.832,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.118609125140862e-06,
+      "loss": 0.8128,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.117569171831178e-06,
+      "loss": 0.8298,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.116528736828712e-06,
+      "loss": 0.8984,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.115487820443296e-06,
+      "loss": 0.809,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.114446422984901e-06,
+      "loss": 0.805,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.113404544763643e-06,
+      "loss": 0.5706,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.112362186089781e-06,
+      "loss": 0.779,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.111319347273717e-06,
+      "loss": 0.8231,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.110276028625995e-06,
+      "loss": 0.7551,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.109232230457304e-06,
+      "loss": 0.7965,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.108187953078474e-06,
+      "loss": 0.5462,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.10714319680048e-06,
+      "loss": 0.7571,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.106097961934434e-06,
+      "loss": 0.7717,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.105052248791599e-06,
+      "loss": 0.895,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.104006057683373e-06,
+      "loss": 0.8664,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.102959388921301e-06,
+      "loss": 0.8411,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.101912242817066e-06,
+      "loss": 0.8385,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.100864619682499e-06,
+      "loss": 0.8286,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.099816519829566e-06,
+      "loss": 0.8314,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.098767943570382e-06,
+      "loss": 0.796,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.097718891217198e-06,
+      "loss": 0.8095,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.09666936308241e-06,
+      "loss": 0.794,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.095619359478556e-06,
+      "loss": 0.7531,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.094568880718312e-06,
+      "loss": 0.8803,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.093517927114501e-06,
+      "loss": 0.784,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.092466498980084e-06,
+      "loss": 0.9103,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.091414596628161e-06,
+      "loss": 0.828,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.090362220371978e-06,
+      "loss": 0.7305,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.089309370524921e-06,
+      "loss": 0.7663,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.088256047400514e-06,
+      "loss": 0.7909,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.087202251312426e-06,
+      "loss": 0.8207,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.086147982574462e-06,
+      "loss": 0.8537,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.085093241500574e-06,
+      "loss": 0.7602,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.08403802840485e-06,
+      "loss": 0.891,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.082982343601519e-06,
+      "loss": 0.8527,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.081926187404953e-06,
+      "loss": 0.8029,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.080869560129661e-06,
+      "loss": 0.8148,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.079812462090295e-06,
+      "loss": 0.7417,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.078754893601646e-06,
+      "loss": 0.8641,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0776968549786455e-06,
+      "loss": 0.839,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.076638346536363e-06,
+      "loss": 0.7792,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.075579368590013e-06,
+      "loss": 0.702,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.074519921454943e-06,
+      "loss": 0.754,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.073460005446647e-06,
+      "loss": 0.8112,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.072399620880752e-06,
+      "loss": 0.703,
+      "step": 1811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.07133876807303e-06,
+      "loss": 0.7939,
+      "step": 1812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.07027744733939e-06,
+      "loss": 0.8283,
+      "step": 1813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.06921565899588e-06,
+      "loss": 0.801,
+      "step": 1814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.068153403358689e-06,
+      "loss": 0.7271,
+      "step": 1815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.067090680744143e-06,
+      "loss": 0.8249,
+      "step": 1816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.066027491468708e-06,
+      "loss": 0.8676,
+      "step": 1817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.06496383584899e-06,
+      "loss": 0.8102,
+      "step": 1818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0638997142017325e-06,
+      "loss": 0.5526,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.062835126843818e-06,
+      "loss": 0.8728,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.061770074092268e-06,
+      "loss": 0.8528,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.060704556264242e-06,
+      "loss": 0.8039,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.05963857367704e-06,
+      "loss": 0.8413,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.058572126648097e-06,
+      "loss": 0.7479,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0575052154949905e-06,
+      "loss": 0.8314,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.056437840535431e-06,
+      "loss": 0.8128,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.055370002087272e-06,
+      "loss": 0.7061,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.054301700468503e-06,
+      "loss": 0.8449,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.053232935997251e-06,
+      "loss": 0.7907,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.052163708991781e-06,
+      "loss": 0.8397,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.051094019770496e-06,
+      "loss": 0.7358,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.050023868651938e-06,
+      "loss": 0.8073,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.048953255954784e-06,
+      "loss": 0.7731,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.04788218199785e-06,
+      "loss": 0.7477,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.046810647100088e-06,
+      "loss": 0.7957,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.045738651580591e-06,
+      "loss": 0.9122,
+      "step": 1836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.044666195758583e-06,
+      "loss": 0.8652,
+      "step": 1837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.043593279953431e-06,
+      "loss": 0.8242,
+      "step": 1838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.042519904484635e-06,
+      "loss": 0.8448,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.041446069671835e-06,
+      "loss": 0.8234,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.040371775834805e-06,
+      "loss": 0.7708,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.039297023293458e-06,
+      "loss": 0.8175,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.038221812367841e-06,
+      "loss": 0.9026,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.037146143378139e-06,
+      "loss": 0.9211,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.036070016644675e-06,
+      "loss": 0.7944,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.034993432487906e-06,
+      "loss": 0.7384,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.033916391228425e-06,
+      "loss": 0.7969,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.032838893186963e-06,
+      "loss": 0.8507,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.031760938684385e-06,
+      "loss": 0.8292,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.030682528041695e-06,
+      "loss": 0.7217,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.029603661580029e-06,
+      "loss": 0.5883,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.02852433962066e-06,
+      "loss": 0.7941,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.027444562484999e-06,
+      "loss": 0.7501,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.02636433049459e-06,
+      "loss": 0.8733,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0252836439711126e-06,
+      "loss": 0.7485,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.024202503236383e-06,
+      "loss": 0.5421,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.023120908612352e-06,
+      "loss": 0.8745,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0220388604211055e-06,
+      "loss": 0.8713,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.020956358984865e-06,
+      "loss": 0.8566,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.019873404625986e-06,
+      "loss": 0.7922,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.018789997666958e-06,
+      "loss": 0.7942,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.017706138430409e-06,
+      "loss": 0.7822,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.016621827239099e-06,
+      "loss": 0.8174,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.015537064415922e-06,
+      "loss": 0.8007,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.014451850283906e-06,
+      "loss": 0.8813,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.013366185166218e-06,
+      "loss": 0.8698,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.012280069386154e-06,
+      "loss": 0.8388,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.011193503267147e-06,
+      "loss": 0.9141,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.010106487132763e-06,
+      "loss": 0.8214,
+      "step": 1869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.009019021306703e-06,
+      "loss": 0.7226,
+      "step": 1870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.007931106112802e-06,
+      "loss": 0.8471,
+      "step": 1871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.006842741875026e-06,
+      "loss": 0.7825,
+      "step": 1872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0057539289174786e-06,
+      "loss": 0.7482,
+      "step": 1873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.004664667564395e-06,
+      "loss": 0.7869,
+      "step": 1874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.003574958140144e-06,
+      "loss": 0.7756,
+      "step": 1875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.002484800969228e-06,
+      "loss": 0.8368,
+      "step": 1876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.001394196376282e-06,
+      "loss": 0.8388,
+      "step": 1877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.000303144686076e-06,
+      "loss": 0.8703,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999211646223512e-06,
+      "loss": 0.8106,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.998119701313627e-06,
+      "loss": 0.7968,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.997027310281585e-06,
+      "loss": 0.8895,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.995934473452691e-06,
+      "loss": 0.8035,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9948411911523764e-06,
+      "loss": 0.7462,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.993747463706208e-06,
+      "loss": 0.8456,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.992653291439886e-06,
+      "loss": 0.8029,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.991558674679241e-06,
+      "loss": 0.8187,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.990463613750236e-06,
+      "loss": 0.8144,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.989368108978968e-06,
+      "loss": 0.8442,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.988272160691665e-06,
+      "loss": 0.8525,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9871757692146875e-06,
+      "loss": 0.8054,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.986078934874528e-06,
+      "loss": 0.879,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.98498165799781e-06,
+      "loss": 0.8223,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.983883938911291e-06,
+      "loss": 0.5516,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9827857779418555e-06,
+      "loss": 0.7381,
+      "step": 1894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.981687175416525e-06,
+      "loss": 0.5978,
+      "step": 1895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.980588131662451e-06,
+      "loss": 0.8318,
+      "step": 1896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.979488647006912e-06,
+      "loss": 0.8511,
+      "step": 1897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.978388721777326e-06,
+      "loss": 0.7871,
+      "step": 1898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977288356301235e-06,
+      "loss": 0.5476,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.976187550906315e-06,
+      "loss": 0.8275,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.975086305920373e-06,
+      "loss": 0.7843,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.973984621671346e-06,
+      "loss": 0.8795,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9728824984873035e-06,
+      "loss": 0.8526,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.971779936696444e-06,
+      "loss": 0.7839,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.970676936627097e-06,
+      "loss": 0.7674,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.969573498607722e-06,
+      "loss": 0.757,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.968469622966912e-06,
+      "loss": 0.7342,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.967365310033385e-06,
+      "loss": 0.8547,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.966260560135996e-06,
+      "loss": 0.853,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.965155373603723e-06,
+      "loss": 0.8191,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.964049750765678e-06,
+      "loss": 0.7689,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.962943691951104e-06,
+      "loss": 0.8873,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.961837197489369e-06,
+      "loss": 0.8806,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.960730267709977e-06,
+      "loss": 0.7674,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.959622902942557e-06,
+      "loss": 0.7815,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9585151035168685e-06,
+      "loss": 0.7405,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.957406869762802e-06,
+      "loss": 0.7801,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.956298202010376e-06,
+      "loss": 0.87,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.955189100589738e-06,
+      "loss": 0.8171,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.954079565831166e-06,
+      "loss": 0.728,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.952969598065066e-06,
+      "loss": 0.7157,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.951859197621972e-06,
+      "loss": 0.851,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.950748364832549e-06,
+      "loss": 0.7457,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.94963710002759e-06,
+      "loss": 0.8053,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.948525403538017e-06,
+      "loss": 0.5596,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.947413275694878e-06,
+      "loss": 0.8695,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.946300716829352e-06,
+      "loss": 0.8372,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.945187727272747e-06,
+      "loss": 0.8653,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.944074307356496e-06,
+      "loss": 0.7723,
+      "step": 1929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.942960457412163e-06,
+      "loss": 0.8335,
+      "step": 1930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.94184617777144e-06,
+      "loss": 0.8552,
+      "step": 1931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.940731468766146e-06,
+      "loss": 0.8625,
+      "step": 1932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.939616330728227e-06,
+      "loss": 0.821,
+      "step": 1933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.938500763989759e-06,
+      "loss": 0.8566,
+      "step": 1934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.937384768882943e-06,
+      "loss": 0.7959,
+      "step": 1935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.936268345740111e-06,
+      "loss": 0.7874,
+      "step": 1936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.935151494893719e-06,
+      "loss": 0.8214,
+      "step": 1937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.934034216676352e-06,
+      "loss": 0.7406,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9329165114207225e-06,
+      "loss": 0.8174,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.93179837945967e-06,
+      "loss": 0.8031,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9306798211261595e-06,
+      "loss": 0.8069,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.929560836753285e-06,
+      "loss": 0.7312,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9284414266742675e-06,
+      "loss": 0.6896,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9273215912224534e-06,
+      "loss": 0.8949,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.926201330731317e-06,
+      "loss": 0.8125,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.925080645534457e-06,
+      "loss": 0.8314,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.923959535965601e-06,
+      "loss": 0.7583,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.922838002358602e-06,
+      "loss": 0.7925,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9217160450474395e-06,
+      "loss": 0.7911,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.920593664366218e-06,
+      "loss": 0.8083,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.91947086064917e-06,
+      "loss": 0.7775,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.918347634230653e-06,
+      "loss": 0.8681,
+      "step": 1952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.91722398544515e-06,
+      "loss": 0.841,
+      "step": 1953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.916099914627271e-06,
+      "loss": 0.7404,
+      "step": 1954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.91497542211175e-06,
+      "loss": 0.7863,
+      "step": 1955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.913850508233448e-06,
+      "loss": 0.8081,
+      "step": 1956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.91272517332735e-06,
+      "loss": 0.8376,
+      "step": 1957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.911599417728568e-06,
+      "loss": 0.7869,
+      "step": 1958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.910473241772338e-06,
+      "loss": 0.8423,
+      "step": 1959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.909346645794024e-06,
+      "loss": 0.8414,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.908219630129109e-06,
+      "loss": 0.8448,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.907092195113208e-06,
+      "loss": 0.8206,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.905964341082057e-06,
+      "loss": 0.8308,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.904836068371516e-06,
+      "loss": 0.7633,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9037073773175725e-06,
+      "loss": 0.5521,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.902578268256337e-06,
+      "loss": 0.8615,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.901448741524044e-06,
+      "loss": 0.8089,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.900318797457055e-06,
+      "loss": 0.8728,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.899188436391852e-06,
+      "loss": 0.784,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.898057658665042e-06,
+      "loss": 0.8102,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8969264646133605e-06,
+      "loss": 0.7965,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.895794854573661e-06,
+      "loss": 0.7763,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.894662828882924e-06,
+      "loss": 0.8642,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.893530387878253e-06,
+      "loss": 0.746,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8923975318968764e-06,
+      "loss": 0.8754,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.891264261276145e-06,
+      "loss": 0.7384,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.890130576353531e-06,
+      "loss": 0.8043,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.888996477466636e-06,
+      "loss": 0.8241,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.887861964953179e-06,
+      "loss": 0.8068,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.886727039151004e-06,
+      "loss": 0.8321,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.885591700398078e-06,
+      "loss": 0.7933,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.884455949032493e-06,
+      "loss": 0.795,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.883319785392461e-06,
+      "loss": 0.5208,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.882183209816318e-06,
+      "loss": 0.6516,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8810462226425236e-06,
+      "loss": 0.8656,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.879908824209658e-06,
+      "loss": 0.8581,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.878771014856425e-06,
+      "loss": 0.8538,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.877632794921651e-06,
+      "loss": 0.7533,
+      "step": 1988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.876494164744284e-06,
+      "loss": 0.5757,
+      "step": 1989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.875355124663395e-06,
+      "loss": 0.8849,
+      "step": 1990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.874215675018175e-06,
+      "loss": 0.8696,
+      "step": 1991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.87307581614794e-06,
+      "loss": 0.8173,
+      "step": 1992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.871935548392126e-06,
+      "loss": 0.8432,
+      "step": 1993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.870794872090292e-06,
+      "loss": 0.5429,
+      "step": 1994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.869653787582116e-06,
+      "loss": 0.7345,
+      "step": 1995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.868512295207401e-06,
+      "loss": 0.5745,
+      "step": 1996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8673703953060685e-06,
+      "loss": 0.8801,
+      "step": 1997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8662280882181616e-06,
+      "loss": 0.7399,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.865085374283849e-06,
+      "loss": 0.847,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.863942253843415e-06,
+      "loss": 0.7856,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.862798727237268e-06,
+      "loss": 0.5463,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.861654794805936e-06,
+      "loss": 0.8446,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8605104568900685e-06,
+      "loss": 0.8531,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8593657138304365e-06,
+      "loss": 0.8531,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.85822056596793e-06,
+      "loss": 0.8153,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.85707501364356e-06,
+      "loss": 0.7816,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.855929057198459e-06,
+      "loss": 0.7929,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.854782696973879e-06,
+      "loss": 0.5245,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.853635933311193e-06,
+      "loss": 0.7968,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.852488766551892e-06,
+      "loss": 0.8283,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.851341197037591e-06,
+      "loss": 0.8299,
+      "step": 2011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.85019322511002e-06,
+      "loss": 0.7959,
+      "step": 2012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.849044851111031e-06,
+      "loss": 0.8035,
+      "step": 2013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.847896075382599e-06,
+      "loss": 0.7732,
+      "step": 2014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.846746898266813e-06,
+      "loss": 0.8477,
+      "step": 2015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8455973201058856e-06,
+      "loss": 0.8076,
+      "step": 2016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.844447341242147e-06,
+      "loss": 0.8062,
+      "step": 2017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8432969620180485e-06,
+      "loss": 0.8611,
+      "step": 2018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.842146182776157e-06,
+      "loss": 0.7005,
+      "step": 2019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8409950038591615e-06,
+      "loss": 0.7191,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.839843425609869e-06,
+      "loss": 0.8517,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8386914483712085e-06,
+      "loss": 0.8142,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.837539072486222e-06,
+      "loss": 0.7403,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8363862982980746e-06,
+      "loss": 0.8503,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.835233126150048e-06,
+      "loss": 0.7703,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8340795563855425e-06,
+      "loss": 0.8468,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.83292558934808e-06,
+      "loss": 0.8532,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.831771225381296e-06,
+      "loss": 0.7771,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.830616464828947e-06,
+      "loss": 0.7988,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8294613080349056e-06,
+      "loss": 0.8993,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.828305755343166e-06,
+      "loss": 0.5578,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.827149807097836e-06,
+      "loss": 0.8323,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.825993463643144e-06,
+      "loss": 0.7259,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.824836725323436e-06,
+      "loss": 0.7182,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.823679592483173e-06,
+      "loss": 0.8186,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.822522065466937e-06,
+      "loss": 0.731,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.821364144619425e-06,
+      "loss": 0.8443,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.820205830285452e-06,
+      "loss": 0.8098,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8190471228099505e-06,
+      "loss": 0.8013,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.81788802253797e-06,
+      "loss": 0.7268,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.816728529814675e-06,
+      "loss": 0.8825,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8155686449853514e-06,
+      "loss": 0.7829,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.814408368395396e-06,
+      "loss": 0.8332,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.813247700390328e-06,
+      "loss": 0.7789,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8120866413157786e-06,
+      "loss": 0.8511,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8109251915174984e-06,
+      "loss": 0.8531,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8097633513413522e-06,
+      "loss": 0.7851,
+      "step": 2047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8086011211333234e-06,
+      "loss": 0.8067,
+      "step": 2048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.80743850123951e-06,
+      "loss": 0.7683,
+      "step": 2049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.806275492006126e-06,
+      "loss": 0.798,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8051120937795006e-06,
+      "loss": 0.8643,
+      "step": 2051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8039483069060824e-06,
+      "loss": 0.8506,
+      "step": 2052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8027841317324298e-06,
+      "loss": 0.8242,
+      "step": 2053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8016195686052237e-06,
+      "loss": 0.8487,
+      "step": 2054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8004546178712554e-06,
+      "loss": 0.8103,
+      "step": 2055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7992892798774338e-06,
+      "loss": 0.9031,
+      "step": 2056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.798123554970783e-06,
+      "loss": 0.814,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.79695744349844e-06,
+      "loss": 0.8572,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7957909458076614e-06,
+      "loss": 0.8143,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7946240622458135e-06,
+      "loss": 0.8836,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.793456793160384e-06,
+      "loss": 0.7976,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7922891388989674e-06,
+      "loss": 0.8027,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7911210998092807e-06,
+      "loss": 0.7358,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7899526762391504e-06,
+      "loss": 0.9021,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7887838685365195e-06,
+      "loss": 0.8758,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.787614677049445e-06,
+      "loss": 0.8453,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.786445102126097e-06,
+      "loss": 0.5567,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7852751441147633e-06,
+      "loss": 0.7723,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.784104803363842e-06,
+      "loss": 0.8408,
+      "step": 2069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.782934080221845e-06,
+      "loss": 0.8426,
+      "step": 2070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.781762975037403e-06,
+      "loss": 0.8671,
+      "step": 2071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7805914881592544e-06,
+      "loss": 0.5348,
+      "step": 2072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7794196199362555e-06,
+      "loss": 0.7784,
+      "step": 2073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7782473707173744e-06,
+      "loss": 0.807,
+      "step": 2074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.777074740851693e-06,
+      "loss": 0.8652,
+      "step": 2075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7759017306884055e-06,
+      "loss": 0.7727,
+      "step": 2076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.774728340576821e-06,
+      "loss": 0.828,
+      "step": 2077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7735545708663613e-06,
+      "loss": 0.8495,
+      "step": 2078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.772380421906561e-06,
+      "loss": 0.8215,
+      "step": 2079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7712058940470675e-06,
+      "loss": 0.7969,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7700309876376407e-06,
+      "loss": 0.8517,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.768855703028155e-06,
+      "loss": 0.7297,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7676800405685937e-06,
+      "loss": 0.8293,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7665040006090573e-06,
+      "loss": 0.853,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.765327583499756e-06,
+      "loss": 0.8243,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.764150789591012e-06,
+      "loss": 0.757,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.762973619233261e-06,
+      "loss": 0.7935,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.761796072777051e-06,
+      "loss": 0.8546,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.76061815057304e-06,
+      "loss": 0.8497,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7594398529719988e-06,
+      "loss": 0.7806,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7582611803248124e-06,
+      "loss": 0.795,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7570821329824735e-06,
+      "loss": 0.8361,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7559027112960898e-06,
+      "loss": 0.8227,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7547229156168776e-06,
+      "loss": 0.8099,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.753542746296167e-06,
+      "loss": 0.8789,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.752362203685398e-06,
+      "loss": 0.801,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.751181288136121e-06,
+      "loss": 0.8456,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.8786,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.748818339628809e-06,
+      "loss": 0.871,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7476363073744305e-06,
+      "loss": 0.8629,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7464539035888605e-06,
+      "loss": 0.7977,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7452711286242055e-06,
+      "loss": 0.8221,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7440879828326805e-06,
+      "loss": 0.7451,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7429044665666126e-06,
+      "loss": 0.8226,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7417205801784396e-06,
+      "loss": 0.7646,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.740536324020707e-06,
+      "loss": 0.8419,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.739351698446074e-06,
+      "loss": 0.8178,
+      "step": 2107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.738166703807306e-06,
+      "loss": 0.8373,
+      "step": 2108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7369813404572818e-06,
+      "loss": 0.7881,
+      "step": 2109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.735795608748989e-06,
+      "loss": 0.8061,
+      "step": 2110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7346095090355223e-06,
+      "loss": 0.9297,
+      "step": 2111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7334230416700888e-06,
+      "loss": 0.8079,
+      "step": 2112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.732236207006006e-06,
+      "loss": 0.7836,
+      "step": 2113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.731049005396696e-06,
+      "loss": 0.8473,
+      "step": 2114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7298614371956958e-06,
+      "loss": 0.808,
+      "step": 2115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7286735027566484e-06,
+      "loss": 0.8167,
+      "step": 2116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.727485202433306e-06,
+      "loss": 0.8572,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.72629653657953e-06,
+      "loss": 0.8147,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7251075055492925e-06,
+      "loss": 0.835,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.72391810969667e-06,
+      "loss": 0.7965,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7227283493758537e-06,
+      "loss": 0.7831,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.721538224941137e-06,
+      "loss": 0.7476,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.720347736746927e-06,
+      "loss": 0.7781,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7191568851477353e-06,
+      "loss": 0.8289,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.717965670498185e-06,
+      "loss": 0.7945,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7167740931530026e-06,
+      "loss": 0.7671,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.715582153467029e-06,
+      "loss": 0.8118,
+      "step": 2127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7143898517952078e-06,
+      "loss": 0.885,
+      "step": 2128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7131971884925923e-06,
+      "loss": 0.7885,
+      "step": 2129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.712004163914344e-06,
+      "loss": 0.8829,
+      "step": 2130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7108107784157297e-06,
+      "loss": 0.9154,
+      "step": 2131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.709617032352128e-06,
+      "loss": 0.7976,
+      "step": 2132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.70842292607902e-06,
+      "loss": 0.8507,
+      "step": 2133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.707228459951998e-06,
+      "loss": 0.82,
+      "step": 2134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.706033634326758e-06,
+      "loss": 0.7311,
+      "step": 2135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7048384495591067e-06,
+      "loss": 0.8927,
+      "step": 2136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.703642906004954e-06,
+      "loss": 0.8192,
+      "step": 2137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.702447004020319e-06,
+      "loss": 0.8754,
+      "step": 2138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7012507439613284e-06,
+      "loss": 0.8104,
+      "step": 2139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.700054126184212e-06,
+      "loss": 0.7423,
+      "step": 2140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6988571510453103e-06,
+      "loss": 0.7802,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.697659818901067e-06,
+      "loss": 0.8483,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6964621301080333e-06,
+      "loss": 0.7918,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.695264085022867e-06,
+      "loss": 0.7153,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6940656840023314e-06,
+      "loss": 0.8682,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6928669274032963e-06,
+      "loss": 0.7824,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6916678155827375e-06,
+      "loss": 0.8479,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6904683488977342e-06,
+      "loss": 0.7784,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6892685277054762e-06,
+      "loss": 0.8214,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6880683523632538e-06,
+      "loss": 0.8096,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.686867823228466e-06,
+      "loss": 0.7634,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6856669406586143e-06,
+      "loss": 0.8453,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6844657050113096e-06,
+      "loss": 0.8024,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.683264116644263e-06,
+      "loss": 0.7292,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6820621759152957e-06,
+      "loss": 0.7973,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6808598831823294e-06,
+      "loss": 0.8498,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6796572388033936e-06,
+      "loss": 0.7853,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.678454243136621e-06,
+      "loss": 0.8007,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6772508965402487e-06,
+      "loss": 0.5307,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.676047199372621e-06,
+      "loss": 0.8757,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6748431519921817e-06,
+      "loss": 0.7713,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6736387547574847e-06,
+      "loss": 0.7549,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.672434008027182e-06,
+      "loss": 0.771,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.671228912160036e-06,
+      "loss": 0.7033,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6700234675149064e-06,
+      "loss": 0.8045,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6688176744507637e-06,
+      "loss": 0.7825,
+      "step": 2166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6676115333266765e-06,
+      "loss": 0.5448,
+      "step": 2167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6664050445018194e-06,
+      "loss": 0.8145,
+      "step": 2168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6651982083354707e-06,
+      "loss": 0.8244,
+      "step": 2169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6639910251870125e-06,
+      "loss": 0.8036,
+      "step": 2170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6627834954159286e-06,
+      "loss": 0.7621,
+      "step": 2171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.661575619381807e-06,
+      "loss": 0.7688,
+      "step": 2172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6603673974443386e-06,
+      "loss": 0.7783,
+      "step": 2173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6591588299633184e-06,
+      "loss": 0.8026,
+      "step": 2174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6579499172986425e-06,
+      "loss": 0.7839,
+      "step": 2175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.656740659810312e-06,
+      "loss": 0.7901,
+      "step": 2176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6555310578584267e-06,
+      "loss": 0.7525,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.654321111803194e-06,
+      "loss": 0.7869,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.65311082200492e-06,
+      "loss": 0.8419,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6519001888240156e-06,
+      "loss": 0.7996,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.650689212620992e-06,
+      "loss": 0.793,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.649477893756464e-06,
+      "loss": 0.8159,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6482662325911483e-06,
+      "loss": 0.8163,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.647054229485862e-06,
+      "loss": 0.7677,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6458418848015255e-06,
+      "loss": 0.7383,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.644629198899161e-06,
+      "loss": 0.7724,
+      "step": 2186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.643416172139892e-06,
+      "loss": 0.8309,
+      "step": 2187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.642202804884943e-06,
+      "loss": 0.845,
+      "step": 2188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.64098909749564e-06,
+      "loss": 0.8128,
+      "step": 2189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.639775050333412e-06,
+      "loss": 0.7952,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6385606637597847e-06,
+      "loss": 0.8552,
+      "step": 2191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6373459381363917e-06,
+      "loss": 0.8615,
+      "step": 2192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.636130873824961e-06,
+      "loss": 0.8024,
+      "step": 2193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.634915471187325e-06,
+      "loss": 0.7287,
+      "step": 2194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6336997305854168e-06,
+      "loss": 0.5064,
+      "step": 2195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.632483652381269e-06,
+      "loss": 0.9071,
+      "step": 2196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6312672369370145e-06,
+      "loss": 0.8653,
+      "step": 2197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.630050484614888e-06,
+      "loss": 0.8402,
+      "step": 2198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.628833395777224e-06,
+      "loss": 0.8053,
+      "step": 2199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6276159707864565e-06,
+      "loss": 0.8421,
+      "step": 2200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6263982100051197e-06,
+      "loss": 0.792,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.62518011379585e-06,
+      "loss": 0.8313,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6239616825213798e-06,
+      "loss": 0.7854,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6227429165445442e-06,
+      "loss": 0.7869,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6215238162282783e-06,
+      "loss": 0.7882,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6203043819356144e-06,
+      "loss": 0.8129,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6190846140296848e-06,
+      "loss": 0.8301,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6178645128737235e-06,
+      "loss": 0.8465,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6166440788310603e-06,
+      "loss": 0.7566,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6154233122651272e-06,
+      "loss": 0.7781,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.614202213539454e-06,
+      "loss": 0.7826,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6129807830176677e-06,
+      "loss": 0.823,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.611759021063498e-06,
+      "loss": 0.7571,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.610536928040769e-06,
+      "loss": 0.8698,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.609314504313407e-06,
+      "loss": 0.8794,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.608091750245434e-06,
+      "loss": 0.8178,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6068686662009723e-06,
+      "loss": 0.8695,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6056452525442408e-06,
+      "loss": 0.5779,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6044215096395583e-06,
+      "loss": 0.8027,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6031974378513414e-06,
+      "loss": 0.7585,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6019730375441025e-06,
+      "loss": 0.7762,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.600748309082455e-06,
+      "loss": 0.7032,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5995232528311063e-06,
+      "loss": 0.811,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5982978691548666e-06,
+      "loss": 0.7679,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.597072158418636e-06,
+      "loss": 0.7042,
+      "step": 2225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.595846120987422e-06,
+      "loss": 0.7334,
+      "step": 2226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.59461975722632e-06,
+      "loss": 0.8474,
+      "step": 2227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5933930675005276e-06,
+      "loss": 0.6072,
+      "step": 2228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.592166052175338e-06,
+      "loss": 0.7829,
+      "step": 2229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5909387116161425e-06,
+      "loss": 0.8061,
+      "step": 2230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.589711046188428e-06,
+      "loss": 0.8408,
+      "step": 2231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5884830562577784e-06,
+      "loss": 0.8241,
+      "step": 2232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.587254742189876e-06,
+      "loss": 0.7918,
+      "step": 2233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5860261043504945e-06,
+      "loss": 0.8534,
+      "step": 2234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.584797143105512e-06,
+      "loss": 0.7848,
+      "step": 2235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.583567858820895e-06,
+      "loss": 0.6534,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5823382518627115e-06,
+      "loss": 0.8574,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.581108322597123e-06,
+      "loss": 0.8386,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.579878071390388e-06,
+      "loss": 0.8181,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5786474986088605e-06,
+      "loss": 0.7402,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.577416604618991e-06,
+      "loss": 0.8068,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5761853897873235e-06,
+      "loss": 0.8864,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.574953854480501e-06,
+      "loss": 0.825,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5737219990652584e-06,
+      "loss": 0.7855,
+      "step": 2244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.572489823908428e-06,
+      "loss": 0.7869,
+      "step": 2245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5712573293769383e-06,
+      "loss": 0.8,
+      "step": 2246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5700245158378094e-06,
+      "loss": 0.8553,
+      "step": 2247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5687913836581596e-06,
+      "loss": 0.7996,
+      "step": 2248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5675579332052002e-06,
+      "loss": 0.7257,
+      "step": 2249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.566324164846239e-06,
+      "loss": 0.759,
+      "step": 2250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5650900789486762e-06,
+      "loss": 0.7791,
+      "step": 2251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5638556758800093e-06,
+      "loss": 0.8052,
+      "step": 2252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.562620956007827e-06,
+      "loss": 0.7906,
+      "step": 2253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5613859196998163e-06,
+      "loss": 0.5397,
+      "step": 2254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.560150567323754e-06,
+      "loss": 0.7562,
+      "step": 2255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.558914899247516e-06,
+      "loss": 0.8758,
+      "step": 2256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5576789158390663e-06,
+      "loss": 0.9075,
+      "step": 2257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5564426174664685e-06,
+      "loss": 0.7747,
+      "step": 2258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5552060044978763e-06,
+      "loss": 0.7198,
+      "step": 2259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.553969077301538e-06,
+      "loss": 0.7727,
+      "step": 2260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5527318362457973e-06,
+      "loss": 0.8555,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5514942816990877e-06,
+      "loss": 0.7421,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5502564140299407e-06,
+      "loss": 0.691,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5490182336069755e-06,
+      "loss": 0.8393,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.547779740798911e-06,
+      "loss": 0.8222,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.546540935974552e-06,
+      "loss": 0.8431,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5453018195028014e-06,
+      "loss": 0.8231,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.544062391752654e-06,
+      "loss": 0.8196,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5428226530931953e-06,
+      "loss": 0.747,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5415826038936056e-06,
+      "loss": 0.812,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.540342244523156e-06,
+      "loss": 0.7429,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5391015753512116e-06,
+      "loss": 0.5786,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5378605967472277e-06,
+      "loss": 0.7638,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.536619309080754e-06,
+      "loss": 0.7676,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5353777127214308e-06,
+      "loss": 0.8203,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.53413580803899e-06,
+      "loss": 0.8162,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.532893595403256e-06,
+      "loss": 0.8433,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5316510751841463e-06,
+      "loss": 0.8542,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5304082477516677e-06,
+      "loss": 0.8193,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.529165113475919e-06,
+      "loss": 0.7811,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5279216727270914e-06,
+      "loss": 0.7617,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5266779258754665e-06,
+      "loss": 0.7318,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5254338732914167e-06,
+      "loss": 0.7918,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.524189515345407e-06,
+      "loss": 0.7937,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.522944852407991e-06,
+      "loss": 0.8112,
+      "step": 2285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5216998848498164e-06,
+      "loss": 0.8306,
+      "step": 2286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5204546130416173e-06,
+      "loss": 0.7807,
+      "step": 2287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.519209037354222e-06,
+      "loss": 0.8299,
+      "step": 2288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.517963158158549e-06,
+      "loss": 0.8815,
+      "step": 2289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5167169758256047e-06,
+      "loss": 0.8107,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5154704907264887e-06,
+      "loss": 0.7538,
+      "step": 2291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.514223703232387e-06,
+      "loss": 0.8467,
+      "step": 2292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5129766137145814e-06,
+      "loss": 0.8114,
+      "step": 2293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5117292225444373e-06,
+      "loss": 0.8514,
+      "step": 2294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.510481530093415e-06,
+      "loss": 0.8943,
+      "step": 2295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.50923353673306e-06,
+      "loss": 0.7742,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5079852428350124e-06,
+      "loss": 0.7761,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.506736648770999e-06,
+      "loss": 0.779,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5054877549128334e-06,
+      "loss": 0.8027,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5042385616324243e-06,
+      "loss": 0.8153,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.502989069301765e-06,
+      "loss": 0.8357,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.50173927829294e-06,
+      "loss": 0.7287,
+      "step": 2302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.500489188978122e-06,
+      "loss": 0.7519,
+      "step": 2303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4992388017295727e-06,
+      "loss": 0.8297,
+      "step": 2304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4979881169196418e-06,
+      "loss": 0.779,
+      "step": 2305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4967371349207694e-06,
+      "loss": 0.7533,
+      "step": 2306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.495485856105483e-06,
+      "loss": 0.7651,
+      "step": 2307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.494234280846397e-06,
+      "loss": 0.7551,
+      "step": 2308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.492982409516217e-06,
+      "loss": 0.7866,
+      "step": 2309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4917302424877353e-06,
+      "loss": 0.5464,
+      "step": 2310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4904777801338314e-06,
+      "loss": 0.5913,
+      "step": 2311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4892250228274736e-06,
+      "loss": 0.8136,
+      "step": 2312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.48797197094172e-06,
+      "loss": 0.7388,
+      "step": 2313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4867186248497128e-06,
+      "loss": 0.7056,
+      "step": 2314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4854649849246825e-06,
+      "loss": 0.8598,
+      "step": 2315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4842110515399496e-06,
+      "loss": 0.8758,
+      "step": 2316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.482956825068921e-06,
+      "loss": 0.5159,
+      "step": 2317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.481702305885089e-06,
+      "loss": 0.7982,
+      "step": 2318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4804474943620357e-06,
+      "loss": 0.8007,
+      "step": 2319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4791923908734276e-06,
+      "loss": 0.8807,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.477936995793021e-06,
+      "loss": 0.6884,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.476681309494656e-06,
+      "loss": 0.8165,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.475425332352263e-06,
+      "loss": 0.8391,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.474169064739854e-06,
+      "loss": 0.8374,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4729125070315337e-06,
+      "loss": 0.8436,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.471655659601488e-06,
+      "loss": 0.8794,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.470398522823992e-06,
+      "loss": 0.8762,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.469141097073405e-06,
+      "loss": 0.8653,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.467883382724175e-06,
+      "loss": 0.7397,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4666253801508326e-06,
+      "loss": 0.847,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4653670897279967e-06,
+      "loss": 0.8444,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4641085118303706e-06,
+      "loss": 0.8169,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4628496468327456e-06,
+      "loss": 0.8143,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4615904951099944e-06,
+      "loss": 0.8343,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4603310570370772e-06,
+      "loss": 0.7694,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.459071332989043e-06,
+      "loss": 0.8082,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4578113233410187e-06,
+      "loss": 0.8799,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4565510284682223e-06,
+      "loss": 0.499,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.455290448745954e-06,
+      "loss": 0.7419,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.454029584549599e-06,
+      "loss": 0.7933,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4527684362546283e-06,
+      "loss": 0.7636,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.451507004236596e-06,
+      "loss": 0.7903,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4502452888711424e-06,
+      "loss": 0.544,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4489832905339904e-06,
+      "loss": 0.7921,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.447721009600949e-06,
+      "loss": 0.8145,
+      "step": 2345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.446458446447909e-06,
+      "loss": 0.8581,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.445195601450848e-06,
+      "loss": 0.8461,
+      "step": 2347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4439324749858257e-06,
+      "loss": 0.8271,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.442669067428985e-06,
+      "loss": 0.8364,
+      "step": 2349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.441405379156556e-06,
+      "loss": 0.756,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4401414105448476e-06,
+      "loss": 0.8396,
+      "step": 2351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.438877161970256e-06,
+      "loss": 0.8383,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.437612633809259e-06,
+      "loss": 0.8657,
+      "step": 2353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.436347826438417e-06,
+      "loss": 0.8468,
+      "step": 2354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.435082740234376e-06,
+      "loss": 0.7701,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4338173755738624e-06,
+      "loss": 0.7731,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4325517328336877e-06,
+      "loss": 0.7278,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4312858123907438e-06,
+      "loss": 0.8145,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.430019614622009e-06,
+      "loss": 0.8906,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.428753139904539e-06,
+      "loss": 0.5844,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.427486388615477e-06,
+      "loss": 0.7931,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4262193611320454e-06,
+      "loss": 0.8958,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.42495205783155e-06,
+      "loss": 0.7429,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4236844790913792e-06,
+      "loss": 0.7916,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4224166252890014e-06,
+      "loss": 0.7736,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4211484968019703e-06,
+      "loss": 0.8462,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.419880094007918e-06,
+      "loss": 0.7768,
+      "step": 2367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4186114172845603e-06,
+      "loss": 0.7819,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4173424670096933e-06,
+      "loss": 0.7533,
+      "step": 2369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.416073243561196e-06,
+      "loss": 0.822,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4148037473170275e-06,
+      "loss": 0.8469,
+      "step": 2371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.413533978655229e-06,
+      "loss": 0.7654,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4122639379539224e-06,
+      "loss": 0.8148,
+      "step": 2373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.41099362559131e-06,
+      "loss": 0.7169,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.409723041945676e-06,
+      "loss": 0.8074,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4084521873953856e-06,
+      "loss": 0.7841,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.407181062318884e-06,
+      "loss": 0.8707,
+      "step": 2377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4059096670946956e-06,
+      "loss": 0.7113,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4046380021014288e-06,
+      "loss": 0.8079,
+      "step": 2379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4033660677177695e-06,
+      "loss": 0.7748,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4020938643224834e-06,
+      "loss": 0.8461,
+      "step": 2381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4008213922944187e-06,
+      "loss": 0.8146,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3995486520125027e-06,
+      "loss": 0.8299,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3982756438557413e-06,
+      "loss": 0.524,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3970023682032222e-06,
+      "loss": 0.7257,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3957288254341097e-06,
+      "loss": 0.7963,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3944550159276517e-06,
+      "loss": 0.8696,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.393180940063172e-06,
+      "loss": 0.8064,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.391906598220077e-06,
+      "loss": 0.7763,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3906319907778486e-06,
+      "loss": 0.7493,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.38935711811605e-06,
+      "loss": 0.8154,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3880819806143243e-06,
+      "loss": 0.7937,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3868065786523906e-06,
+      "loss": 0.7915,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.38553091261005e-06,
+      "loss": 0.7626,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.384254982867179e-06,
+      "loss": 0.7543,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3829787898037352e-06,
+      "loss": 0.7286,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3817023337997533e-06,
+      "loss": 0.771,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.380425615235348e-06,
+      "loss": 0.7879,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3791486344907078e-06,
+      "loss": 0.7605,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3778713919461053e-06,
+      "loss": 0.8499,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3765938879818865e-06,
+      "loss": 0.8188,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.375316122978477e-06,
+      "loss": 0.7888,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.374038097316381e-06,
+      "loss": 0.7729,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3727598113761776e-06,
+      "loss": 0.8606,
+      "step": 2404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3714812655385265e-06,
+      "loss": 0.822,
+      "step": 2405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3702024601841627e-06,
+      "loss": 0.7376,
+      "step": 2406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3689233956938994e-06,
+      "loss": 0.8183,
+      "step": 2407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3676440724486263e-06,
+      "loss": 0.7672,
+      "step": 2408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3663644908293115e-06,
+      "loss": 0.8282,
+      "step": 2409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.365084651216998e-06,
+      "loss": 0.7681,
+      "step": 2410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3638045539928075e-06,
+      "loss": 0.8174,
+      "step": 2411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3625241995379377e-06,
+      "loss": 0.832,
+      "step": 2412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3612435882336623e-06,
+      "loss": 0.5308,
+      "step": 2413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3599627204613326e-06,
+      "loss": 0.7718,
+      "step": 2414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3586815966023762e-06,
+      "loss": 0.8236,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.357400217038295e-06,
+      "loss": 0.7948,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.356118582150669e-06,
+      "loss": 0.7406,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.354836692321154e-06,
+      "loss": 0.8015,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3535545479314825e-06,
+      "loss": 0.8316,
+      "step": 2419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3522721493634603e-06,
+      "loss": 0.8338,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.350989496998971e-06,
+      "loss": 0.8009,
+      "step": 2421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.349706591219973e-06,
+      "loss": 0.782,
+      "step": 2422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3484234324085e-06,
+      "loss": 0.7244,
+      "step": 2423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.347140020946663e-06,
+      "loss": 0.7273,
+      "step": 2424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3458563572166443e-06,
+      "loss": 0.8317,
+      "step": 2425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.344572441600705e-06,
+      "loss": 0.7557,
+      "step": 2426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.343288274481179e-06,
+      "loss": 0.8039,
+      "step": 2427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3420038562404777e-06,
+      "loss": 0.7495,
+      "step": 2428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3407191872610834e-06,
+      "loss": 0.785,
+      "step": 2429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3394342679255563e-06,
+      "loss": 0.8276,
+      "step": 2430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3381490986165297e-06,
+      "loss": 0.7765,
+      "step": 2431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.336863679716712e-06,
+      "loss": 0.8328,
+      "step": 2432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3355780116088852e-06,
+      "loss": 0.8509,
+      "step": 2433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.334292094675906e-06,
+      "loss": 0.7709,
+      "step": 2434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.333005929300705e-06,
+      "loss": 0.7726,
+      "step": 2435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3317195158662872e-06,
+      "loss": 0.7311,
+      "step": 2436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.330432854755731e-06,
+      "loss": 0.5549,
+      "step": 2437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3291459463521887e-06,
+      "loss": 0.5439,
+      "step": 2438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3278587910388865e-06,
+      "loss": 0.5543,
+      "step": 2439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3265713891991234e-06,
+      "loss": 0.8631,
+      "step": 2440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.325283741216273e-06,
+      "loss": 0.8051,
+      "step": 2441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.323995847473781e-06,
+      "loss": 0.8275,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3227077083551664e-06,
+      "loss": 0.8915,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3214193242440216e-06,
+      "loss": 0.8456,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.320130695524014e-06,
+      "loss": 0.7075,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3188418225788786e-06,
+      "loss": 0.7349,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3175527057924295e-06,
+      "loss": 0.7771,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.316263345548548e-06,
+      "loss": 0.839,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.314973742231191e-06,
+      "loss": 0.6375,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3136838962243877e-06,
+      "loss": 0.7608,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3123938079122375e-06,
+      "loss": 0.7981,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.311103477678914e-06,
+      "loss": 0.7738,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3098129059086624e-06,
+      "loss": 0.8861,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3085220929857997e-06,
+      "loss": 0.7903,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3072310392947127e-06,
+      "loss": 0.7981,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3059397452198644e-06,
+      "loss": 0.8305,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3046482111457843e-06,
+      "loss": 0.8456,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3033564374570787e-06,
+      "loss": 0.7601,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3020644245384193e-06,
+      "loss": 0.8745,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3007721727745545e-06,
+      "loss": 0.8047,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2994796825502996e-06,
+      "loss": 0.6741,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2981869542505447e-06,
+      "loss": 0.8214,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.296893988260247e-06,
+      "loss": 0.7787,
+      "step": 2463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2956007849644374e-06,
+      "loss": 0.7606,
+      "step": 2464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2943073447482155e-06,
+      "loss": 0.7168,
+      "step": 2465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2930136679967534e-06,
+      "loss": 0.8157,
+      "step": 2466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2917197550952924e-06,
+      "loss": 0.71,
+      "step": 2467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2904256064291427e-06,
+      "loss": 0.784,
+      "step": 2468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2891312223836898e-06,
+      "loss": 0.7973,
+      "step": 2469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.287836603344382e-06,
+      "loss": 0.7826,
+      "step": 2470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.286541749696744e-06,
+      "loss": 0.7654,
+      "step": 2471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2852466618263667e-06,
+      "loss": 0.8424,
+      "step": 2472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2839513401189115e-06,
+      "loss": 0.7579,
+      "step": 2473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.28265578496011e-06,
+      "loss": 0.7693,
+      "step": 2474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.281359996735764e-06,
+      "loss": 0.7835,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.280063975831742e-06,
+      "loss": 0.856,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2787677226339854e-06,
+      "loss": 0.6959,
+      "step": 2477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2774712375285017e-06,
+      "loss": 0.8108,
+      "step": 2478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2761745209013684e-06,
+      "loss": 0.7487,
+      "step": 2479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.274877573138734e-06,
+      "loss": 0.7496,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2735803946268117e-06,
+      "loss": 0.7633,
+      "step": 2481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.272282985751888e-06,
+      "loss": 0.7824,
+      "step": 2482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.270985346900314e-06,
+      "loss": 0.7514,
+      "step": 2483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.269687478458512e-06,
+      "loss": 0.7572,
+      "step": 2484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.268389380812971e-06,
+      "loss": 0.5968,
+      "step": 2485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.267091054350251e-06,
+      "loss": 0.8612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2657924994569744e-06,
+      "loss": 0.8444,
+      "step": 2487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2644937165198397e-06,
+      "loss": 0.6928,
+      "step": 2488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.263194705925605e-06,
+      "loss": 0.7938,
+      "step": 2489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.261895468061103e-06,
+      "loss": 0.7844,
+      "step": 2490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2605960033132296e-06,
+      "loss": 0.7716,
+      "step": 2491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2592963120689506e-06,
+      "loss": 0.8141,
+      "step": 2492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.257996394715299e-06,
+      "loss": 0.6863,
+      "step": 2493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.256696251639373e-06,
+      "loss": 0.7972,
+      "step": 2494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2553958832283404e-06,
+      "loss": 0.7614,
+      "step": 2495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.254095289869436e-06,
+      "loss": 0.847,
+      "step": 2496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2527944719499614e-06,
+      "loss": 0.7601,
+      "step": 2497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2514934298572832e-06,
+      "loss": 0.7671,
+      "step": 2498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2501921639788375e-06,
+      "loss": 0.8006,
+      "step": 2499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2488906747021236e-06,
+      "loss": 0.741,
+      "step": 2500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.247588962414713e-06,
+      "loss": 0.8239,
+      "step": 2501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.246287027504237e-06,
+      "loss": 0.7926,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2449848703583977e-06,
+      "loss": 0.8137,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2436824913649607e-06,
+      "loss": 0.7603,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2423798909117604e-06,
+      "loss": 0.7853,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2410770693866946e-06,
+      "loss": 0.799,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.239774027177729e-06,
+      "loss": 0.7836,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.238470764672893e-06,
+      "loss": 0.7952,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.237167282260283e-06,
+      "loss": 0.5272,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2358635803280608e-06,
+      "loss": 0.7873,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.234559659264453e-06,
+      "loss": 0.8329,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2332555194577516e-06,
+      "loss": 0.8315,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.231951161296314e-06,
+      "loss": 0.7022,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.230646585168562e-06,
+      "loss": 0.8721,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2293417914629838e-06,
+      "loss": 0.8706,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2280367805681307e-06,
+      "loss": 0.7381,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.226731552872619e-06,
+      "loss": 0.804,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2254261087651315e-06,
+      "loss": 0.7591,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2241204486344114e-06,
+      "loss": 0.7203,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2228145728692716e-06,
+      "loss": 0.7826,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.221508481858583e-06,
+      "loss": 0.8387,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.220202175991287e-06,
+      "loss": 0.75,
+      "step": 2522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.218895655656384e-06,
+      "loss": 0.7086,
+      "step": 2523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2175889212429408e-06,
+      "loss": 0.7314,
+      "step": 2524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2162819731400876e-06,
+      "loss": 0.8842,
+      "step": 2525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.214974811737017e-06,
+      "loss": 0.8279,
+      "step": 2526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.213667437422987e-06,
+      "loss": 0.7602,
+      "step": 2527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2123598505873167e-06,
+      "loss": 0.8626,
+      "step": 2528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2110520516193923e-06,
+      "loss": 0.8407,
+      "step": 2529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2097440409086576e-06,
+      "loss": 0.8038,
+      "step": 2530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.208435818844625e-06,
+      "loss": 0.7995,
+      "step": 2531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.207127385816867e-06,
+      "loss": 0.8359,
+      "step": 2532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2058187422150183e-06,
+      "loss": 0.8077,
+      "step": 2533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2045098884287785e-06,
+      "loss": 0.7285,
+      "step": 2534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.203200824847907e-06,
+      "loss": 0.8566,
+      "step": 2535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2018915518622297e-06,
+      "loss": 0.5201,
+      "step": 2536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2005820698616303e-06,
+      "loss": 0.7555,
+      "step": 2537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1992723792360582e-06,
+      "loss": 0.8382,
+      "step": 2538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1979624803755227e-06,
+      "loss": 0.8199,
+      "step": 2539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.196652373670097e-06,
+      "loss": 0.7419,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1953420595099138e-06,
+      "loss": 0.8886,
+      "step": 2541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1940315382851707e-06,
+      "loss": 0.7894,
+      "step": 2542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.192720810386124e-06,
+      "loss": 0.5772,
+      "step": 2543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.191409876203094e-06,
+      "loss": 0.7982,
+      "step": 2544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.19009873612646e-06,
+      "loss": 0.6883,
+      "step": 2545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.188787390546664e-06,
+      "loss": 0.7378,
+      "step": 2546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1874758398542104e-06,
+      "loss": 0.8024,
+      "step": 2547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1861640844396624e-06,
+      "loss": 0.5104,
+      "step": 2548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1848521246936443e-06,
+      "loss": 0.793,
+      "step": 2549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.183539961006844e-06,
+      "loss": 0.812,
+      "step": 2550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1822275937700053e-06,
+      "loss": 0.8808,
+      "step": 2551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1809150233739382e-06,
+      "loss": 0.5687,
+      "step": 2552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1796022502095097e-06,
+      "loss": 0.8176,
+      "step": 2553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1782892746676474e-06,
+      "loss": 0.7433,
+      "step": 2554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1769760971393406e-06,
+      "loss": 0.8559,
+      "step": 2555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1756627180156362e-06,
+      "loss": 0.7336,
+      "step": 2556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1743491376876456e-06,
+      "loss": 0.7905,
+      "step": 2557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1730353565465343e-06,
+      "loss": 0.525,
+      "step": 2558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1717213749835335e-06,
+      "loss": 0.7923,
+      "step": 2559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.170407193389928e-06,
+      "loss": 0.5281,
+      "step": 2560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1690928121570688e-06,
+      "loss": 0.7858,
+      "step": 2561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.16777823167636e-06,
+      "loss": 0.745,
+      "step": 2562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1664634523392702e-06,
+      "loss": 0.6905,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.165148474537324e-06,
+      "loss": 0.7858,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1638332986621057e-06,
+      "loss": 0.8428,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.162517925105259e-06,
+      "loss": 0.7888,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1612023542584875e-06,
+      "loss": 0.7945,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1598865865135513e-06,
+      "loss": 0.7921,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1585706222622715e-06,
+      "loss": 0.7668,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.157254461896525e-06,
+      "loss": 0.8142,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1559381058082505e-06,
+      "loss": 0.842,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1546215543894415e-06,
+      "loss": 0.7635,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.153304808032152e-06,
+      "loss": 0.8096,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.151987867128493e-06,
+      "loss": 0.7839,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.150670732070635e-06,
+      "loss": 0.7198,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1493534032508034e-06,
+      "loss": 0.7488,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.148035881061283e-06,
+      "loss": 0.729,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.146718165894418e-06,
+      "loss": 0.8346,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1454002581426068e-06,
+      "loss": 0.7958,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1440821581983073e-06,
+      "loss": 0.8543,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1427638664540327e-06,
+      "loss": 0.7066,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1414453833023557e-06,
+      "loss": 0.8078,
+      "step": 2582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1401267091359043e-06,
+      "loss": 0.8189,
+      "step": 2583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.138807844347365e-06,
+      "loss": 0.8247,
+      "step": 2584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.137488789329478e-06,
+      "loss": 0.7937,
+      "step": 2585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.136169544475044e-06,
+      "loss": 0.7943,
+      "step": 2586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.134850110176918e-06,
+      "loss": 0.8216,
+      "step": 2587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.13353048682801e-06,
+      "loss": 0.7972,
+      "step": 2588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1322106748212906e-06,
+      "loss": 0.7236,
+      "step": 2589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1308906745497818e-06,
+      "loss": 0.8708,
+      "step": 2590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1295704864065664e-06,
+      "loss": 0.549,
+      "step": 2591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.128250110784778e-06,
+      "loss": 0.7829,
+      "step": 2592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1269295480776108e-06,
+      "loss": 0.8483,
+      "step": 2593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1256087986783113e-06,
+      "loss": 0.5576,
+      "step": 2594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1242878629801837e-06,
+      "loss": 0.7245,
+      "step": 2595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.122966741376586e-06,
+      "loss": 0.895,
+      "step": 2596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1216454342609338e-06,
+      "loss": 0.7705,
+      "step": 2597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.120323942026695e-06,
+      "loss": 0.767,
+      "step": 2598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1190022650673944e-06,
+      "loss": 0.7164,
+      "step": 2599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1176804037766136e-06,
+      "loss": 0.7915,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1163583585479844e-06,
+      "loss": 0.7827,
+      "step": 2601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1150361297751985e-06,
+      "loss": 0.8186,
+      "step": 2602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1137137178519983e-06,
+      "loss": 0.7853,
+      "step": 2603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1123911231721825e-06,
+      "loss": 0.8045,
+      "step": 2604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1110683461296044e-06,
+      "loss": 0.8139,
+      "step": 2605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1097453871181705e-06,
+      "loss": 0.7395,
+      "step": 2606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1084222465318433e-06,
+      "loss": 0.8798,
+      "step": 2607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.107098924764637e-06,
+      "loss": 0.8015,
+      "step": 2608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.105775422210622e-06,
+      "loss": 0.8966,
+      "step": 2609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.104451739263921e-06,
+      "loss": 0.8269,
+      "step": 2610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1031278763187113e-06,
+      "loss": 0.8326,
+      "step": 2611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1018038337692236e-06,
+      "loss": 0.7729,
+      "step": 2612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1004796120097413e-06,
+      "loss": 0.754,
+      "step": 2613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0991552114346026e-06,
+      "loss": 0.8293,
+      "step": 2614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.097830632438197e-06,
+      "loss": 0.7616,
+      "step": 2615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.09650587541497e-06,
+      "loss": 0.6961,
+      "step": 2616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.095180940759417e-06,
+      "loss": 0.8005,
+      "step": 2617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0938558288660886e-06,
+      "loss": 0.7745,
+      "step": 2618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0925305401295874e-06,
+      "loss": 0.7543,
+      "step": 2619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0912050749445664e-06,
+      "loss": 0.794,
+      "step": 2620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.089879433705737e-06,
+      "loss": 0.9016,
+      "step": 2621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0885536168078554e-06,
+      "loss": 0.6827,
+      "step": 2622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.087227624645738e-06,
+      "loss": 0.7762,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0859014576142455e-06,
+      "loss": 0.7737,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0845751161082983e-06,
+      "loss": 0.7547,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0832486005228625e-06,
+      "loss": 0.7492,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.081921911252958e-06,
+      "loss": 0.8086,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0805950486936603e-06,
+      "loss": 0.8019,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.07926801324009e-06,
+      "loss": 0.7916,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.077940805287425e-06,
+      "loss": 0.7753,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0766134252308887e-06,
+      "loss": 0.7667,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.075285873465763e-06,
+      "loss": 0.8599,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0739581503873746e-06,
+      "loss": 0.7481,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.072630256391103e-06,
+      "loss": 0.7016,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0713021918723806e-06,
+      "loss": 0.7343,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.069973957226689e-06,
+      "loss": 0.8336,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0686455528495603e-06,
+      "loss": 0.8128,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.067316979136577e-06,
+      "loss": 0.722,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0659882364833725e-06,
+      "loss": 0.7828,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0646593252856316e-06,
+      "loss": 0.7914,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.063330245939088e-06,
+      "loss": 0.8018,
+      "step": 2641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0620009988395245e-06,
+      "loss": 0.8229,
+      "step": 2642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0606715843827755e-06,
+      "loss": 0.758,
+      "step": 2643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0593420029647254e-06,
+      "loss": 0.8373,
+      "step": 2644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0580122549813067e-06,
+      "loss": 0.7951,
+      "step": 2645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.056682340828503e-06,
+      "loss": 0.8822,
+      "step": 2646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.055352260902347e-06,
+      "loss": 0.792,
+      "step": 2647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0540220155989203e-06,
+      "loss": 0.8011,
+      "step": 2648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.052691605314355e-06,
+      "loss": 0.7702,
+      "step": 2649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0513610304448295e-06,
+      "loss": 0.5575,
+      "step": 2650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0500302913865747e-06,
+      "loss": 0.7258,
+      "step": 2651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.048699388535868e-06,
+      "loss": 0.8057,
+      "step": 2652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.047368322289037e-06,
+      "loss": 0.7322,
+      "step": 2653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0460370930424566e-06,
+      "loss": 0.8166,
+      "step": 2654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0447057011925514e-06,
+      "loss": 0.796,
+      "step": 2655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0433741471357937e-06,
+      "loss": 0.7828,
+      "step": 2656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0420424312687052e-06,
+      "loss": 0.5748,
+      "step": 2657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0407105539878547e-06,
+      "loss": 0.7343,
+      "step": 2658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.039378515689858e-06,
+      "loss": 0.7645,
+      "step": 2659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0380463167713826e-06,
+      "loss": 0.5274,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0367139576291386e-06,
+      "loss": 0.7934,
+      "step": 2661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0353814386598895e-06,
+      "loss": 0.7454,
+      "step": 2662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0340487602604414e-06,
+      "loss": 0.7647,
+      "step": 2663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.032715922827651e-06,
+      "loss": 0.7513,
+      "step": 2664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.031382926758421e-06,
+      "loss": 0.6972,
+      "step": 2665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.030049772449702e-06,
+      "loss": 0.7688,
+      "step": 2666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0287164602984913e-06,
+      "loss": 0.7368,
+      "step": 2667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0273829907018332e-06,
+      "loss": 0.7922,
+      "step": 2668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.026049364056819e-06,
+      "loss": 0.7649,
+      "step": 2669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.024715580760587e-06,
+      "loss": 0.8372,
+      "step": 2670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0233816412103224e-06,
+      "loss": 0.8134,
+      "step": 2671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0220475458032553e-06,
+      "loss": 0.821,
+      "step": 2672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.020713294936664e-06,
+      "loss": 0.8237,
+      "step": 2673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.019378889007873e-06,
+      "loss": 0.797,
+      "step": 2674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0180443284142512e-06,
+      "loss": 0.7851,
+      "step": 2675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0167096135532166e-06,
+      "loss": 0.8511,
+      "step": 2676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.015374744822229e-06,
+      "loss": 0.7562,
+      "step": 2677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0140397226187984e-06,
+      "loss": 0.8069,
+      "step": 2678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.012704547340478e-06,
+      "loss": 0.7797,
+      "step": 2679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0113692193848663e-06,
+      "loss": 0.8656,
+      "step": 2680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0100337391496085e-06,
+      "loss": 0.8053,
+      "step": 2681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0086981070323945e-06,
+      "loss": 0.6856,
+      "step": 2682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0073623234309602e-06,
+      "loss": 0.7779,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.006026388743085e-06,
+      "loss": 0.7707,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0046903033665953e-06,
+      "loss": 0.7987,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0033540676993606e-06,
+      "loss": 0.6932,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0020176821392965e-06,
+      "loss": 0.7633,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0006811470843626e-06,
+      "loss": 0.8086,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.999344462932562e-06,
+      "loss": 0.8006,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.998007630081945e-06,
+      "loss": 0.8677,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.996670648930603e-06,
+      "loss": 0.8138,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.995333519876674e-06,
+      "loss": 0.8297,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9939962433183384e-06,
+      "loss": 0.8117,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.992658819653822e-06,
+      "loss": 0.7028,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.991321249281392e-06,
+      "loss": 0.7988,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.989983532599363e-06,
+      "loss": 0.782,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9886456700060897e-06,
+      "loss": 0.8183,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.987307661899972e-06,
+      "loss": 0.5605,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.985969508679452e-06,
+      "loss": 0.7711,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.984631210743017e-06,
+      "loss": 0.7466,
+      "step": 2700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9832927684891953e-06,
+      "loss": 0.8357,
+      "step": 2701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9819541823165587e-06,
+      "loss": 0.7048,
+      "step": 2702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9806154526237233e-06,
+      "loss": 0.7305,
+      "step": 2703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9792765798093466e-06,
+      "loss": 0.7841,
+      "step": 2704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.977937564272128e-06,
+      "loss": 0.7783,
+      "step": 2705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.976598406410811e-06,
+      "loss": 0.7937,
+      "step": 2706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.975259106624181e-06,
+      "loss": 0.7824,
+      "step": 2707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9739196653110645e-06,
+      "loss": 0.8449,
+      "step": 2708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9725800828703324e-06,
+      "loss": 0.7002,
+      "step": 2709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9712403597008956e-06,
+      "loss": 0.5768,
+      "step": 2710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9699004962017076e-06,
+      "loss": 0.8297,
+      "step": 2711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9685604927717644e-06,
+      "loss": 0.7093,
+      "step": 2712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.967220349810102e-06,
+      "loss": 0.8182,
+      "step": 2713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9658800677158e-06,
+      "loss": 0.5821,
+      "step": 2714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9645396468879783e-06,
+      "loss": 0.8244,
+      "step": 2715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.963199087725797e-06,
+      "loss": 0.8437,
+      "step": 2716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9618583906284603e-06,
+      "loss": 0.8334,
+      "step": 2717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9605175559952105e-06,
+      "loss": 0.7761,
+      "step": 2718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959176584225333e-06,
+      "loss": 0.7995,
+      "step": 2719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9578354757181526e-06,
+      "loss": 0.7384,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9564942308730356e-06,
+      "loss": 0.7966,
+      "step": 2721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.955152850089389e-06,
+      "loss": 0.7903,
+      "step": 2722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9538113337666583e-06,
+      "loss": 0.8077,
+      "step": 2723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9524696823043343e-06,
+      "loss": 0.7814,
+      "step": 2724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.951127896101941e-06,
+      "loss": 0.4911,
+      "step": 2725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9497859755590495e-06,
+      "loss": 0.7093,
+      "step": 2726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9484439210752656e-06,
+      "loss": 0.8153,
+      "step": 2727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9471017330502387e-06,
+      "loss": 0.7972,
+      "step": 2728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9457594118836553e-06,
+      "loss": 0.7972,
+      "step": 2729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9444169579752417e-06,
+      "loss": 0.7818,
+      "step": 2730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9430743717247673e-06,
+      "loss": 0.7464,
+      "step": 2731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9417316535320355e-06,
+      "loss": 0.7773,
+      "step": 2732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.940388803796895e-06,
+      "loss": 0.7647,
+      "step": 2733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9390458229192257e-06,
+      "loss": 0.7847,
+      "step": 2734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9377027112989565e-06,
+      "loss": 0.776,
+      "step": 2735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9363594693360466e-06,
+      "loss": 0.8126,
+      "step": 2736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9350160974304974e-06,
+      "loss": 0.8398,
+      "step": 2737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.933672595982351e-06,
+      "loss": 0.8756,
+      "step": 2738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9323289653916843e-06,
+      "loss": 0.7113,
+      "step": 2739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9309852060586154e-06,
+      "loss": 0.7699,
+      "step": 2740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9296413183832994e-06,
+      "loss": 0.8346,
+      "step": 2741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.92829730276593e-06,
+      "loss": 0.7823,
+      "step": 2742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.926953159606739e-06,
+      "loss": 0.7747,
+      "step": 2743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.925608889305997e-06,
+      "loss": 0.8808,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.92426449226401e-06,
+      "loss": 0.8179,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9229199688811248e-06,
+      "loss": 0.8078,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.921575319557724e-06,
+      "loss": 0.576,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.920230544694228e-06,
+      "loss": 0.8105,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9188856446910947e-06,
+      "loss": 0.5456,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9175406199488193e-06,
+      "loss": 0.7933,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9161954708679335e-06,
+      "loss": 0.7669,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9148501978490086e-06,
+      "loss": 0.7814,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9135048012926474e-06,
+      "loss": 0.7964,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.912159281599497e-06,
+      "loss": 0.7524,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9108136391702335e-06,
+      "loss": 0.7489,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.909467874405576e-06,
+      "loss": 0.794,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9081219877062743e-06,
+      "loss": 0.8538,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.90677597947312e-06,
+      "loss": 0.7043,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9054298501069377e-06,
+      "loss": 0.9198,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.904083600008587e-06,
+      "loss": 0.8108,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9027372295789674e-06,
+      "loss": 0.7766,
+      "step": 2761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.90139073921901e-06,
+      "loss": 0.5621,
+      "step": 2762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9000441293296856e-06,
+      "loss": 0.8162,
+      "step": 2763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8986974003119955e-06,
+      "loss": 0.877,
+      "step": 2764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8973505525669824e-06,
+      "loss": 0.7569,
+      "step": 2765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8960035864957205e-06,
+      "loss": 0.6589,
+      "step": 2766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.89465650249932e-06,
+      "loss": 0.764,
+      "step": 2767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8933093009789263e-06,
+      "loss": 0.5087,
+      "step": 2768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.89196198233572e-06,
+      "loss": 0.7854,
+      "step": 2769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8906145469709174e-06,
+      "loss": 0.5377,
+      "step": 2770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8892669952857676e-06,
+      "loss": 0.7268,
+      "step": 2771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8879193276815556e-06,
+      "loss": 0.798,
+      "step": 2772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.886571544559601e-06,
+      "loss": 0.7924,
+      "step": 2773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8852236463212567e-06,
+      "loss": 0.7848,
+      "step": 2774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8838756333679118e-06,
+      "loss": 0.7531,
+      "step": 2775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8825275061009883e-06,
+      "loss": 0.7156,
+      "step": 2776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8811792649219418e-06,
+      "loss": 0.5595,
+      "step": 2777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.879830910232263e-06,
+      "loss": 0.7885,
+      "step": 2778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.878482442433475e-06,
+      "loss": 0.8521,
+      "step": 2779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8771338619271367e-06,
+      "loss": 0.8061,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8757851691148377e-06,
+      "loss": 0.5386,
+      "step": 2781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8744363643982043e-06,
+      "loss": 0.8237,
+      "step": 2782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.873087448178893e-06,
+      "loss": 0.8495,
+      "step": 2783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.871738420858596e-06,
+      "loss": 0.8177,
+      "step": 2784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.870389282839036e-06,
+      "loss": 0.8019,
+      "step": 2785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.869040034521972e-06,
+      "loss": 0.7789,
+      "step": 2786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8676906763091934e-06,
+      "loss": 0.7748,
+      "step": 2787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.866341208602522e-06,
+      "loss": 0.7565,
+      "step": 2788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8649916318038155e-06,
+      "loss": 0.71,
+      "step": 2789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.863641946314959e-06,
+      "loss": 0.7902,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8622921525378744e-06,
+      "loss": 0.728,
+      "step": 2791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8609422508745134e-06,
+      "loss": 0.796,
+      "step": 2792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8595922417268612e-06,
+      "loss": 0.8629,
+      "step": 2793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8582421254969337e-06,
+      "loss": 0.8382,
+      "step": 2794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.856891902586781e-06,
+      "loss": 0.7233,
+      "step": 2795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8555415733984802e-06,
+      "loss": 0.8165,
+      "step": 2796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.854191138334147e-06,
+      "loss": 0.7931,
+      "step": 2797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8528405977959224e-06,
+      "loss": 0.7747,
+      "step": 2798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.851489952185982e-06,
+      "loss": 0.7827,
+      "step": 2799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.850139201906531e-06,
+      "loss": 0.7691,
+      "step": 2800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.848788347359808e-06,
+      "loss": 0.8036,
+      "step": 2801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.847437388948081e-06,
+      "loss": 0.8494,
+      "step": 2802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.846086327073649e-06,
+      "loss": 0.7915,
+      "step": 2803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8447351621388417e-06,
+      "loss": 0.8024,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8433838945460207e-06,
+      "loss": 0.7484,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8420325246975773e-06,
+      "loss": 0.6957,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.840681052995933e-06,
+      "loss": 0.7641,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8393294798435393e-06,
+      "loss": 0.7748,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.83797780564288e-06,
+      "loss": 0.7857,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8366260307964666e-06,
+      "loss": 0.8345,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8352741557068415e-06,
+      "loss": 0.5541,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8339221807765764e-06,
+      "loss": 0.7033,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.832570106408275e-06,
+      "loss": 0.8748,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8312179330045674e-06,
+      "loss": 0.7486,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.829865660968115e-06,
+      "loss": 0.8442,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8285132907016087e-06,
+      "loss": 0.7552,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8271608226077684e-06,
+      "loss": 0.7668,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.825808257089342e-06,
+      "loss": 0.8077,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.824455594549108e-06,
+      "loss": 0.7844,
+      "step": 2819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8231028353898727e-06,
+      "loss": 0.7548,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8217499800144725e-06,
+      "loss": 0.74,
+      "step": 2821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8203970288257702e-06,
+      "loss": 0.7699,
+      "step": 2822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.81904398222666e-06,
+      "loss": 0.8571,
+      "step": 2823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.817690840620062e-06,
+      "loss": 0.782,
+      "step": 2824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8163376044089264e-06,
+      "loss": 0.8604,
+      "step": 2825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8149842739962295e-06,
+      "loss": 0.8386,
+      "step": 2826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8136308497849785e-06,
+      "loss": 0.5419,
+      "step": 2827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.812277332178206e-06,
+      "loss": 0.7537,
+      "step": 2828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8109237215789737e-06,
+      "loss": 0.8252,
+      "step": 2829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.80957001839037e-06,
+      "loss": 0.7822,
+      "step": 2830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8082162230155123e-06,
+      "loss": 0.5469,
+      "step": 2831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8068623358575447e-06,
+      "loss": 0.7335,
+      "step": 2832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.805508357319638e-06,
+      "loss": 0.7645,
+      "step": 2833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8041542878049922e-06,
+      "loss": 0.7914,
+      "step": 2834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8028001277168302e-06,
+      "loss": 0.7015,
+      "step": 2835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8014458774584086e-06,
+      "loss": 0.8203,
+      "step": 2836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.800091537433004e-06,
+      "loss": 0.8296,
+      "step": 2837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.798737108043924e-06,
+      "loss": 0.746,
+      "step": 2838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7973825896945013e-06,
+      "loss": 0.8138,
+      "step": 2839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7960279827880943e-06,
+      "loss": 0.8355,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7946732877280903e-06,
+      "loss": 0.7335,
+      "step": 2841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7933185049179e-06,
+      "loss": 0.5571,
+      "step": 2842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7919636347609622e-06,
+      "loss": 0.833,
+      "step": 2843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7906086776607412e-06,
+      "loss": 0.5526,
+      "step": 2844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7892536340207268e-06,
+      "loss": 0.8528,
+      "step": 2845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7878985042444343e-06,
+      "loss": 0.7679,
+      "step": 2846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7865432887354065e-06,
+      "loss": 0.8258,
+      "step": 2847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7851879878972092e-06,
+      "loss": 0.5383,
+      "step": 2848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.783832602133435e-06,
+      "loss": 0.8554,
+      "step": 2849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7824771318477023e-06,
+      "loss": 0.8271,
+      "step": 2850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7811215774436527e-06,
+      "loss": 0.7996,
+      "step": 2851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7797659393249556e-06,
+      "loss": 0.7549,
+      "step": 2852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7784102178953033e-06,
+      "loss": 0.7848,
+      "step": 2853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.777054413558412e-06,
+      "loss": 0.8066,
+      "step": 2854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7756985267180263e-06,
+      "loss": 0.7904,
+      "step": 2855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7743425577779116e-06,
+      "loss": 0.8076,
+      "step": 2856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7729865071418605e-06,
+      "loss": 0.5593,
+      "step": 2857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7716303752136865e-06,
+      "loss": 0.7839,
+      "step": 2858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7702741623972322e-06,
+      "loss": 0.51,
+      "step": 2859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.76891786909636e-06,
+      "loss": 0.7992,
+      "step": 2860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.767561495714957e-06,
+      "loss": 0.7043,
+      "step": 2861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7662050426569377e-06,
+      "loss": 0.7471,
+      "step": 2862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.764848510326235e-06,
+      "loss": 0.8007,
+      "step": 2863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.76349189912681e-06,
+      "loss": 0.7143,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.762135209462643e-06,
+      "loss": 0.7817,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.760778441737743e-06,
+      "loss": 0.7637,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7594215963561373e-06,
+      "loss": 0.8649,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7580646737218774e-06,
+      "loss": 0.8289,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7567076742390406e-06,
+      "loss": 0.8002,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7553505983117247e-06,
+      "loss": 0.7772,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.75399344634405e-06,
+      "loss": 0.7619,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7526362187401603e-06,
+      "loss": 0.811,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7512789159042217e-06,
+      "loss": 0.529,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7499215382404232e-06,
+      "loss": 0.772,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7485640861529756e-06,
+      "loss": 0.7794,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.747206560046112e-06,
+      "loss": 0.8271,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7458489603240873e-06,
+      "loss": 0.7652,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7444912873911784e-06,
+      "loss": 0.688,
+      "step": 2878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7431335416516835e-06,
+      "loss": 0.7214,
+      "step": 2879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7417757235099247e-06,
+      "loss": 0.7558,
+      "step": 2880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7404178333702426e-06,
+      "loss": 0.7789,
+      "step": 2881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.739059871637001e-06,
+      "loss": 0.8495,
+      "step": 2882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7377018387145847e-06,
+      "loss": 0.8091,
+      "step": 2883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7363437350073996e-06,
+      "loss": 0.7567,
+      "step": 2884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7349855609198738e-06,
+      "loss": 0.7761,
+      "step": 2885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7336273168564536e-06,
+      "loss": 0.7471,
+      "step": 2886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.732269003221609e-06,
+      "loss": 0.7657,
+      "step": 2887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.730910620419829e-06,
+      "loss": 0.6981,
+      "step": 2888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.729552168855624e-06,
+      "loss": 0.7121,
+      "step": 2889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7281936489335254e-06,
+      "loss": 0.8445,
+      "step": 2890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.726835061058083e-06,
+      "loss": 0.7808,
+      "step": 2891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7254764056338693e-06,
+      "loss": 0.7782,
+      "step": 2892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.724117683065474e-06,
+      "loss": 0.8461,
+      "step": 2893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.72275889375751e-06,
+      "loss": 0.8036,
+      "step": 2894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.721400038114607e-06,
+      "loss": 0.782,
+      "step": 2895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7200411165414187e-06,
+      "loss": 0.5611,
+      "step": 2896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7186821294426118e-06,
+      "loss": 0.8517,
+      "step": 2897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.71732307722288e-06,
+      "loss": 0.4664,
+      "step": 2898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7159639602869308e-06,
+      "loss": 0.7426,
+      "step": 2899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7146047790394936e-06,
+      "loss": 0.7247,
+      "step": 2900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7132455338853163e-06,
+      "loss": 0.8721,
+      "step": 2901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.711886225229166e-06,
+      "loss": 0.6909,
+      "step": 2902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7105268534758284e-06,
+      "loss": 0.6768,
+      "step": 2903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7091674190301077e-06,
+      "loss": 0.7956,
+      "step": 2904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7078079222968277e-06,
+      "loss": 0.8054,
+      "step": 2905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7064483636808314e-06,
+      "loss": 0.8208,
+      "step": 2906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.705088743586977e-06,
+      "loss": 0.8182,
+      "step": 2907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7037290624201447e-06,
+      "loss": 0.882,
+      "step": 2908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.702369320585231e-06,
+      "loss": 0.7936,
+      "step": 2909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.70100951848715e-06,
+      "loss": 0.8148,
+      "step": 2910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6996496565308356e-06,
+      "loss": 0.7358,
+      "step": 2911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6982897351212377e-06,
+      "loss": 0.5359,
+      "step": 2912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.696929754663325e-06,
+      "loss": 0.7787,
+      "step": 2913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.695569715562083e-06,
+      "loss": 0.7691,
+      "step": 2914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6942096182225163e-06,
+      "loss": 0.8498,
+      "step": 2915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6928494630496436e-06,
+      "loss": 0.8587,
+      "step": 2916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6914892504485045e-06,
+      "loss": 0.8699,
+      "step": 2917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.690128980824154e-06,
+      "loss": 0.8611,
+      "step": 2918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.688768654581664e-06,
+      "loss": 0.8524,
+      "step": 2919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6874082721261226e-06,
+      "loss": 0.7622,
+      "step": 2920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.686047833862636e-06,
+      "loss": 0.7637,
+      "step": 2921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.684687340196327e-06,
+      "loss": 0.7705,
+      "step": 2922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.683326791532333e-06,
+      "loss": 0.8152,
+      "step": 2923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6819661882758114e-06,
+      "loss": 0.8313,
+      "step": 2924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6806055308319305e-06,
+      "loss": 0.7964,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6792448196058807e-06,
+      "loss": 0.7977,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.677884055002864e-06,
+      "loss": 0.8084,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6765232374281006e-06,
+      "loss": 0.6849,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.675162367286825e-06,
+      "loss": 0.7811,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6738014449842886e-06,
+      "loss": 0.7585,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.672440470925757e-06,
+      "loss": 0.8619,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6710794455165135e-06,
+      "loss": 0.7742,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.669718369161854e-06,
+      "loss": 0.5077,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6683572422670914e-06,
+      "loss": 0.8565,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6669960652375526e-06,
+      "loss": 0.7993,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.665634838478579e-06,
+      "loss": 0.7788,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6642735623955306e-06,
+      "loss": 0.8132,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6629122373937765e-06,
+      "loss": 0.7604,
+      "step": 2938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6615508638787037e-06,
+      "loss": 0.774,
+      "step": 2939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6601894422557133e-06,
+      "loss": 0.8113,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6588279729302197e-06,
+      "loss": 0.8076,
+      "step": 2941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.657466456307653e-06,
+      "loss": 0.798,
+      "step": 2942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6561048927934562e-06,
+      "loss": 0.5739,
+      "step": 2943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.654743282793087e-06,
+      "loss": 0.7594,
+      "step": 2944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6533816267120162e-06,
+      "loss": 0.7299,
+      "step": 2945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.652019924955729e-06,
+      "loss": 0.8522,
+      "step": 2946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.650658177929723e-06,
+      "loss": 0.7723,
+      "step": 2947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6492963860395116e-06,
+      "loss": 0.8335,
+      "step": 2948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6479345496906194e-06,
+      "loss": 0.7587,
+      "step": 2949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.646572669288585e-06,
+      "loss": 0.8022,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6452107452389603e-06,
+      "loss": 0.7529,
+      "step": 2951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6438487779473104e-06,
+      "loss": 0.8399,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642486767819212e-06,
+      "loss": 0.756,
+      "step": 2953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6411247152602564e-06,
+      "loss": 0.8256,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639762620676045e-06,
+      "loss": 0.8182,
+      "step": 2955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6384004844721955e-06,
+      "loss": 0.786,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637038307054334e-06,
+      "loss": 0.8223,
+      "step": 2957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.635676088828102e-06,
+      "loss": 0.8286,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.63431383019915e-06,
+      "loss": 0.824,
+      "step": 2959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6329515315731453e-06,
+      "loss": 0.7247,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6315891933557606e-06,
+      "loss": 0.7694,
+      "step": 2961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6302268159526865e-06,
+      "loss": 0.7992,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6288643997696223e-06,
+      "loss": 0.7417,
+      "step": 2963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627501945212278e-06,
+      "loss": 0.7939,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.626139452686378e-06,
+      "loss": 0.7476,
+      "step": 2965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6247769225976544e-06,
+      "loss": 0.7727,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6234143553518548e-06,
+      "loss": 0.7018,
+      "step": 2967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.622051751354734e-06,
+      "loss": 0.7657,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.620689111012059e-06,
+      "loss": 0.7975,
+      "step": 2969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6193264347296084e-06,
+      "loss": 0.7769,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6179637229131698e-06,
+      "loss": 0.8485,
+      "step": 2971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.616600975968544e-06,
+      "loss": 0.8492,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.61523819430154e-06,
+      "loss": 0.817,
+      "step": 2973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6138753783179786e-06,
+      "loss": 0.7476,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.612512528423689e-06,
+      "loss": 0.7643,
+      "step": 2975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6111496450245126e-06,
+      "loss": 0.7987,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.609786728526299e-06,
+      "loss": 0.8027,
+      "step": 2977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6084237793349095e-06,
+      "loss": 0.8144,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.607060797856213e-06,
+      "loss": 0.8026,
+      "step": 2979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6056977844960897e-06,
+      "loss": 0.7961,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.604334739660428e-06,
+      "loss": 0.7333,
+      "step": 2981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.602971663755127e-06,
+      "loss": 0.8372,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.601608557186095e-06,
+      "loss": 0.762,
+      "step": 2983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.600245420359249e-06,
+      "loss": 0.7855,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.598882253680512e-06,
+      "loss": 0.7726,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5975190575558223e-06,
+      "loss": 0.7756,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5961558323911205e-06,
+      "loss": 0.8377,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5947925785923624e-06,
+      "loss": 0.777,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5934292965655046e-06,
+      "loss": 0.7503,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5920659867165194e-06,
+      "loss": 0.8007,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5907026494513815e-06,
+      "loss": 0.8214,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.589339285176079e-06,
+      "loss": 0.7883,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5879758942966043e-06,
+      "loss": 0.7897,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5866124772189582e-06,
+      "loss": 0.7626,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.585249034349152e-06,
+      "loss": 0.7766,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5838855660932006e-06,
+      "loss": 0.8092,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5825220728571303e-06,
+      "loss": 0.7497,
+      "step": 2997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5811585550469714e-06,
+      "loss": 0.7456,
+      "step": 2998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.579795013068765e-06,
+      "loss": 0.7616,
+      "step": 2999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5784314473285565e-06,
+      "loss": 0.817,
+      "step": 3000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5770678582323995e-06,
+      "loss": 0.792,
+      "step": 3001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5757042461863546e-06,
+      "loss": 0.7633,
+      "step": 3002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.574340611596489e-06,
+      "loss": 0.8441,
+      "step": 3003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.572976954868878e-06,
+      "loss": 0.7203,
+      "step": 3004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.571613276409601e-06,
+      "loss": 0.744,
+      "step": 3005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570249576624746e-06,
+      "loss": 0.7909,
+      "step": 3006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5688858559204056e-06,
+      "loss": 0.5208,
+      "step": 3007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.56752211470268e-06,
+      "loss": 0.7613,
+      "step": 3008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.566158353377675e-06,
+      "loss": 0.7461,
+      "step": 3009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5647945723515028e-06,
+      "loss": 0.7752,
+      "step": 3010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5634307720302807e-06,
+      "loss": 0.7965,
+      "step": 3011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5620669528201324e-06,
+      "loss": 0.9,
+      "step": 3012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5607031151271865e-06,
+      "loss": 0.8056,
+      "step": 3013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5593392593575784e-06,
+      "loss": 0.7155,
+      "step": 3014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.557975385917447e-06,
+      "loss": 0.7631,
+      "step": 3015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.556611495212939e-06,
+      "loss": 0.7645,
+      "step": 3016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5552475876502024e-06,
+      "loss": 0.7724,
+      "step": 3017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5538836636353958e-06,
+      "loss": 0.7517,
+      "step": 3018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.552519723574676e-06,
+      "loss": 0.7875,
+      "step": 3019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5511557678742104e-06,
+      "loss": 0.7328,
+      "step": 3020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5497917969401685e-06,
+      "loss": 0.7235,
+      "step": 3021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.548427811178724e-06,
+      "loss": 0.8027,
+      "step": 3022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.547063810996056e-06,
+      "loss": 0.7662,
+      "step": 3023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5456997967983465e-06,
+      "loss": 0.7475,
+      "step": 3024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.544335768991784e-06,
+      "loss": 0.8287,
+      "step": 3025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5429717279825583e-06,
+      "loss": 0.7216,
+      "step": 3026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.541607674176866e-06,
+      "loss": 0.7028,
+      "step": 3027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.540243607980904e-06,
+      "loss": 0.8279,
+      "step": 3028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.538879529800878e-06,
+      "loss": 0.779,
+      "step": 3029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5375154400429907e-06,
+      "loss": 0.7133,
+      "step": 3030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5361513391134547e-06,
+      "loss": 0.7931,
+      "step": 3031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.534787227418481e-06,
+      "loss": 0.526,
+      "step": 3032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5334231053642864e-06,
+      "loss": 0.8709,
+      "step": 3033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5320589733570905e-06,
+      "loss": 0.7779,
+      "step": 3034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.530694831803115e-06,
+      "loss": 0.7904,
+      "step": 3035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.529330681108585e-06,
+      "loss": 0.763,
+      "step": 3036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.527966521679729e-06,
+      "loss": 0.8306,
+      "step": 3037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5266023539227764e-06,
+      "loss": 0.7818,
+      "step": 3038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5252381782439605e-06,
+      "loss": 0.6853,
+      "step": 3039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5238739950495166e-06,
+      "loss": 0.7878,
+      "step": 3040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5225098047456825e-06,
+      "loss": 0.6829,
+      "step": 3041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.521145607738697e-06,
+      "loss": 0.8252,
+      "step": 3042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5197814044348022e-06,
+      "loss": 0.7718,
+      "step": 3043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5184171952402415e-06,
+      "loss": 0.8033,
+      "step": 3044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.51705298056126e-06,
+      "loss": 0.7074,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5156887608041047e-06,
+      "loss": 0.7468,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.514324536375023e-06,
+      "loss": 0.7516,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.512960307680266e-06,
+      "loss": 0.7642,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5115960751260843e-06,
+      "loss": 0.7732,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5102318391187296e-06,
+      "loss": 0.8057,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.508867600064456e-06,
+      "loss": 0.7147,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5075033583695162e-06,
+      "loss": 0.7713,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5061391144401664e-06,
+      "loss": 0.6208,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5047748686826616e-06,
+      "loss": 0.7767,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.503410621503258e-06,
+      "loss": 0.8065,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.502046373308212e-06,
+      "loss": 0.8037,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5006821245037805e-06,
+      "loss": 0.7564,
+      "step": 3057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.49931787549622e-06,
+      "loss": 0.7489,
+      "step": 3058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4979536266917885e-06,
+      "loss": 0.8079,
+      "step": 3059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.496589378496743e-06,
+      "loss": 0.7782,
+      "step": 3060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4952251313173393e-06,
+      "loss": 0.7619,
+      "step": 3061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4938608855598336e-06,
+      "loss": 0.8026,
+      "step": 3062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4924966416304846e-06,
+      "loss": 0.7098,
+      "step": 3063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.491132399935545e-06,
+      "loss": 0.8045,
+      "step": 3064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.489768160881271e-06,
+      "loss": 0.7003,
+      "step": 3065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.488403924873916e-06,
+      "loss": 0.7097,
+      "step": 3066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4870396923197347e-06,
+      "loss": 0.7758,
+      "step": 3067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4856754636249776e-06,
+      "loss": 0.7828,
+      "step": 3068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.484311239195896e-06,
+      "loss": 0.8071,
+      "step": 3069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.482947019438742e-06,
+      "loss": 0.6824,
+      "step": 3070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4815828047597597e-06,
+      "loss": 0.7742,
+      "step": 3071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4802185955651986e-06,
+      "loss": 0.7537,
+      "step": 3072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.478854392261303e-06,
+      "loss": 0.7768,
+      "step": 3073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4774901952543188e-06,
+      "loss": 0.7629,
+      "step": 3074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.476126004950484e-06,
+      "loss": 0.7774,
+      "step": 3075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4747618217560403e-06,
+      "loss": 0.7092,
+      "step": 3076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.473397646077225e-06,
+      "loss": 0.7773,
+      "step": 3077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.472033478320272e-06,
+      "loss": 0.7712,
+      "step": 3078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4706693188914156e-06,
+      "loss": 0.7366,
+      "step": 3079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.469305168196886e-06,
+      "loss": 0.7802,
+      "step": 3080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4679410266429112e-06,
+      "loss": 0.7134,
+      "step": 3081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4665768946357144e-06,
+      "loss": 0.7577,
+      "step": 3082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4652127725815198e-06,
+      "loss": 0.7961,
+      "step": 3083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.463848660886546e-06,
+      "loss": 0.7464,
+      "step": 3084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.46248455995701e-06,
+      "loss": 0.7646,
+      "step": 3085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.461120470199123e-06,
+      "loss": 0.6864,
+      "step": 3086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.459756392019096e-06,
+      "loss": 0.7906,
+      "step": 3087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4583923258231352e-06,
+      "loss": 0.7984,
+      "step": 3088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4570282720174425e-06,
+      "loss": 0.8125,
+      "step": 3089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4556642310082166e-06,
+      "loss": 0.8085,
+      "step": 3090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.454300203201654e-06,
+      "loss": 0.7674,
+      "step": 3091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.452936189003945e-06,
+      "loss": 0.8317,
+      "step": 3092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4515721888212767e-06,
+      "loss": 0.7679,
+      "step": 3093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4502082030598315e-06,
+      "loss": 0.774,
+      "step": 3094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.44884423212579e-06,
+      "loss": 0.7142,
+      "step": 3095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4474802764253244e-06,
+      "loss": 0.7485,
+      "step": 3096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4461163363646046e-06,
+      "loss": 0.8498,
+      "step": 3097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.444752412349797e-06,
+      "loss": 0.7107,
+      "step": 3098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.443388504787062e-06,
+      "loss": 0.8127,
+      "step": 3099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4420246140825533e-06,
+      "loss": 0.7814,
+      "step": 3100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4406607406424216e-06,
+      "loss": 0.5466,
+      "step": 3101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4392968848728147e-06,
+      "loss": 0.5385,
+      "step": 3102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4379330471798684e-06,
+      "loss": 0.8243,
+      "step": 3103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.43656922796972e-06,
+      "loss": 0.7879,
+      "step": 3104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.435205427648497e-06,
+      "loss": 0.8588,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4338416466223257e-06,
+      "loss": 0.5112,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4324778852973206e-06,
+      "loss": 0.8299,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4311141440795956e-06,
+      "loss": 0.8636,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4297504233752558e-06,
+      "loss": 0.8202,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4283867235903995e-06,
+      "loss": 0.7053,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4270230451311227e-06,
+      "loss": 0.7668,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4256593884035112e-06,
+      "loss": 0.7965,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4242957538136467e-06,
+      "loss": 0.7558,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4229321417676013e-06,
+      "loss": 0.6683,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4215685526714444e-06,
+      "loss": 0.717,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.420204986931236e-06,
+      "loss": 0.8087,
+      "step": 3116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4188414449530294e-06,
+      "loss": 0.5672,
+      "step": 3117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.41747792714287e-06,
+      "loss": 0.808,
+      "step": 3118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4161144339068e-06,
+      "loss": 0.789,
+      "step": 3119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4147509656508487e-06,
+      "loss": 0.7687,
+      "step": 3120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.413387522781042e-06,
+      "loss": 0.7844,
+      "step": 3121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.412024105703396e-06,
+      "loss": 0.8517,
+      "step": 3122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4106607148239213e-06,
+      "loss": 0.8133,
+      "step": 3123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4092973505486193e-06,
+      "loss": 0.8056,
+      "step": 3124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4079340132834815e-06,
+      "loss": 0.7438,
+      "step": 3125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.406570703434496e-06,
+      "loss": 0.7044,
+      "step": 3126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4052074214076393e-06,
+      "loss": 0.8052,
+      "step": 3127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.40384416760888e-06,
+      "loss": 0.8233,
+      "step": 3128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4024809424441786e-06,
+      "loss": 0.7545,
+      "step": 3129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.401117746319488e-06,
+      "loss": 0.5591,
+      "step": 3130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.399754579640753e-06,
+      "loss": 0.7363,
+      "step": 3131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3983914428139058e-06,
+      "loss": 0.7821,
+      "step": 3132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.397028336244873e-06,
+      "loss": 0.8073,
+      "step": 3133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.395665260339573e-06,
+      "loss": 0.8111,
+      "step": 3134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3943022155039115e-06,
+      "loss": 0.7951,
+      "step": 3135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.392939202143788e-06,
+      "loss": 0.5647,
+      "step": 3136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.391576220665091e-06,
+      "loss": 0.7168,
+      "step": 3137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3902132714737014e-06,
+      "loss": 0.7349,
+      "step": 3138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3888503549754883e-06,
+      "loss": 0.7244,
+      "step": 3139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3874874715763115e-06,
+      "loss": 0.7935,
+      "step": 3140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3861246216820226e-06,
+      "loss": 0.7822,
+      "step": 3141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3847618056984603e-06,
+      "loss": 0.7942,
+      "step": 3142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3833990240314564e-06,
+      "loss": 0.7885,
+      "step": 3143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.38203627708683e-06,
+      "loss": 0.7676,
+      "step": 3144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3806735652703933e-06,
+      "loss": 0.8346,
+      "step": 3145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.379310888987942e-06,
+      "loss": 0.8047,
+      "step": 3146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.377948248645267e-06,
+      "loss": 0.8289,
+      "step": 3147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.376585644648146e-06,
+      "loss": 0.7143,
+      "step": 3148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375223077402346e-06,
+      "loss": 0.8077,
+      "step": 3149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.373860547313623e-06,
+      "loss": 0.8016,
+      "step": 3150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3724980547877225e-06,
+      "loss": 0.8021,
+      "step": 3151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3711356002303794e-06,
+      "loss": 0.7645,
+      "step": 3152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.369773184047314e-06,
+      "loss": 0.7714,
+      "step": 3153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36841080664424e-06,
+      "loss": 0.8844,
+      "step": 3154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3670484684268563e-06,
+      "loss": 0.7565,
+      "step": 3155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3656861698008504e-06,
+      "loss": 0.7058,
+      "step": 3156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.364323911171899e-06,
+      "loss": 0.8625,
+      "step": 3157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3629616929456666e-06,
+      "loss": 0.7739,
+      "step": 3158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3615995155278053e-06,
+      "loss": 0.7512,
+      "step": 3159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3602373793239554e-06,
+      "loss": 0.762,
+      "step": 3160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.358875284739744e-06,
+      "loss": 0.6782,
+      "step": 3161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.357513232180788e-06,
+      "loss": 0.7901,
+      "step": 3162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3561512220526904e-06,
+      "loss": 0.7522,
+      "step": 3163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35478925476104e-06,
+      "loss": 0.7782,
+      "step": 3164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.353427330711415e-06,
+      "loss": 0.8134,
+      "step": 3165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.352065450309381e-06,
+      "loss": 0.7432,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3507036139604893e-06,
+      "loss": 0.7241,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3493418220702775e-06,
+      "loss": 1.3204,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3479800750442715e-06,
+      "loss": 0.7702,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3466183732879846e-06,
+      "loss": 0.7783,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.345256717206914e-06,
+      "loss": 0.8118,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3438951072065437e-06,
+      "loss": 0.7962,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.342533543692348e-06,
+      "loss": 0.7466,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3411720270697807e-06,
+      "loss": 0.7683,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3398105577442875e-06,
+      "loss": 0.8834,
+      "step": 3175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3384491361212963e-06,
+      "loss": 0.8322,
+      "step": 3176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3370877626062248e-06,
+      "loss": 0.8512,
+      "step": 3177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.33572643760447e-06,
+      "loss": 0.7686,
+      "step": 3178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.334365161521421e-06,
+      "loss": 1.1059,
+      "step": 3179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.333003934762449e-06,
+      "loss": 0.8319,
+      "step": 3180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.331642757732909e-06,
+      "loss": 0.7979,
+      "step": 3181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3302816308381463e-06,
+      "loss": 0.7754,
+      "step": 3182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3289205544834873e-06,
+      "loss": 0.7914,
+      "step": 3183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.327559529074244e-06,
+      "loss": 0.7395,
+      "step": 3184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3261985550157123e-06,
+      "loss": 0.7702,
+      "step": 3185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.324837632713176e-06,
+      "loss": 0.792,
+      "step": 3186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3234767625719006e-06,
+      "loss": 0.7744,
+      "step": 3187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.322115944997137e-06,
+      "loss": 0.8179,
+      "step": 3188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3207551803941197e-06,
+      "loss": 0.8193,
+      "step": 3189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31939446916807e-06,
+      "loss": 0.8319,
+      "step": 3190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3180338117241903e-06,
+      "loss": 0.8018,
+      "step": 3191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3166732084676676e-06,
+      "loss": 0.8314,
+      "step": 3192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3153126598036733e-06,
+      "loss": 0.7551,
+      "step": 3193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3139521661373647e-06,
+      "loss": 0.8385,
+      "step": 3194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3125917278738782e-06,
+      "loss": 0.7598,
+      "step": 3195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.311231345418337e-06,
+      "loss": 0.7948,
+      "step": 3196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3098710191758463e-06,
+      "loss": 0.7756,
+      "step": 3197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.308510749551496e-06,
+      "loss": 0.7931,
+      "step": 3198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.307150536950357e-06,
+      "loss": 0.7727,
+      "step": 3199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3057903817774845e-06,
+      "loss": 0.7685,
+      "step": 3200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.304430284437917e-06,
+      "loss": 0.8085,
+      "step": 3201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.303070245336676e-06,
+      "loss": 0.7465,
+      "step": 3202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.301710264878763e-06,
+      "loss": 0.8364,
+      "step": 3203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3003503434691648e-06,
+      "loss": 0.8167,
+      "step": 3204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.298990481512851e-06,
+      "loss": 0.7849,
+      "step": 3205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29763067941477e-06,
+      "loss": 0.8025,
+      "step": 3206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2962709375798557e-06,
+      "loss": 0.7429,
+      "step": 3207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.294911256413023e-06,
+      "loss": 0.7698,
+      "step": 3208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2935516363191695e-06,
+      "loss": 0.8317,
+      "step": 3209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2921920777031727e-06,
+      "loss": 0.8041,
+      "step": 3210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2908325809698927e-06,
+      "loss": 0.6446,
+      "step": 3211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2894731465241733e-06,
+      "loss": 0.7774,
+      "step": 3212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2881137747708347e-06,
+      "loss": 2.3385,
+      "step": 3213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.286754466114684e-06,
+      "loss": 0.7923,
+      "step": 3214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2853952209605068e-06,
+      "loss": 0.7903,
+      "step": 3215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2840360397130705e-06,
+      "loss": 0.7515,
+      "step": 3216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2826769227771207e-06,
+      "loss": 0.8387,
+      "step": 3217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2813178705573882e-06,
+      "loss": 0.8431,
+      "step": 3218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.279958883458583e-06,
+      "loss": 0.7997,
+      "step": 3219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2785999618853932e-06,
+      "loss": 0.7329,
+      "step": 3220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2772411062424905e-06,
+      "loss": 0.7782,
+      "step": 3221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2758823169345263e-06,
+      "loss": 0.8813,
+      "step": 3222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.274523594366132e-06,
+      "loss": 0.8237,
+      "step": 3223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2731649389419173e-06,
+      "loss": 0.7845,
+      "step": 3224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.271806351066475e-06,
+      "loss": 0.7364,
+      "step": 3225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.270447831144376e-06,
+      "loss": 0.762,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2690893795801715e-06,
+      "loss": 0.826,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2677309967783916e-06,
+      "loss": 0.8434,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.266372683143547e-06,
+      "loss": 0.7453,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2650144390801275e-06,
+      "loss": 0.7674,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.263656264992601e-06,
+      "loss": 0.819,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2622981612854157e-06,
+      "loss": 0.7984,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2609401283630007e-06,
+      "loss": 0.8491,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2595821666297587e-06,
+      "loss": 0.7723,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2582242764900765e-06,
+      "loss": 0.8498,
+      "step": 3235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2568664583483165e-06,
+      "loss": 2.2165,
+      "step": 3236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.255508712608823e-06,
+      "loss": 0.7859,
+      "step": 3237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2541510396759136e-06,
+      "loss": 0.8139,
+      "step": 3238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2527934399538886e-06,
+      "loss": 0.7738,
+      "step": 3239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2514359138470244e-06,
+      "loss": 0.8015,
+      "step": 3240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.250078461759577e-06,
+      "loss": 0.8341,
+      "step": 3241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.248721084095779e-06,
+      "loss": 0.8414,
+      "step": 3242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2473637812598406e-06,
+      "loss": 0.7042,
+      "step": 3243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2460065536559518e-06,
+      "loss": 0.7512,
+      "step": 3244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2446494016882765e-06,
+      "loss": 0.7171,
+      "step": 3245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2432923257609598e-06,
+      "loss": 0.7394,
+      "step": 3246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2419353262781225e-06,
+      "loss": 0.6876,
+      "step": 3247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2405784036438644e-06,
+      "loss": 0.8039,
+      "step": 3248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2392215582622576e-06,
+      "loss": 0.8079,
+      "step": 3249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.237864790537357e-06,
+      "loss": 0.8203,
+      "step": 3250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2365081008731913e-06,
+      "loss": 0.6836,
+      "step": 3251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.235151489673766e-06,
+      "loss": 0.8093,
+      "step": 3252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2337949573430627e-06,
+      "loss": 0.755,
+      "step": 3253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.232438504285043e-06,
+      "loss": 0.8231,
+      "step": 3254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2310821309036413e-06,
+      "loss": 0.756,
+      "step": 3255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2297258376027686e-06,
+      "loss": 0.7517,
+      "step": 3256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.228369624786314e-06,
+      "loss": 0.8225,
+      "step": 3257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.227013492858141e-06,
+      "loss": 0.7782,
+      "step": 3258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2256574422220896e-06,
+      "loss": 2.5253,
+      "step": 3259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.224301473281974e-06,
+      "loss": 1.8627,
+      "step": 3260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.222945586441588e-06,
+      "loss": 0.8158,
+      "step": 3261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.221589782104698e-06,
+      "loss": 1.305,
+      "step": 3262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2202340606750453e-06,
+      "loss": 0.8408,
+      "step": 3263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.218878422556347e-06,
+      "loss": 0.7939,
+      "step": 3264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2175228681522985e-06,
+      "loss": 0.7749,
+      "step": 3265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2161673978665654e-06,
+      "loss": 0.7861,
+      "step": 3266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.214812012102791e-06,
+      "loss": 0.8212,
+      "step": 3267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2134567112645935e-06,
+      "loss": 0.8069,
+      "step": 3268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.212101495755566e-06,
+      "loss": 0.7874,
+      "step": 3269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2107463659792737e-06,
+      "loss": 0.7775,
+      "step": 3270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.209391322339259e-06,
+      "loss": 0.8054,
+      "step": 3271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.208036365239039e-06,
+      "loss": 0.8139,
+      "step": 3272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2066814950821007e-06,
+      "loss": 0.8439,
+      "step": 3273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.205326712271911e-06,
+      "loss": 0.7412,
+      "step": 3274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2039720172119056e-06,
+      "loss": 0.7589,
+      "step": 3275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2026174103055004e-06,
+      "loss": 0.7965,
+      "step": 3276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2012628919560768e-06,
+      "loss": 0.7521,
+      "step": 3277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1999084625669965e-06,
+      "loss": 0.77,
+      "step": 3278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1985541225415914e-06,
+      "loss": 0.8174,
+      "step": 3279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.19719987228317e-06,
+      "loss": 0.817,
+      "step": 3280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.195845712195009e-06,
+      "loss": 0.7927,
+      "step": 3281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1944916426803626e-06,
+      "loss": 0.8568,
+      "step": 3282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1931376641424566e-06,
+      "loss": 0.7915,
+      "step": 3283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.191783776984488e-06,
+      "loss": 0.807,
+      "step": 3284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.190429981609631e-06,
+      "loss": 0.7485,
+      "step": 3285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1890762784210275e-06,
+      "loss": 0.7157,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.187722667821795e-06,
+      "loss": 0.7497,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.186369150215022e-06,
+      "loss": 0.7969,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.185015726003771e-06,
+      "loss": 0.7606,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1836623955910744e-06,
+      "loss": 0.8156,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1823091593799385e-06,
+      "loss": 0.7594,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.18095601777334e-06,
+      "loss": 0.7582,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1796029711742297e-06,
+      "loss": 0.7819,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1782500199855284e-06,
+      "loss": 0.7726,
+      "step": 3294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1768971646101277e-06,
+      "loss": 0.7621,
+      "step": 3295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1755444054508924e-06,
+      "loss": 0.6786,
+      "step": 3296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.174191742910659e-06,
+      "loss": 0.774,
+      "step": 3297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1728391773922325e-06,
+      "loss": 0.7805,
+      "step": 3298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1714867092983918e-06,
+      "loss": 0.7901,
+      "step": 3299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.170134339031885e-06,
+      "loss": 0.7908,
+      "step": 3300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1687820669954334e-06,
+      "loss": 0.8012,
+      "step": 3301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1674298935917255e-06,
+      "loss": 0.722,
+      "step": 3302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.166077819223423e-06,
+      "loss": 0.7869,
+      "step": 3303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.16472584429316e-06,
+      "loss": 0.7945,
+      "step": 3304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.163373969203534e-06,
+      "loss": 0.7882,
+      "step": 3305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1620221943571206e-06,
+      "loss": 0.7446,
+      "step": 3306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1606705201564607e-06,
+      "loss": 0.7971,
+      "step": 3307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1593189470040683e-06,
+      "loss": 0.7186,
+      "step": 3308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1579674753024236e-06,
+      "loss": 0.7245,
+      "step": 3309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1566161054539797e-06,
+      "loss": 0.8786,
+      "step": 3310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1552648378611595e-06,
+      "loss": 0.7715,
+      "step": 3311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.153913672926352e-06,
+      "loss": 0.7792,
+      "step": 3312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1525626110519197e-06,
+      "loss": 0.817,
+      "step": 3313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1512116526401928e-06,
+      "loss": 0.8331,
+      "step": 3314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1498607980934703e-06,
+      "loss": 0.8431,
+      "step": 3315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.148510047814019e-06,
+      "loss": 0.7043,
+      "step": 3316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1471594022040785e-06,
+      "loss": 0.7486,
+      "step": 3317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1458088616658536e-06,
+      "loss": 0.801,
+      "step": 3318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.14445842660152e-06,
+      "loss": 0.7123,
+      "step": 3319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1431080974132203e-06,
+      "loss": 0.7965,
+      "step": 3320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1417578745030667e-06,
+      "loss": 0.877,
+      "step": 3321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1404077582731396e-06,
+      "loss": 0.8231,
+      "step": 3322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1390577491254874e-06,
+      "loss": 0.7301,
+      "step": 3323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.137707847462126e-06,
+      "loss": 0.7566,
+      "step": 3324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1363580536850415e-06,
+      "loss": 0.7948,
+      "step": 3325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1350083681961857e-06,
+      "loss": 0.9253,
+      "step": 3326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1336587913974784e-06,
+      "loss": 0.7716,
+      "step": 3327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.132309323690807e-06,
+      "loss": 0.6633,
+      "step": 3328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1309599654780285e-06,
+      "loss": 0.782,
+      "step": 3329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1296107171609643e-06,
+      "loss": 0.8438,
+      "step": 3330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1282615791414046e-06,
+      "loss": 0.7222,
+      "step": 3331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.126912551821107e-06,
+      "loss": 0.7747,
+      "step": 3332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1255636356017965e-06,
+      "loss": 0.8993,
+      "step": 3333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1242148308851627e-06,
+      "loss": 0.745,
+      "step": 3334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1228661380728637e-06,
+      "loss": 0.7891,
+      "step": 3335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.121517557566526e-06,
+      "loss": 0.8015,
+      "step": 3336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.120169089767738e-06,
+      "loss": 0.8282,
+      "step": 3337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1188207350780586e-06,
+      "loss": 0.8227,
+      "step": 3338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1174724938990117e-06,
+      "loss": 0.7556,
+      "step": 3339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1161243666320886e-06,
+      "loss": 0.6831,
+      "step": 3340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1147763536787437e-06,
+      "loss": 0.737,
+      "step": 3341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1134284554404e-06,
+      "loss": 0.833,
+      "step": 3342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1120806723184457e-06,
+      "loss": 0.8707,
+      "step": 3343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1107330047142333e-06,
+      "loss": 0.8345,
+      "step": 3344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1093854530290834e-06,
+      "loss": 0.8353,
+      "step": 3345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1080380176642802e-06,
+      "loss": 0.8601,
+      "step": 3346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.106690699021075e-06,
+      "loss": 0.7336,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1053434975006804e-06,
+      "loss": 0.8141,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.10399641350428e-06,
+      "loss": 0.6992,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1026494474330184e-06,
+      "loss": 0.75,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.101302599688005e-06,
+      "loss": 0.8296,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0999558706703156e-06,
+      "loss": 0.8179,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.09860926078099e-06,
+      "loss": 0.8243,
+      "step": 3353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0972627704210335e-06,
+      "loss": 0.823,
+      "step": 3354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0959163999914135e-06,
+      "loss": 0.6877,
+      "step": 3355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.094570149893063e-06,
+      "loss": 0.7623,
+      "step": 3356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.09322402052688e-06,
+      "loss": 0.7863,
+      "step": 3357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.091878012293726e-06,
+      "loss": 0.7416,
+      "step": 3358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.090532125594425e-06,
+      "loss": 0.7948,
+      "step": 3359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.089186360829767e-06,
+      "loss": 0.8526,
+      "step": 3360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0878407184005045e-06,
+      "loss": 0.778,
+      "step": 3361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.086495198707353e-06,
+      "loss": 0.7538,
+      "step": 3362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0851498021509927e-06,
+      "loss": 0.7796,
+      "step": 3363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0838045291320665e-06,
+      "loss": 0.738,
+      "step": 3364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0824593800511815e-06,
+      "loss": 0.7928,
+      "step": 3365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.081114355308906e-06,
+      "loss": 0.8518,
+      "step": 3366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0797694553057724e-06,
+      "loss": 0.7931,
+      "step": 3367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0784246804422766e-06,
+      "loss": 0.7602,
+      "step": 3368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0770800311188756e-06,
+      "loss": 0.7899,
+      "step": 3369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0757355077359904e-06,
+      "loss": 0.7058,
+      "step": 3370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0743911106940035e-06,
+      "loss": 0.7549,
+      "step": 3371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0730468403932613e-06,
+      "loss": 0.7766,
+      "step": 3372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0717026972340707e-06,
+      "loss": 0.761,
+      "step": 3373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0703586816167014e-06,
+      "loss": 0.7509,
+      "step": 3374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.069014793941386e-06,
+      "loss": 0.8395,
+      "step": 3375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0676710346083165e-06,
+      "loss": 0.7593,
+      "step": 3376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.06632740401765e-06,
+      "loss": 0.7752,
+      "step": 3377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0649839025695026e-06,
+      "loss": 0.7623,
+      "step": 3378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.063640530663955e-06,
+      "loss": 0.8164,
+      "step": 3379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0622972887010443e-06,
+      "loss": 0.7778,
+      "step": 3380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.060954177080774e-06,
+      "loss": 0.7184,
+      "step": 3381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0596111962031068e-06,
+      "loss": 0.7186,
+      "step": 3382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.058268346467965e-06,
+      "loss": 0.7625,
+      "step": 3383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.056925628275233e-06,
+      "loss": 0.8227,
+      "step": 3384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0555830420247583e-06,
+      "loss": 0.8125,
+      "step": 3385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0542405881163463e-06,
+      "loss": 0.7598,
+      "step": 3386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.052898266949762e-06,
+      "loss": 0.821,
+      "step": 3387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0515560789247348e-06,
+      "loss": 0.7193,
+      "step": 3388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0502140244409517e-06,
+      "loss": 0.8223,
+      "step": 3389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0488721038980596e-06,
+      "loss": 0.828,
+      "step": 3390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.047530317695667e-06,
+      "loss": 0.782,
+      "step": 3391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.046188666233342e-06,
+      "loss": 0.7689,
+      "step": 3392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0448471499106126e-06,
+      "loss": 0.824,
+      "step": 3393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0435057691269657e-06,
+      "loss": 0.6767,
+      "step": 3394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.042164524281848e-06,
+      "loss": 0.8419,
+      "step": 3395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0408234157746675e-06,
+      "loss": 0.7862,
+      "step": 3396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.03948244400479e-06,
+      "loss": 0.7464,
+      "step": 3397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0381416093715405e-06,
+      "loss": 0.7904,
+      "step": 3398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.036800912274203e-06,
+      "loss": 0.8193,
+      "step": 3399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0354603531120225e-06,
+      "loss": 0.7774,
+      "step": 3400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0341199322842003e-06,
+      "loss": 0.753,
+      "step": 3401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0327796501898984e-06,
+      "loss": 0.7726,
+      "step": 3402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.031439507228236e-06,
+      "loss": 0.7739,
+      "step": 3403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.030099503798293e-06,
+      "loss": 0.816,
+      "step": 3404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.028759640299105e-06,
+      "loss": 0.7876,
+      "step": 3405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0274199171296676e-06,
+      "loss": 0.686,
+      "step": 3406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0260803346889364e-06,
+      "loss": 0.8161,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.02474089337582e-06,
+      "loss": 0.8187,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.02340159358919e-06,
+      "loss": 0.7787,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0220624357278722e-06,
+      "loss": 0.8206,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0207234201906546e-06,
+      "loss": 0.7584,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.019384547376277e-06,
+      "loss": 0.7753,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0180458176834417e-06,
+      "loss": 0.7775,
+      "step": 3413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.016707231510806e-06,
+      "loss": 0.7537,
+      "step": 3414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.015368789256984e-06,
+      "loss": 0.7065,
+      "step": 3415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0140304913205485e-06,
+      "loss": 0.6463,
+      "step": 3416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0126923381000293e-06,
+      "loss": 0.8333,
+      "step": 3417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0113543299939116e-06,
+      "loss": 0.7279,
+      "step": 3418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0100164674006376e-06,
+      "loss": 0.8274,
+      "step": 3419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.008678750718608e-06,
+      "loss": 0.774,
+      "step": 3420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0073411803461795e-06,
+      "loss": 0.8035,
+      "step": 3421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0060037566816625e-06,
+      "loss": 0.7199,
+      "step": 3422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0046664801233263e-06,
+      "loss": 0.8015,
+      "step": 3423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.003329351069397e-06,
+      "loss": 0.8363,
+      "step": 3424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.001992369918056e-06,
+      "loss": 0.7871,
+      "step": 3425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0006555370674388e-06,
+      "loss": 0.7252,
+      "step": 3426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9993188529156382e-06,
+      "loss": 0.7889,
+      "step": 3427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9979823178607044e-06,
+      "loss": 0.761,
+      "step": 3428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.99664593230064e-06,
+      "loss": 0.8252,
+      "step": 3429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9953096966334055e-06,
+      "loss": 0.772,
+      "step": 3430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.993973611256915e-06,
+      "loss": 0.788,
+      "step": 3431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.992637676569041e-06,
+      "loss": 0.8476,
+      "step": 3432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9913018929676064e-06,
+      "loss": 0.7751,
+      "step": 3433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.989966260850392e-06,
+      "loss": 0.7066,
+      "step": 3434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.988630780615134e-06,
+      "loss": 0.7595,
+      "step": 3435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9872954526595228e-06,
+      "loss": 0.76,
+      "step": 3436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.985960277381202e-06,
+      "loss": 0.8597,
+      "step": 3437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.984625255177771e-06,
+      "loss": 0.7491,
+      "step": 3438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.983290386446785e-06,
+      "loss": 0.7396,
+      "step": 3439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.981955671585749e-06,
+      "loss": 0.7594,
+      "step": 3440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980621110992128e-06,
+      "loss": 0.7614,
+      "step": 3441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.979286705063336e-06,
+      "loss": 0.7988,
+      "step": 3442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9779524541967455e-06,
+      "loss": 0.7223,
+      "step": 3443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9766183587896784e-06,
+      "loss": 0.8069,
+      "step": 3444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9752844192394133e-06,
+      "loss": 0.7829,
+      "step": 3445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.973950635943182e-06,
+      "loss": 0.7547,
+      "step": 3446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9726170092981676e-06,
+      "loss": 0.7953,
+      "step": 3447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9712835397015095e-06,
+      "loss": 0.7649,
+      "step": 3448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9699502275502984e-06,
+      "loss": 0.7894,
+      "step": 3449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9686170732415802e-06,
+      "loss": 0.7378,
+      "step": 3450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9672840771723495e-06,
+      "loss": 0.7845,
+      "step": 3451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.965951239739559e-06,
+      "loss": 0.7309,
+      "step": 3452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9646185613401118e-06,
+      "loss": 0.7995,
+      "step": 3453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.963286042370862e-06,
+      "loss": 0.6168,
+      "step": 3454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619536832286183e-06,
+      "loss": 0.789,
+      "step": 3455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960621484310142e-06,
+      "loss": 0.7734,
+      "step": 3456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9592894460121465e-06,
+      "loss": 0.7378,
+      "step": 3457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9579575687312956e-06,
+      "loss": 0.7969,
+      "step": 3458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9566258528642063e-06,
+      "loss": 0.7951,
+      "step": 3459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9552942988074494e-06,
+      "loss": 0.7888,
+      "step": 3460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9539629069575442e-06,
+      "loss": 0.8133,
+      "step": 3461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9526316777109636e-06,
+      "loss": 0.6336,
+      "step": 3462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9513006114641326e-06,
+      "loss": 0.7452,
+      "step": 3463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.949969708613426e-06,
+      "loss": 0.7639,
+      "step": 3464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9486389695551713e-06,
+      "loss": 0.7067,
+      "step": 3465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9473083946856458e-06,
+      "loss": 0.8315,
+      "step": 3466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9459779844010805e-06,
+      "loss": 0.6642,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9446477390976536e-06,
+      "loss": 0.6536,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9433176591714974e-06,
+      "loss": 0.8052,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9419877450186937e-06,
+      "loss": 0.7932,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.940657997035276e-06,
+      "loss": 0.7295,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9393284156172253e-06,
+      "loss": 0.7105,
+      "step": 3472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9379990011604767e-06,
+      "loss": 0.79,
+      "step": 3473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9366697540609125e-06,
+      "loss": 0.7695,
+      "step": 3474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.935340674714369e-06,
+      "loss": 0.7527,
+      "step": 3475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.934011763516628e-06,
+      "loss": 0.7596,
+      "step": 3476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.932683020863424e-06,
+      "loss": 0.7865,
+      "step": 3477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9313544471504414e-06,
+      "loss": 0.8352,
+      "step": 3478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.930026042773312e-06,
+      "loss": 0.8093,
+      "step": 3479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.92869780812762e-06,
+      "loss": 0.8426,
+      "step": 3480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.927369743608897e-06,
+      "loss": 0.72,
+      "step": 3481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.926041849612627e-06,
+      "loss": 0.7698,
+      "step": 3482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9247141265342374e-06,
+      "loss": 0.8329,
+      "step": 3483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.923386574769111e-06,
+      "loss": 0.6596,
+      "step": 3484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.922059194712577e-06,
+      "loss": 0.808,
+      "step": 3485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.920731986759911e-06,
+      "loss": 0.8233,
+      "step": 3486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9194049513063405e-06,
+      "loss": 0.8236,
+      "step": 3487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.918078088747042e-06,
+      "loss": 0.7456,
+      "step": 3488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.916751399477139e-06,
+      "loss": 0.7327,
+      "step": 3489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.915424883891703e-06,
+      "loss": 0.7504,
+      "step": 3490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9140985423857545e-06,
+      "loss": 0.7381,
+      "step": 3491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.912772375354263e-06,
+      "loss": 0.8272,
+      "step": 3492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.911446383192145e-06,
+      "loss": 0.8365,
+      "step": 3493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.910120566294264e-06,
+      "loss": 0.8058,
+      "step": 3494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.908794925055433e-06,
+      "loss": 0.7702,
+      "step": 3495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9074694598704142e-06,
+      "loss": 0.7177,
+      "step": 3496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9061441711339122e-06,
+      "loss": 0.8301,
+      "step": 3497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.904819059240583e-06,
+      "loss": 0.7496,
+      "step": 3498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9034941245850308e-06,
+      "loss": 0.7277,
+      "step": 3499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9021693675618036e-06,
+      "loss": 0.7491,
+      "step": 3500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9008447885653986e-06,
+      "loss": 0.7537,
+      "step": 3501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8995203879902592e-06,
+      "loss": 0.8038,
+      "step": 3502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8981961662307775e-06,
+      "loss": 0.7513,
+      "step": 3503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8968721236812893e-06,
+      "loss": 0.6929,
+      "step": 3504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8955482607360797e-06,
+      "loss": 0.8484,
+      "step": 3505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8942245777893791e-06,
+      "loss": 0.7555,
+      "step": 3506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8929010752353635e-06,
+      "loss": 0.7874,
+      "step": 3507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8915777534681575e-06,
+      "loss": 0.8209,
+      "step": 3508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8902546128818295e-06,
+      "loss": 0.6186,
+      "step": 3509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.888931653870397e-06,
+      "loss": 0.741,
+      "step": 3510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8876088768278182e-06,
+      "loss": 0.7852,
+      "step": 3511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8862862821480023e-06,
+      "loss": 0.8302,
+      "step": 3512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8849638702248017e-06,
+      "loss": 0.8269,
+      "step": 3513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8836416414520162e-06,
+      "loss": 0.7569,
+      "step": 3514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.882319596223387e-06,
+      "loss": 0.7405,
+      "step": 3515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8809977349326056e-06,
+      "loss": 0.6611,
+      "step": 3516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8796760579733063e-06,
+      "loss": 0.6958,
+      "step": 3517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8783545657390673e-06,
+      "loss": 0.7527,
+      "step": 3518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8770332586234147e-06,
+      "loss": 0.741,
+      "step": 3519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8757121370198172e-06,
+      "loss": 0.5897,
+      "step": 3520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8743912013216897e-06,
+      "loss": 0.7632,
+      "step": 3521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.87307045192239e-06,
+      "loss": 0.7163,
+      "step": 3522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8717498892152224e-06,
+      "loss": 0.764,
+      "step": 3523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.870429513593435e-06,
+      "loss": 0.5809,
+      "step": 3524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.869109325450219e-06,
+      "loss": 0.7886,
+      "step": 3525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8677893251787103e-06,
+      "loss": 0.7915,
+      "step": 3526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8664695131719901e-06,
+      "loss": 0.5887,
+      "step": 3527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8651498898230835e-06,
+      "loss": 0.7693,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8638304555249568e-06,
+      "loss": 0.7502,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8625112106705218e-06,
+      "loss": 0.7299,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8611921556526358e-06,
+      "loss": 0.7593,
+      "step": 3531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8598732908640959e-06,
+      "loss": 0.7574,
+      "step": 3532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.858554616697645e-06,
+      "loss": 0.7499,
+      "step": 3533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8572361335459674e-06,
+      "loss": 0.7425,
+      "step": 3534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8559178418016937e-06,
+      "loss": 0.7858,
+      "step": 3535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545997418573939e-06,
+      "loss": 0.7378,
+      "step": 3536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532818341055819e-06,
+      "loss": 0.7285,
+      "step": 3537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519641189387178e-06,
+      "loss": 0.6921,
+      "step": 3538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506465967491976e-06,
+      "loss": 0.56,
+      "step": 3539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8493292679293662e-06,
+      "loss": 0.7935,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8480121328715068e-06,
+      "loss": 0.8348,
+      "step": 3541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8466951919678491e-06,
+      "loss": 0.7562,
+      "step": 3542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8453784456105591e-06,
+      "loss": 0.7861,
+      "step": 3543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8440618941917504e-06,
+      "loss": 0.8121,
+      "step": 3544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8427455381034758e-06,
+      "loss": 0.8002,
+      "step": 3545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8414293777377293e-06,
+      "loss": 0.7448,
+      "step": 3546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.840113413486449e-06,
+      "loss": 0.8285,
+      "step": 3547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8387976457415133e-06,
+      "loss": 0.7288,
+      "step": 3548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.837482074894742e-06,
+      "loss": 0.729,
+      "step": 3549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8361667013378953e-06,
+      "loss": 0.7818,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.834851525462677e-06,
+      "loss": 0.6817,
+      "step": 3551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8335365476607304e-06,
+      "loss": 0.7272,
+      "step": 3552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.832221768323641e-06,
+      "loss": 0.7903,
+      "step": 3553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8309071878429319e-06,
+      "loss": 0.7766,
+      "step": 3554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.829592806610072e-06,
+      "loss": 0.7576,
+      "step": 3555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.828278625016468e-06,
+      "loss": 0.8169,
+      "step": 3556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8269646434534663e-06,
+      "loss": 0.7413,
+      "step": 3557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8256508623123548e-06,
+      "loss": 0.7162,
+      "step": 3558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8243372819843635e-06,
+      "loss": 0.8136,
+      "step": 3559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8230239028606605e-06,
+      "loss": 0.7994,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8217107253323532e-06,
+      "loss": 0.7957,
+      "step": 3561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8203977497904905e-06,
+      "loss": 0.742,
+      "step": 3562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8190849766260622e-06,
+      "loss": 0.6398,
+      "step": 3563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8177724062299951e-06,
+      "loss": 0.7882,
+      "step": 3564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.816460038993157e-06,
+      "loss": 0.7858,
+      "step": 3565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8151478753063556e-06,
+      "loss": 0.7635,
+      "step": 3566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8138359155603386e-06,
+      "loss": 0.7582,
+      "step": 3567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8125241601457902e-06,
+      "loss": 0.8169,
+      "step": 3568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8112126094533357e-06,
+      "loss": 0.7903,
+      "step": 3569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8099012638735413e-06,
+      "loss": 0.749,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.808590123796907e-06,
+      "loss": 0.8656,
+      "step": 3571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8072791896138764e-06,
+      "loss": 0.7261,
+      "step": 3572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8059684617148295e-06,
+      "loss": 0.7469,
+      "step": 3573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8046579404900869e-06,
+      "loss": 0.7445,
+      "step": 3574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8033476263299044e-06,
+      "loss": 0.7657,
+      "step": 3575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8020375196244783e-06,
+      "loss": 0.7626,
+      "step": 3576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8007276207639435e-06,
+      "loss": 0.8416,
+      "step": 3577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7994179301383704e-06,
+      "loss": 0.7507,
+      "step": 3578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7981084481377713e-06,
+      "loss": 0.673,
+      "step": 3579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7967991751520932e-06,
+      "loss": 0.7302,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7954901115712232e-06,
+      "loss": 0.7768,
+      "step": 3581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.794181257784982e-06,
+      "loss": 0.7776,
+      "step": 3582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7928726141831338e-06,
+      "loss": 0.8075,
+      "step": 3583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7915641811553758e-06,
+      "loss": 0.825,
+      "step": 3584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7902559590913432e-06,
+      "loss": 0.7575,
+      "step": 3585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7889479483806088e-06,
+      "loss": 0.7902,
+      "step": 3586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7876401494126834e-06,
+      "loss": 0.8405,
+      "step": 3587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7863325625770144e-06,
+      "loss": 0.7636,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7850251882629844e-06,
+      "loss": 0.6208,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.783718026859913e-06,
+      "loss": 0.7372,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7824110787570599e-06,
+      "loss": 0.7891,
+      "step": 3591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7811043443436172e-06,
+      "loss": 0.7483,
+      "step": 3592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7797978240087135e-06,
+      "loss": 0.7685,
+      "step": 3593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.778491518141417e-06,
+      "loss": 0.7462,
+      "step": 3594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7771854271307299e-06,
+      "loss": 0.8021,
+      "step": 3595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.775879551365589e-06,
+      "loss": 0.744,
+      "step": 3596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7745738912348693e-06,
+      "loss": 0.79,
+      "step": 3597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7732684471273809e-06,
+      "loss": 0.7591,
+      "step": 3598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7719632194318704e-06,
+      "loss": 0.6872,
+      "step": 3599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7706582085370173e-06,
+      "loss": 0.7224,
+      "step": 3600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7693534148314382e-06,
+      "loss": 0.7718,
+      "step": 3601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.768048838703687e-06,
+      "loss": 0.7279,
+      "step": 3602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7667444805422495e-06,
+      "loss": 0.7995,
+      "step": 3603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.765440340735548e-06,
+      "loss": 0.7451,
+      "step": 3604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7641364196719396e-06,
+      "loss": 0.787,
+      "step": 3605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7628327177397176e-06,
+      "loss": 0.7855,
+      "step": 3606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7615292353271079e-06,
+      "loss": 0.7568,
+      "step": 3607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.760225972822272e-06,
+      "loss": 0.8815,
+      "step": 3608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7589229306133065e-06,
+      "loss": 0.5583,
+      "step": 3609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7576201090882406e-06,
+      "loss": 0.6975,
+      "step": 3610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.75631750863504e-06,
+      "loss": 0.795,
+      "step": 3611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7550151296416034e-06,
+      "loss": 0.7648,
+      "step": 3612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7537129724957644e-06,
+      "loss": 0.7244,
+      "step": 3613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.752411037585288e-06,
+      "loss": 0.7271,
+      "step": 3614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7511093252978762e-06,
+      "loss": 0.734,
+      "step": 3615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.749807836021164e-06,
+      "loss": 0.7876,
+      "step": 3616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7485065701427178e-06,
+      "loss": 0.7159,
+      "step": 3617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7472055280500392e-06,
+      "loss": 0.8089,
+      "step": 3618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.745904710130564e-06,
+      "loss": 0.712,
+      "step": 3619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7446041167716605e-06,
+      "loss": 0.5569,
+      "step": 3620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.743303748360628e-06,
+      "loss": 0.6952,
+      "step": 3621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7420036052847022e-06,
+      "loss": 0.7186,
+      "step": 3622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7407036879310502e-06,
+      "loss": 0.767,
+      "step": 3623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7394039966867713e-06,
+      "loss": 0.732,
+      "step": 3624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7381045319388972e-06,
+      "loss": 0.8449,
+      "step": 3625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.736805294074395e-06,
+      "loss": 0.775,
+      "step": 3626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7355062834801618e-06,
+      "loss": 0.7823,
+      "step": 3627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.734207500543026e-06,
+      "loss": 0.7578,
+      "step": 3628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.73290894564975e-06,
+      "loss": 0.7895,
+      "step": 3629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7316106191870288e-06,
+      "loss": 0.7414,
+      "step": 3630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7303125215414887e-06,
+      "loss": 0.8407,
+      "step": 3631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7290146530996869e-06,
+      "loss": 0.8712,
+      "step": 3632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7277170142481126e-06,
+      "loss": 0.7515,
+      "step": 3633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7264196053731885e-06,
+      "loss": 0.8019,
+      "step": 3634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7251224268612672e-06,
+      "loss": 0.8565,
+      "step": 3635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723825479098632e-06,
+      "loss": 0.7551,
+      "step": 3636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7225287624714987e-06,
+      "loss": 0.7678,
+      "step": 3637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7212322773660152e-06,
+      "loss": 0.8409,
+      "step": 3638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7199360241682583e-06,
+      "loss": 0.7582,
+      "step": 3639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.718640003264237e-06,
+      "loss": 0.7665,
+      "step": 3640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7173442150398908e-06,
+      "loss": 0.7674,
+      "step": 3641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7160486598810894e-06,
+      "loss": 0.8072,
+      "step": 3642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7147533381736341e-06,
+      "loss": 0.7496,
+      "step": 3643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.713458250303256e-06,
+      "loss": 0.779,
+      "step": 3644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7121633966556188e-06,
+      "loss": 0.7303,
+      "step": 3645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7108687776163113e-06,
+      "loss": 0.5509,
+      "step": 3646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7095743935708569e-06,
+      "loss": 0.5939,
+      "step": 3647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7082802449047093e-06,
+      "loss": 0.7775,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.706986332003247e-06,
+      "loss": 0.7161,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.705692655251785e-06,
+      "loss": 0.7616,
+      "step": 3650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7043992150355637e-06,
+      "loss": 0.5241,
+      "step": 3651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7031060117397545e-06,
+      "loss": 0.7836,
+      "step": 3652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7018130457494564e-06,
+      "loss": 0.726,
+      "step": 3653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7005203174497006e-06,
+      "loss": 0.7781,
+      "step": 3654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6992278272254467e-06,
+      "loss": 0.7314,
+      "step": 3655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6979355754615814e-06,
+      "loss": 0.7238,
+      "step": 3656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6966435625429221e-06,
+      "loss": 0.717,
+      "step": 3657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6953517888542153e-06,
+      "loss": 0.7832,
+      "step": 3658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6940602547801367e-06,
+      "loss": 0.7806,
+      "step": 3659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6927689607052877e-06,
+      "loss": 0.7607,
+      "step": 3660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6914779070142012e-06,
+      "loss": 0.7579,
+      "step": 3661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6901870940913384e-06,
+      "loss": 0.685,
+      "step": 3662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6888965223210866e-06,
+      "loss": 0.775,
+      "step": 3663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6876061920877632e-06,
+      "loss": 0.8105,
+      "step": 3664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6863161037756125e-06,
+      "loss": 0.8277,
+      "step": 3665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6850262577688093e-06,
+      "loss": 0.8048,
+      "step": 3666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6837366544514527e-06,
+      "loss": 0.7819,
+      "step": 3667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6824472942075714e-06,
+      "loss": 0.6954,
+      "step": 3668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6811581774211212e-06,
+      "loss": 0.7475,
+      "step": 3669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.679869304475987e-06,
+      "loss": 0.765,
+      "step": 3670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6785806757559786e-06,
+      "loss": 0.685,
+      "step": 3671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6772922916448336e-06,
+      "loss": 0.611,
+      "step": 3672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6760041525262204e-06,
+      "loss": 0.7227,
+      "step": 3673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6747162587837275e-06,
+      "loss": 0.7389,
+      "step": 3674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6734286108008768e-06,
+      "loss": 0.6418,
+      "step": 3675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6721412089611133e-06,
+      "loss": 0.812,
+      "step": 3676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6708540536478123e-06,
+      "loss": 0.7552,
+      "step": 3677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6695671452442697e-06,
+      "loss": 0.7632,
+      "step": 3678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6682804841337134e-06,
+      "loss": 0.7946,
+      "step": 3679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.666994070699296e-06,
+      "loss": 0.7168,
+      "step": 3680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.665707905324095e-06,
+      "loss": 0.7703,
+      "step": 3681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6644219883911156e-06,
+      "loss": 0.7394,
+      "step": 3682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.663136320283289e-06,
+      "loss": 0.8062,
+      "step": 3683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6618509013834715e-06,
+      "loss": 0.7609,
+      "step": 3684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6605657320744445e-06,
+      "loss": 0.7449,
+      "step": 3685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.659280812738917e-06,
+      "loss": 0.8288,
+      "step": 3686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6579961437595236e-06,
+      "loss": 0.7276,
+      "step": 3687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6567117255188215e-06,
+      "loss": 0.5828,
+      "step": 3688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6554275583992956e-06,
+      "loss": 0.6772,
+      "step": 3689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6541436427833563e-06,
+      "loss": 0.7418,
+      "step": 3690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6528599790533384e-06,
+      "loss": 0.7907,
+      "step": 3691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6515765675915005e-06,
+      "loss": 0.7768,
+      "step": 3692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6502934087800277e-06,
+      "loss": 0.6391,
+      "step": 3693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.64901050300103e-06,
+      "loss": 0.8242,
+      "step": 3694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6477278506365407e-06,
+      "loss": 0.8195,
+      "step": 3695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6464454520685186e-06,
+      "loss": 0.5807,
+      "step": 3696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.645163307678846e-06,
+      "loss": 0.7823,
+      "step": 3697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6438814178493317e-06,
+      "loss": 0.5654,
+      "step": 3698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6425997829617064e-06,
+      "loss": 0.6431,
+      "step": 3699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6413184033976246e-06,
+      "loss": 0.7378,
+      "step": 3700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6400372795386683e-06,
+      "loss": 0.8164,
+      "step": 3701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6387564117663379e-06,
+      "loss": 0.7546,
+      "step": 3702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6374758004620627e-06,
+      "loss": 0.7756,
+      "step": 3703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6361954460071923e-06,
+      "loss": 0.8133,
+      "step": 3704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6349153487830028e-06,
+      "loss": 0.7977,
+      "step": 3705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.633635509170689e-06,
+      "loss": 0.7529,
+      "step": 3706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.632355927551374e-06,
+      "loss": 0.6907,
+      "step": 3707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6310766043061016e-06,
+      "loss": 0.7753,
+      "step": 3708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6297975398158377e-06,
+      "loss": 0.8046,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6285187344614739e-06,
+      "loss": 0.8139,
+      "step": 3710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.627240188623823e-06,
+      "loss": 0.5539,
+      "step": 3711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6259619026836204e-06,
+      "loss": 0.7666,
+      "step": 3712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6246838770215235e-06,
+      "loss": 0.7867,
+      "step": 3713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6234061120181144e-06,
+      "loss": 0.7371,
+      "step": 3714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6221286080538956e-06,
+      "loss": 0.6547,
+      "step": 3715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6208513655092933e-06,
+      "loss": 0.7742,
+      "step": 3716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6195743847646534e-06,
+      "loss": 0.7253,
+      "step": 3717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6182976662002469e-06,
+      "loss": 0.6948,
+      "step": 3718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6170212101962656e-06,
+      "loss": 0.7403,
+      "step": 3719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6157450171328222e-06,
+      "loss": 0.6927,
+      "step": 3720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6144690873899506e-06,
+      "loss": 0.7695,
+      "step": 3721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6131934213476096e-06,
+      "loss": 0.8175,
+      "step": 3722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6119180193856768e-06,
+      "loss": 0.7263,
+      "step": 3723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6106428818839506e-06,
+      "loss": 0.8106,
+      "step": 3724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609368009222152e-06,
+      "loss": 0.8361,
+      "step": 3725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6080934017799244e-06,
+      "loss": 0.7566,
+      "step": 3726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6068190599368286e-06,
+      "loss": 0.7523,
+      "step": 3727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6055449840723485e-06,
+      "loss": 0.7402,
+      "step": 3728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6042711745658907e-06,
+      "loss": 0.7588,
+      "step": 3729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6029976317967795e-06,
+      "loss": 0.7877,
+      "step": 3730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6017243561442593e-06,
+      "loss": 0.8,
+      "step": 3731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6004513479874975e-06,
+      "loss": 0.755,
+      "step": 3732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5991786077055821e-06,
+      "loss": 0.8257,
+      "step": 3733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5979061356775172e-06,
+      "loss": 0.7363,
+      "step": 3734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5966339322822315e-06,
+      "loss": 0.6944,
+      "step": 3735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5953619978985712e-06,
+      "loss": 0.7239,
+      "step": 3736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5940903329053048e-06,
+      "loss": 0.7789,
+      "step": 3737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.592818937681117e-06,
+      "loss": 0.6855,
+      "step": 3738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5915478126046152e-06,
+      "loss": 0.7503,
+      "step": 3739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.590276958054325e-06,
+      "loss": 0.6996,
+      "step": 3740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5890063744086908e-06,
+      "loss": 0.7586,
+      "step": 3741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5877360620460786e-06,
+      "loss": 0.7681,
+      "step": 3742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5864660213447717e-06,
+      "loss": 0.765,
+      "step": 3743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5851962526829733e-06,
+      "loss": 0.783,
+      "step": 3744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5839267564388044e-06,
+      "loss": 0.771,
+      "step": 3745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.582657532990307e-06,
+      "loss": 0.7453,
+      "step": 3746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5813885827154408e-06,
+      "loss": 0.7865,
+      "step": 3747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.580119905992083e-06,
+      "loss": 0.8174,
+      "step": 3748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5788515031980301e-06,
+      "loss": 0.7177,
+      "step": 3749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5775833747109986e-06,
+      "loss": 0.7282,
+      "step": 3750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.576315520908622e-06,
+      "loss": 0.7371,
+      "step": 3751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5750479421684511e-06,
+      "loss": 0.7463,
+      "step": 3752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5737806388679552e-06,
+      "loss": 0.806,
+      "step": 3753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5725136113845239e-06,
+      "loss": 0.825,
+      "step": 3754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5712468600954621e-06,
+      "loss": 0.8276,
+      "step": 3755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5699803853779922e-06,
+      "loss": 0.849,
+      "step": 3756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5687141876092566e-06,
+      "loss": 0.7375,
+      "step": 3757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5674482671663135e-06,
+      "loss": 0.5596,
+      "step": 3758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5661826244261385e-06,
+      "loss": 0.7207,
+      "step": 3759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.564917259765625e-06,
+      "loss": 0.772,
+      "step": 3760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5636521735615835e-06,
+      "loss": 0.7268,
+      "step": 3761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5623873661907425e-06,
+      "loss": 0.734,
+      "step": 3762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.561122838029745e-06,
+      "loss": 0.8295,
+      "step": 3763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5598585894551524e-06,
+      "loss": 0.5971,
+      "step": 3764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5585946208434449e-06,
+      "loss": 0.769,
+      "step": 3765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.557330932571015e-06,
+      "loss": 0.8177,
+      "step": 3766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5560675250141752e-06,
+      "loss": 0.7377,
+      "step": 3767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5548043985491524e-06,
+      "loss": 0.771,
+      "step": 3768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5535415535520915e-06,
+      "loss": 0.5002,
+      "step": 3769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.552278990399052e-06,
+      "loss": 0.7219,
+      "step": 3770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.55101670946601e-06,
+      "loss": 0.8084,
+      "step": 3771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5497547111288588e-06,
+      "loss": 0.7958,
+      "step": 3772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5484929957634045e-06,
+      "loss": 0.8129,
+      "step": 3773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5472315637453725e-06,
+      "loss": 0.7189,
+      "step": 3774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5459704154504013e-06,
+      "loss": 0.716,
+      "step": 3775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5447095512540472e-06,
+      "loss": 0.7653,
+      "step": 3776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5434489715317783e-06,
+      "loss": 0.7615,
+      "step": 3777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5421886766589817e-06,
+      "loss": 0.7486,
+      "step": 3778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5409286670109585e-06,
+      "loss": 0.8916,
+      "step": 3779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.539668942962923e-06,
+      "loss": 0.7838,
+      "step": 3780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5384095048900067e-06,
+      "loss": 0.6956,
+      "step": 3781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5371503531672554e-06,
+      "loss": 0.6942,
+      "step": 3782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5358914881696302e-06,
+      "loss": 0.7747,
+      "step": 3783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5346329102720041e-06,
+      "loss": 0.59,
+      "step": 3784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.533374619849168e-06,
+      "loss": 0.7027,
+      "step": 3785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5321166172758263e-06,
+      "loss": 0.7786,
+      "step": 3786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5308589029265958e-06,
+      "loss": 0.7808,
+      "step": 3787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5296014771760085e-06,
+      "loss": 0.7242,
+      "step": 3788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.528344340398512e-06,
+      "loss": 0.7578,
+      "step": 3789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.527087492968467e-06,
+      "loss": 0.8372,
+      "step": 3790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5258309352601464e-06,
+      "loss": 0.762,
+      "step": 3791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5245746676477376e-06,
+      "loss": 0.7528,
+      "step": 3792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5233186905053438e-06,
+      "loss": 0.7731,
+      "step": 3793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.52206300420698e-06,
+      "loss": 0.769,
+      "step": 3794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5208076091265729e-06,
+      "loss": 0.7466,
+      "step": 3795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5195525056379645e-06,
+      "loss": 0.8346,
+      "step": 3796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5182976941149113e-06,
+      "loss": 0.6997,
+      "step": 3797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.51704317493108e-06,
+      "loss": 0.8188,
+      "step": 3798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5157889484600508e-06,
+      "loss": 0.7387,
+      "step": 3799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5145350150753179e-06,
+      "loss": 0.6898,
+      "step": 3800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.513281375150289e-06,
+      "loss": 0.7413,
+      "step": 3801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5120280290582812e-06,
+      "loss": 0.5995,
+      "step": 3802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5107749771725262e-06,
+      "loss": 0.7218,
+      "step": 3803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5095222198661699e-06,
+      "loss": 0.7486,
+      "step": 3804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5082697575122657e-06,
+      "loss": 0.7116,
+      "step": 3805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5070175904837835e-06,
+      "loss": 0.5276,
+      "step": 3806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.505765719153603e-06,
+      "loss": 0.8102,
+      "step": 3807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5045141438945188e-06,
+      "loss": 0.7657,
+      "step": 3808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5032628650792314e-06,
+      "loss": 0.7806,
+      "step": 3809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5020118830803586e-06,
+      "loss": 0.8112,
+      "step": 3810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.500761198270429e-06,
+      "loss": 0.8066,
+      "step": 3811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4995108110218786e-06,
+      "loss": 0.7684,
+      "step": 3812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4982607217070608e-06,
+      "loss": 0.7892,
+      "step": 3813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4970109306982356e-06,
+      "loss": 0.8242,
+      "step": 3814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.495761438367577e-06,
+      "loss": 0.7154,
+      "step": 3815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4945122450871672e-06,
+      "loss": 0.7216,
+      "step": 3816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4932633512290024e-06,
+      "loss": 0.7202,
+      "step": 3817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4920147571649884e-06,
+      "loss": 0.7351,
+      "step": 3818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4907664632669405e-06,
+      "loss": 0.5794,
+      "step": 3819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.489518469906586e-06,
+      "loss": 0.5826,
+      "step": 3820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4882707774555633e-06,
+      "loss": 0.7856,
+      "step": 3821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.48702338628542e-06,
+      "loss": 0.7377,
+      "step": 3822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4857762967676135e-06,
+      "loss": 0.733,
+      "step": 3823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4845295092735122e-06,
+      "loss": 0.7042,
+      "step": 3824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4832830241743961e-06,
+      "loss": 0.833,
+      "step": 3825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.482036841841452e-06,
+      "loss": 0.7612,
+      "step": 3826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4807909626457785e-06,
+      "loss": 0.845,
+      "step": 3827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4795453869583831e-06,
+      "loss": 0.7815,
+      "step": 3828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4783001151501849e-06,
+      "loss": 0.7253,
+      "step": 3829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4770551475920098e-06,
+      "loss": 0.7083,
+      "step": 3830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4758104846545935e-06,
+      "loss": 0.7684,
+      "step": 3831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4745661267085837e-06,
+      "loss": 0.7347,
+      "step": 3832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4733220741245346e-06,
+      "loss": 0.6798,
+      "step": 3833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4720783272729095e-06,
+      "loss": 0.7891,
+      "step": 3834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4708348865240812e-06,
+      "loss": 0.7995,
+      "step": 3835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4695917522483336e-06,
+      "loss": 0.7449,
+      "step": 3836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.468348924815854e-06,
+      "loss": 0.7928,
+      "step": 3837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4671064045967443e-06,
+      "loss": 0.7614,
+      "step": 3838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4658641919610106e-06,
+      "loss": 0.7685,
+      "step": 3839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4646222872785703e-06,
+      "loss": 0.6965,
+      "step": 3840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4633806909192468e-06,
+      "loss": 0.7923,
+      "step": 3841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.462139403252773e-06,
+      "loss": 0.7698,
+      "step": 3842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4608984246487898e-06,
+      "loss": 0.7949,
+      "step": 3843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4596577554768443e-06,
+      "loss": 0.8106,
+      "step": 3844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4584173961063953e-06,
+      "loss": 0.7223,
+      "step": 3845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4571773469068047e-06,
+      "loss": 0.778,
+      "step": 3846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4559376082473467e-06,
+      "loss": 0.756,
+      "step": 3847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4546981804971994e-06,
+      "loss": 0.684,
+      "step": 3848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4534590640254488e-06,
+      "loss": 0.8127,
+      "step": 3849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4522202592010909e-06,
+      "loss": 0.7549,
+      "step": 3850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4509817663930244e-06,
+      "loss": 0.7717,
+      "step": 3851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4497435859700606e-06,
+      "loss": 0.6742,
+      "step": 3852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4485057183009124e-06,
+      "loss": 0.7362,
+      "step": 3853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4472681637542042e-06,
+      "loss": 0.7457,
+      "step": 3854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4460309226984623e-06,
+      "loss": 0.7521,
+      "step": 3855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.444793995502125e-06,
+      "loss": 0.7755,
+      "step": 3856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4435573825335332e-06,
+      "loss": 0.7991,
+      "step": 3857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4423210841609348e-06,
+      "loss": 0.7442,
+      "step": 3858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.441085100752485e-06,
+      "loss": 0.6858,
+      "step": 3859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4398494326762462e-06,
+      "loss": 0.5347,
+      "step": 3860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4386140803001852e-06,
+      "loss": 0.7759,
+      "step": 3861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4373790439921736e-06,
+      "loss": 0.7862,
+      "step": 3862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4361443241199913e-06,
+      "loss": 0.7188,
+      "step": 3863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4349099210513246e-06,
+      "loss": 0.7096,
+      "step": 3864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4336758351537624e-06,
+      "loss": 0.7419,
+      "step": 3865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4324420667948002e-06,
+      "loss": 0.7424,
+      "step": 3866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4312086163418408e-06,
+      "loss": 0.8073,
+      "step": 3867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4299754841621912e-06,
+      "loss": 0.6945,
+      "step": 3868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4287426706230628e-06,
+      "loss": 0.6419,
+      "step": 3869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4275101760915723e-06,
+      "loss": 0.7769,
+      "step": 3870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4262780009347418e-06,
+      "loss": 0.8144,
+      "step": 3871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4250461455195e-06,
+      "loss": 0.7264,
+      "step": 3872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4238146102126765e-06,
+      "loss": 0.7215,
+      "step": 3873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4225833953810104e-06,
+      "loss": 0.7581,
+      "step": 3874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4213525013911408e-06,
+      "loss": 0.6816,
+      "step": 3875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4201219286096129e-06,
+      "loss": 0.7963,
+      "step": 3876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4188916774028773e-06,
+      "loss": 0.733,
+      "step": 3877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4176617481372896e-06,
+      "loss": 0.8774,
+      "step": 3878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4164321411791063e-06,
+      "loss": 0.7983,
+      "step": 3879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4152028568944892e-06,
+      "loss": 0.5759,
+      "step": 3880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.413973895649505e-06,
+      "loss": 0.7428,
+      "step": 3881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4127452578101252e-06,
+      "loss": 0.6946,
+      "step": 3882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.411516943742222e-06,
+      "loss": 0.7558,
+      "step": 3883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4102889538115725e-06,
+      "loss": 0.7423,
+      "step": 3884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4090612883838573e-06,
+      "loss": 0.7532,
+      "step": 3885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4078339478246623e-06,
+      "loss": 0.7884,
+      "step": 3886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4066069324994735e-06,
+      "loss": 0.7122,
+      "step": 3887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4053802427736808e-06,
+      "loss": 0.7287,
+      "step": 3888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4041538790125792e-06,
+      "loss": 0.7403,
+      "step": 3889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4029278415813634e-06,
+      "loss": 0.7437,
+      "step": 3890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4017021308451351e-06,
+      "loss": 0.7387,
+      "step": 3891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4004767471688937e-06,
+      "loss": 0.753,
+      "step": 3892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3992516909175463e-06,
+      "loss": 0.726,
+      "step": 3893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3980269624558977e-06,
+      "loss": 0.7852,
+      "step": 3894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3968025621486592e-06,
+      "loss": 0.6801,
+      "step": 3895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3955784903604425e-06,
+      "loss": 0.7358,
+      "step": 3896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3943547474557594e-06,
+      "loss": 0.711,
+      "step": 3897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3931313337990283e-06,
+      "loss": 0.7088,
+      "step": 3898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3919082497545665e-06,
+      "loss": 0.7812,
+      "step": 3899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3906854956865942e-06,
+      "loss": 0.7768,
+      "step": 3900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3894630719592313e-06,
+      "loss": 0.7984,
+      "step": 3901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3882409789365024e-06,
+      "loss": 0.7489,
+      "step": 3902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3870192169823328e-06,
+      "loss": 0.7478,
+      "step": 3903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3857977864605476e-06,
+      "loss": 0.7535,
+      "step": 3904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3845766877348732e-06,
+      "loss": 0.7146,
+      "step": 3905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.38335592116894e-06,
+      "loss": 0.7851,
+      "step": 3906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3821354871262777e-06,
+      "loss": 0.7116,
+      "step": 3907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3809153859703165e-06,
+      "loss": 0.7035,
+      "step": 3908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3796956180643867e-06,
+      "loss": 0.5716,
+      "step": 3909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3784761837717219e-06,
+      "loss": 0.7157,
+      "step": 3910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.377257083455456e-06,
+      "loss": 0.7519,
+      "step": 3911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3760383174786213e-06,
+      "loss": 0.7066,
+      "step": 3912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.374819886204151e-06,
+      "loss": 0.7606,
+      "step": 3913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3736017899948812e-06,
+      "loss": 0.8015,
+      "step": 3914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3723840292135444e-06,
+      "loss": 0.7923,
+      "step": 3915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3711666042227772e-06,
+      "loss": 0.7441,
+      "step": 3916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3699495153851126e-06,
+      "loss": 0.8049,
+      "step": 3917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3687327630629865e-06,
+      "loss": 0.802,
+      "step": 3918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3675163476187319e-06,
+      "loss": 0.8398,
+      "step": 3919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3663002694145832e-06,
+      "loss": 0.7675,
+      "step": 3920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3650845288126763e-06,
+      "loss": 0.791,
+      "step": 3921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3638691261750402e-06,
+      "loss": 0.7575,
+      "step": 3922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3626540618636092e-06,
+      "loss": 0.7449,
+      "step": 3923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3614393362402147e-06,
+      "loss": 0.7283,
+      "step": 3924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3602249496665898e-06,
+      "loss": 0.7041,
+      "step": 3925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3590109025043605e-06,
+      "loss": 0.7947,
+      "step": 3926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.357797195115057e-06,
+      "loss": 0.5119,
+      "step": 3927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3565838278601085e-06,
+      "loss": 0.7767,
+      "step": 3928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3553708011008387e-06,
+      "loss": 0.7328,
+      "step": 3929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3541581151984752e-06,
+      "loss": 0.7813,
+      "step": 3930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3529457705141386e-06,
+      "loss": 0.741,
+      "step": 3931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3517337674088527e-06,
+      "loss": 0.8728,
+      "step": 3932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.350522106243536e-06,
+      "loss": 0.8397,
+      "step": 3933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3493107873790085e-06,
+      "loss": 0.7874,
+      "step": 3934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3480998111759858e-06,
+      "loss": 0.7349,
+      "step": 3935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3468891779950805e-06,
+      "loss": 0.7385,
+      "step": 3936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3456788881968064e-06,
+      "loss": 0.7563,
+      "step": 3937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3444689421415741e-06,
+      "loss": 0.7792,
+      "step": 3938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3432593401896899e-06,
+      "loss": 0.7423,
+      "step": 3939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3420500827013577e-06,
+      "loss": 0.7332,
+      "step": 3940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3408411700366814e-06,
+      "loss": 0.7183,
+      "step": 3941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3396326025556616e-06,
+      "loss": 0.8286,
+      "step": 3942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.338424380618194e-06,
+      "loss": 0.7194,
+      "step": 3943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.337216504584072e-06,
+      "loss": 0.7827,
+      "step": 3944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3360089748129877e-06,
+      "loss": 0.7374,
+      "step": 3945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3348017916645295e-06,
+      "loss": 0.7488,
+      "step": 3946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3335949554981814e-06,
+      "loss": 0.7659,
+      "step": 3947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.332388466673324e-06,
+      "loss": 0.7989,
+      "step": 3948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3311823255492363e-06,
+      "loss": 0.6864,
+      "step": 3949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3299765324850936e-06,
+      "loss": 0.7128,
+      "step": 3950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3287710878399657e-06,
+      "loss": 0.7188,
+      "step": 3951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3275659919728182e-06,
+      "loss": 0.737,
+      "step": 3952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3263612452425168e-06,
+      "loss": 0.4886,
+      "step": 3953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3251568480078183e-06,
+      "loss": 0.7754,
+      "step": 3954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3239528006273802e-06,
+      "loss": 0.7192,
+      "step": 3955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3227491034597513e-06,
+      "loss": 0.7977,
+      "step": 3956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.32154575686338e-06,
+      "loss": 0.7901,
+      "step": 3957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3203427611966068e-06,
+      "loss": 0.8245,
+      "step": 3958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3191401168176712e-06,
+      "loss": 0.7628,
+      "step": 3959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3179378240847057e-06,
+      "loss": 0.7063,
+      "step": 3960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3167358833557376e-06,
+      "loss": 0.8198,
+      "step": 3961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3155342949886912e-06,
+      "loss": 0.7051,
+      "step": 3962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3143330593413865e-06,
+      "loss": 0.755,
+      "step": 3963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.313132176771536e-06,
+      "loss": 0.7372,
+      "step": 3964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.311931647636747e-06,
+      "loss": 0.8153,
+      "step": 3965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3107314722945242e-06,
+      "loss": 0.7446,
+      "step": 3966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.309531651102266e-06,
+      "loss": 0.7753,
+      "step": 3967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3083321844172642e-06,
+      "loss": 0.7304,
+      "step": 3968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.307133072596704e-06,
+      "loss": 0.7197,
+      "step": 3969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3059343159976686e-06,
+      "loss": 0.6612,
+      "step": 3970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3047359149771338e-06,
+      "loss": 0.8119,
+      "step": 3971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3035378698919682e-06,
+      "loss": 0.8015,
+      "step": 3972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3023401810989338e-06,
+      "loss": 0.737,
+      "step": 3973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.301142848954691e-06,
+      "loss": 0.6965,
+      "step": 3974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2999458738157883e-06,
+      "loss": 0.7185,
+      "step": 3975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2987492560386722e-06,
+      "loss": 0.7646,
+      "step": 3976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2975529959796812e-06,
+      "loss": 0.6962,
+      "step": 3977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.296357093995047e-06,
+      "loss": 0.6533,
+      "step": 3978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2951615504408944e-06,
+      "loss": 0.6272,
+      "step": 3979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2939663656732418e-06,
+      "loss": 0.7547,
+      "step": 3980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2927715400480035e-06,
+      "loss": 0.7204,
+      "step": 3981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2915770739209804e-06,
+      "loss": 0.7864,
+      "step": 3982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2903829676478723e-06,
+      "loss": 0.8534,
+      "step": 3983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.28918922158427e-06,
+      "loss": 0.7683,
+      "step": 3984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.287995836085657e-06,
+      "loss": 0.5363,
+      "step": 3985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2868028115074088e-06,
+      "loss": 0.5846,
+      "step": 3986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2856101482047929e-06,
+      "loss": 0.773,
+      "step": 3987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.284417846532971e-06,
+      "loss": 0.783,
+      "step": 3988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2832259068469976e-06,
+      "loss": 0.7836,
+      "step": 3989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2820343295018167e-06,
+      "loss": 0.7553,
+      "step": 3990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2808431148522649e-06,
+      "loss": 0.8189,
+      "step": 3991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.279652263253074e-06,
+      "loss": 0.7105,
+      "step": 3992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.278461775058863e-06,
+      "loss": 0.7539,
+      "step": 3993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2772716506241473e-06,
+      "loss": 0.6907,
+      "step": 3994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2760818903033298e-06,
+      "loss": 0.7097,
+      "step": 3995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2748924944507085e-06,
+      "loss": 0.7738,
+      "step": 3996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2737034634204698e-06,
+      "loss": 0.7607,
+      "step": 3997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2725147975666948e-06,
+      "loss": 0.7558,
+      "step": 3998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2713264972433529e-06,
+      "loss": 0.5192,
+      "step": 3999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2701385628043044e-06,
+      "loss": 0.7466,
+      "step": 4000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.268950994603304e-06,
+      "loss": 0.7522,
+      "step": 4001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2677637929939952e-06,
+      "loss": 0.8036,
+      "step": 4002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.266576958329912e-06,
+      "loss": 0.709,
+      "step": 4003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2653904909644785e-06,
+      "loss": 0.7605,
+      "step": 4004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2642043912510115e-06,
+      "loss": 0.8156,
+      "step": 4005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2630186595427186e-06,
+      "loss": 0.7884,
+      "step": 4006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.261833296192695e-06,
+      "loss": 0.7009,
+      "step": 4007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2606483015539272e-06,
+      "loss": 0.707,
+      "step": 4008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2594636759792934e-06,
+      "loss": 0.7181,
+      "step": 4009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2582794198215615e-06,
+      "loss": 0.7099,
+      "step": 4010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2570955334333884e-06,
+      "loss": 0.7408,
+      "step": 4011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2559120171673201e-06,
+      "loss": 0.6551,
+      "step": 4012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2547288713757955e-06,
+      "loss": 0.754,
+      "step": 4013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2535460964111395e-06,
+      "loss": 0.7462,
+      "step": 4014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2523636926255705e-06,
+      "loss": 0.7488,
+      "step": 4015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2511816603711918e-06,
+      "loss": 0.691,
+      "step": 4016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2500000000000007e-06,
+      "loss": 0.7403,
+      "step": 4017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2488187118638794e-06,
+      "loss": 0.7384,
+      "step": 4018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2476377963146035e-06,
+      "loss": 0.729,
+      "step": 4019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2464572537038347e-06,
+      "loss": 0.7414,
+      "step": 4020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2452770843831236e-06,
+      "loss": 0.7749,
+      "step": 4021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.244097288703911e-06,
+      "loss": 0.6691,
+      "step": 4022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2429178670175263e-06,
+      "loss": 0.8032,
+      "step": 4023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.241738819675189e-06,
+      "loss": 0.7365,
+      "step": 4024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2405601470280016e-06,
+      "loss": 0.7348,
+      "step": 4025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2393818494269608e-06,
+      "loss": 0.7544,
+      "step": 4026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.238203927222949e-06,
+      "loss": 0.7929,
+      "step": 4027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2370263807667399e-06,
+      "loss": 0.8208,
+      "step": 4028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2358492104089887e-06,
+      "loss": 0.5479,
+      "step": 4029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2346724165002444e-06,
+      "loss": 0.7127,
+      "step": 4030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.233495999390943e-06,
+      "loss": 0.71,
+      "step": 4031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2323199594314065e-06,
+      "loss": 0.7313,
+      "step": 4032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2311442969718461e-06,
+      "loss": 0.7528,
+      "step": 4033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2299690123623595e-06,
+      "loss": 0.7663,
+      "step": 4034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2287941059529331e-06,
+      "loss": 0.7838,
+      "step": 4035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2276195780934392e-06,
+      "loss": 0.7162,
+      "step": 4036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.226445429133639e-06,
+      "loss": 0.7966,
+      "step": 4037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2252716594231801e-06,
+      "loss": 0.5945,
+      "step": 4038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2240982693115953e-06,
+      "loss": 0.5348,
+      "step": 4039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2229252591483076e-06,
+      "loss": 0.8118,
+      "step": 4040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2217526292826262e-06,
+      "loss": 0.801,
+      "step": 4041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2205803800637454e-06,
+      "loss": 0.7941,
+      "step": 4042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2194085118407463e-06,
+      "loss": 0.6716,
+      "step": 4043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2182370249625973e-06,
+      "loss": 0.5693,
+      "step": 4044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2170659197781553e-06,
+      "loss": 0.8333,
+      "step": 4045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2158951966361599e-06,
+      "loss": 0.7386,
+      "step": 4046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2147248558852374e-06,
+      "loss": 0.7754,
+      "step": 4047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2135548978739026e-06,
+      "loss": 0.7604,
+      "step": 4048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2123853229505556e-06,
+      "loss": 0.7746,
+      "step": 4049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2112161314634813e-06,
+      "loss": 0.7945,
+      "step": 4050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2100473237608498e-06,
+      "loss": 0.7112,
+      "step": 4051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.20887890019072e-06,
+      "loss": 0.754,
+      "step": 4052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2077108611010326e-06,
+      "loss": 0.6998,
+      "step": 4053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2065432068396176e-06,
+      "loss": 0.7704,
+      "step": 4054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2053759377541865e-06,
+      "loss": 0.7941,
+      "step": 4055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.20420905419234e-06,
+      "loss": 0.7789,
+      "step": 4056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2030425565015606e-06,
+      "loss": 0.5961,
+      "step": 4057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2018764450292183e-06,
+      "loss": 0.7552,
+      "step": 4058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2007107201225673e-06,
+      "loss": 0.6735,
+      "step": 4059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.199545382128745e-06,
+      "loss": 0.7675,
+      "step": 4060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1983804313947763e-06,
+      "loss": 0.6501,
+      "step": 4061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1972158682675706e-06,
+      "loss": 0.7893,
+      "step": 4062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1960516930939195e-06,
+      "loss": 0.8356,
+      "step": 4063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1948879062205002e-06,
+      "loss": 0.7882,
+      "step": 4064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1937245079938748e-06,
+      "loss": 0.7318,
+      "step": 4065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1925614987604909e-06,
+      "loss": 0.7614,
+      "step": 4066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1913988788666774e-06,
+      "loss": 0.7847,
+      "step": 4067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1902366486586484e-06,
+      "loss": 0.8341,
+      "step": 4068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1890748084825022e-06,
+      "loss": 0.7138,
+      "step": 4069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.187913358684222e-06,
+      "loss": 0.7643,
+      "step": 4070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1867522996096734e-06,
+      "loss": 0.6675,
+      "step": 4071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1855916316046042e-06,
+      "loss": 0.6876,
+      "step": 4072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1844313550146492e-06,
+      "loss": 0.6813,
+      "step": 4073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1832714701853253e-06,
+      "loss": 0.7313,
+      "step": 4074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1821119774620315e-06,
+      "loss": 0.7776,
+      "step": 4075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.18095287719005e-06,
+      "loss": 0.7827,
+      "step": 4076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.179794169714549e-06,
+      "loss": 0.7094,
+      "step": 4077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1786358553805757e-06,
+      "loss": 0.6769,
+      "step": 4078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.177477934533063e-06,
+      "loss": 0.8026,
+      "step": 4079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1763204075168274e-06,
+      "loss": 0.8053,
+      "step": 4080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1751632746765654e-06,
+      "loss": 0.7196,
+      "step": 4081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1740065363568566e-06,
+      "loss": 0.6522,
+      "step": 4082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.172850192902164e-06,
+      "loss": 0.6882,
+      "step": 4083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1716942446568357e-06,
+      "loss": 0.7326,
+      "step": 4084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1705386919650953e-06,
+      "loss": 0.7793,
+      "step": 4085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.169383535171054e-06,
+      "loss": 0.5231,
+      "step": 4086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.168228774618704e-06,
+      "loss": 0.7736,
+      "step": 4087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1670744106519205e-06,
+      "loss": 0.7719,
+      "step": 4088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1659204436144577e-06,
+      "loss": 0.7329,
+      "step": 4089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.164766873849953e-06,
+      "loss": 0.6938,
+      "step": 4090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1636137017019269e-06,
+      "loss": 0.8553,
+      "step": 4091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1624609275137788e-06,
+      "loss": 0.7689,
+      "step": 4092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1613085516287928e-06,
+      "loss": 0.6915,
+      "step": 4093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1601565743901308e-06,
+      "loss": 0.8412,
+      "step": 4094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1590049961408397e-06,
+      "loss": 0.7408,
+      "step": 4095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.157853817223844e-06,
+      "loss": 0.6817,
+      "step": 4096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.156703037981953e-06,
+      "loss": 0.5729,
+      "step": 4097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.155552658757854e-06,
+      "loss": 0.7653,
+      "step": 4098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1544026798941153e-06,
+      "loss": 0.8046,
+      "step": 4099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1532531017331871e-06,
+      "loss": 0.7268,
+      "step": 4100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.152103924617402e-06,
+      "loss": 0.8042,
+      "step": 4101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.15095514888897e-06,
+      "loss": 0.7201,
+      "step": 4102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1498067748899815e-06,
+      "loss": 0.7416,
+      "step": 4103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1486588029624098e-06,
+      "loss": 0.7315,
+      "step": 4104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1475112334481084e-06,
+      "loss": 0.6748,
+      "step": 4105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.146364066688808e-06,
+      "loss": 0.7732,
+      "step": 4106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1452173030261213e-06,
+      "loss": 0.6842,
+      "step": 4107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.144070942801541e-06,
+      "loss": 0.7465,
+      "step": 4108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1429249863564407e-06,
+      "loss": 0.7542,
+      "step": 4109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1417794340320714e-06,
+      "loss": 0.7697,
+      "step": 4110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1406342861695643e-06,
+      "loss": 0.7437,
+      "step": 4111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1394895431099315e-06,
+      "loss": 0.6848,
+      "step": 4112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1383452051940647e-06,
+      "loss": 0.7555,
+      "step": 4113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1372012727627333e-06,
+      "loss": 0.8317,
+      "step": 4114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1360577461565855e-06,
+      "loss": 0.6923,
+      "step": 4115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1349146257161519e-06,
+      "loss": 0.6306,
+      "step": 4116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1337719117818386e-06,
+      "loss": 0.7523,
+      "step": 4117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1326296046939334e-06,
+      "loss": 0.764,
+      "step": 4118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1314877047926e-06,
+      "loss": 0.7992,
+      "step": 4119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.130346212417885e-06,
+      "loss": 0.7573,
+      "step": 4120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1292051279097085e-06,
+      "loss": 0.7537,
+      "step": 4121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1280644516078742e-06,
+      "loss": 0.7896,
+      "step": 4122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1269241838520608e-06,
+      "loss": 0.7835,
+      "step": 4123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1257843249818257e-06,
+      "loss": 0.5505,
+      "step": 4124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1246448753366056e-06,
+      "loss": 0.7241,
+      "step": 4125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.123505835255716e-06,
+      "loss": 0.761,
+      "step": 4126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1223672050783501e-06,
+      "loss": 0.7145,
+      "step": 4127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1212289851435756e-06,
+      "loss": 0.8372,
+      "step": 4128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1200911757903425e-06,
+      "loss": 0.7119,
+      "step": 4129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.118953777357477e-06,
+      "loss": 0.8213,
+      "step": 4130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1178167901836827e-06,
+      "loss": 0.7598,
+      "step": 4131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1166802146075395e-06,
+      "loss": 0.7597,
+      "step": 4132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1155440509675072e-06,
+      "loss": 0.751,
+      "step": 4133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1144082996019224e-06,
+      "loss": 0.8171,
+      "step": 4134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1132729608489967e-06,
+      "loss": 0.7209,
+      "step": 4135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.112138035046822e-06,
+      "loss": 0.7499,
+      "step": 4136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.111003522533365e-06,
+      "loss": 0.7872,
+      "step": 4137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1098694236464688e-06,
+      "loss": 0.8072,
+      "step": 4138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1087357387238556e-06,
+      "loss": 0.7245,
+      "step": 4139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1076024681031242e-06,
+      "loss": 0.7307,
+      "step": 4140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1064696121217476e-06,
+      "loss": 0.7077,
+      "step": 4141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1053371711170768e-06,
+      "loss": 0.7703,
+      "step": 4142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1042051454263395e-06,
+      "loss": 0.8387,
+      "step": 4143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1030735353866404e-06,
+      "loss": 0.7122,
+      "step": 4144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1019423413349587e-06,
+      "loss": 0.76,
+      "step": 4145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1008115636081494e-06,
+      "loss": 0.8043,
+      "step": 4146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0996812025429454e-06,
+      "loss": 0.7029,
+      "step": 4147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.098551258475956e-06,
+      "loss": 0.8109,
+      "step": 4148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0974217317436638e-06,
+      "loss": 0.7288,
+      "step": 4149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.096292622682428e-06,
+      "loss": 0.6903,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.095163931628484e-06,
+      "loss": 0.768,
+      "step": 4151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.094035658917944e-06,
+      "loss": 0.7045,
+      "step": 4152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0929078048867928e-06,
+      "loss": 0.5357,
+      "step": 4153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0917803698708912e-06,
+      "loss": 0.7333,
+      "step": 4154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0906533542059774e-06,
+      "loss": 0.7104,
+      "step": 4155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.089526758227662e-06,
+      "loss": 0.7242,
+      "step": 4156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.088400582271433e-06,
+      "loss": 0.704,
+      "step": 4157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0872748266726508e-06,
+      "loss": 0.738,
+      "step": 4158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0861494917665533e-06,
+      "loss": 0.7296,
+      "step": 4159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0850245778882503e-06,
+      "loss": 0.7747,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0839000853727298e-06,
+      "loss": 0.7268,
+      "step": 4161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0827760145548507e-06,
+      "loss": 0.6703,
+      "step": 4162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0816523657693476e-06,
+      "loss": 0.6953,
+      "step": 4163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0805291393508298e-06,
+      "loss": 0.7593,
+      "step": 4164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0794063356337824e-06,
+      "loss": 0.7945,
+      "step": 4165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.078283954952562e-06,
+      "loss": 0.6998,
+      "step": 4166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0771619976413985e-06,
+      "loss": 0.851,
+      "step": 4167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.076040464034399e-06,
+      "loss": 0.7478,
+      "step": 4168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0749193544655435e-06,
+      "loss": 0.7929,
+      "step": 4169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.073798669268684e-06,
+      "loss": 0.7941,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0726784087775465e-06,
+      "loss": 0.7933,
+      "step": 4171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0715585733257322e-06,
+      "loss": 0.7162,
+      "step": 4172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.070439163246715e-06,
+      "loss": 0.7094,
+      "step": 4173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0693201788738418e-06,
+      "loss": 0.7721,
+      "step": 4174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0682016205403312e-06,
+      "loss": 0.7894,
+      "step": 4175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0670834885792786e-06,
+      "loss": 0.7341,
+      "step": 4176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0659657833236487e-06,
+      "loss": 0.7158,
+      "step": 4177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0648485051062821e-06,
+      "loss": 0.5947,
+      "step": 4178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0637316542598893e-06,
+      "loss": 0.7991,
+      "step": 4179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0626152311170573e-06,
+      "loss": 0.8083,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0614992360102414e-06,
+      "loss": 0.76,
+      "step": 4181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0603836692717728e-06,
+      "loss": 0.6644,
+      "step": 4182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0592685312338544e-06,
+      "loss": 0.7329,
+      "step": 4183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0581538222285607e-06,
+      "loss": 0.716,
+      "step": 4184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0570395425878371e-06,
+      "loss": 0.7605,
+      "step": 4185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0559256926435044e-06,
+      "loss": 0.4937,
+      "step": 4186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0548122727272554e-06,
+      "loss": 0.7055,
+      "step": 4187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0536992831706492e-06,
+      "loss": 0.7686,
+      "step": 4188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.052586724305123e-06,
+      "loss": 0.7058,
+      "step": 4189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0514745964619832e-06,
+      "loss": 0.7627,
+      "step": 4190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0503628999724096e-06,
+      "loss": 0.7409,
+      "step": 4191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.049251635167451e-06,
+      "loss": 0.7679,
+      "step": 4192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0481408023780282e-06,
+      "loss": 0.7601,
+      "step": 4193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.047030401934935e-06,
+      "loss": 0.7323,
+      "step": 4194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0459204341688342e-06,
+      "loss": 0.7136,
+      "step": 4195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0448108994102627e-06,
+      "loss": 0.7335,
+      "step": 4196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0437017979896245e-06,
+      "loss": 0.7504,
+      "step": 4197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.042593130237199e-06,
+      "loss": 0.7532,
+      "step": 4198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.041484896483132e-06,
+      "loss": 0.5752,
+      "step": 4199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0403770970574441e-06,
+      "loss": 0.7377,
+      "step": 4200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0392697322900243e-06,
+      "loss": 0.7034,
+      "step": 4201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0381628025106317e-06,
+      "loss": 0.7456,
+      "step": 4202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0370563080488972e-06,
+      "loss": 0.7875,
+      "step": 4203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.035950249234323e-06,
+      "loss": 0.7919,
+      "step": 4204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0348446263962786e-06,
+      "loss": 0.7011,
+      "step": 4205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.033739439864005e-06,
+      "loss": 0.7722,
+      "step": 4206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0326346899666143e-06,
+      "loss": 0.7245,
+      "step": 4207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0315303770330887e-06,
+      "loss": 0.8017,
+      "step": 4208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0304265013922785e-06,
+      "loss": 0.7451,
+      "step": 4209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0293230633729037e-06,
+      "loss": 0.6554,
+      "step": 4210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0282200633035563e-06,
+      "loss": 0.8118,
+      "step": 4211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0271175015126967e-06,
+      "loss": 0.7582,
+      "step": 4212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0260153783286544e-06,
+      "loss": 0.7847,
+      "step": 4213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0249136940796273e-06,
+      "loss": 0.712,
+      "step": 4214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0238124490936854e-06,
+      "loss": 0.827,
+      "step": 4215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0227116436987649e-06,
+      "loss": 0.7452,
+      "step": 4216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0216112782226745e-06,
+      "loss": 0.6825,
+      "step": 4217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0205113529930874e-06,
+      "loss": 0.7731,
+      "step": 4218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0194118683375502e-06,
+      "loss": 0.7922,
+      "step": 4219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0183128245834752e-06,
+      "loss": 0.7111,
+      "step": 4220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0172142220581455e-06,
+      "loss": 0.7652,
+      "step": 4221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0161160610887102e-06,
+      "loss": 0.7614,
+      "step": 4222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0150183420021905e-06,
+      "loss": 0.7318,
+      "step": 4223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0139210651254723e-06,
+      "loss": 0.8052,
+      "step": 4224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.012824230785313e-06,
+      "loss": 0.7566,
+      "step": 4225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0117278393083357e-06,
+      "loss": 0.7609,
+      "step": 4226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0106318910210325e-06,
+      "loss": 0.8081,
+      "step": 4227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0095363862497643e-06,
+      "loss": 0.6814,
+      "step": 4228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0084413253207591e-06,
+      "loss": 0.7158,
+      "step": 4229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0073467085601153e-06,
+      "loss": 0.7866,
+      "step": 4230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0062525362937923e-06,
+      "loss": 0.544,
+      "step": 4231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0051588088476238e-06,
+      "loss": 0.5403,
+      "step": 4232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0040655265473098e-06,
+      "loss": 0.7506,
+      "step": 4233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0029726897184158e-06,
+      "loss": 0.5566,
+      "step": 4234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0018802986863743e-06,
+      "loss": 0.7418,
+      "step": 4235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0007883537764876e-06,
+      "loss": 0.7132,
+      "step": 4236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.996968553139245e-07,
+      "loss": 0.7696,
+      "step": 4237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.986058036237185e-07,
+      "loss": 0.7708,
+      "step": 4238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.975151990307733e-07,
+      "loss": 0.7462,
+      "step": 4239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.964250418598577e-07,
+      "loss": 0.7658,
+      "step": 4240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.953353324356058e-07,
+      "loss": 0.7816,
+      "step": 4241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.942460710825214e-07,
+      "loss": 0.8155,
+      "step": 4242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.931572581249742e-07,
+      "loss": 0.7659,
+      "step": 4243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.920688938871992e-07,
+      "loss": 0.6463,
+      "step": 4244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.909809786932966e-07,
+      "loss": 0.7479,
+      "step": 4245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.898935128672366e-07,
+      "loss": 0.6981,
+      "step": 4246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.888064967328533e-07,
+      "loss": 0.6905,
+      "step": 4247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.877199306138469e-07,
+      "loss": 0.7106,
+      "step": 4248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.866338148337827e-07,
+      "loss": 0.739,
+      "step": 4249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855481497160937e-07,
+      "loss": 0.7605,
+      "step": 4250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.844629355840794e-07,
+      "loss": 0.7729,
+      "step": 4251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.833781727609024e-07,
+      "loss": 0.6711,
+      "step": 4252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.822938615695909e-07,
+      "loss": 0.7635,
+      "step": 4253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.812100023330422e-07,
+      "loss": 0.7825,
+      "step": 4254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.80126595374015e-07,
+      "loss": 0.7893,
+      "step": 4255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.790436410151357e-07,
+      "loss": 0.5523,
+      "step": 4256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.779611395788942e-07,
+      "loss": 0.7196,
+      "step": 4257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.768790913876486e-07,
+      "loss": 0.7383,
+      "step": 4258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.757974967636172e-07,
+      "loss": 0.4943,
+      "step": 4259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.747163560288883e-07,
+      "loss": 0.8207,
+      "step": 4260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.736356695054108e-07,
+      "loss": 0.7379,
+      "step": 4261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.72555437515002e-07,
+      "loss": 0.8385,
+      "step": 4262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.714756603793405e-07,
+      "loss": 0.6912,
+      "step": 4263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.703963384199728e-07,
+      "loss": 0.758,
+      "step": 4264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.693174719583068e-07,
+      "loss": 0.7704,
+      "step": 4265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.682390613156153e-07,
+      "loss": 0.8367,
+      "step": 4266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.671611068130375e-07,
+      "loss": 0.5385,
+      "step": 4267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.660836087715754e-07,
+      "loss": 0.7956,
+      "step": 4268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.65006567512095e-07,
+      "loss": 0.5379,
+      "step": 4269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.639299833553253e-07,
+      "loss": 0.7636,
+      "step": 4270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.628538566218607e-07,
+      "loss": 0.7299,
+      "step": 4271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.617781876321598e-07,
+      "loss": 0.8206,
+      "step": 4272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.607029767065432e-07,
+      "loss": 0.7212,
+      "step": 4273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.59628224165195e-07,
+      "loss": 0.7542,
+      "step": 4274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.58553930328165e-07,
+      "loss": 0.7336,
+      "step": 4275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.574800955153652e-07,
+      "loss": 0.6787,
+      "step": 4276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.564067200465702e-07,
+      "loss": 0.7014,
+      "step": 4277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.553338042414175e-07,
+      "loss": 0.749,
+      "step": 4278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.542613484194108e-07,
+      "loss": 0.7573,
+      "step": 4279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.53189352899912e-07,
+      "loss": 0.7686,
+      "step": 4280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.521178180021512e-07,
+      "loss": 0.7286,
+      "step": 4281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.510467440452164e-07,
+      "loss": 0.7781,
+      "step": 4282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.499761313480627e-07,
+      "loss": 0.7453,
+      "step": 4283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.489059802295039e-07,
+      "loss": 0.7888,
+      "step": 4284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.47836291008219e-07,
+      "loss": 0.4974,
+      "step": 4285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.467670640027501e-07,
+      "loss": 0.7163,
+      "step": 4286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.456982995314976e-07,
+      "loss": 0.71,
+      "step": 4287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.44629997912728e-07,
+      "loss": 0.8223,
+      "step": 4288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.435621594645689e-07,
+      "loss": 0.6343,
+      "step": 4289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.424947845050111e-07,
+      "loss": 0.7565,
+      "step": 4290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.414278733519036e-07,
+      "loss": 0.8409,
+      "step": 4291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.403614263229607e-07,
+      "loss": 0.8093,
+      "step": 4292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.392954437357588e-07,
+      "loss": 0.6869,
+      "step": 4293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.382299259077329e-07,
+      "loss": 0.6847,
+      "step": 4294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.371648731561833e-07,
+      "loss": 0.7462,
+      "step": 4295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.361002857982684e-07,
+      "loss": 0.5423,
+      "step": 4296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.350361641510111e-07,
+      "loss": 0.735,
+      "step": 4297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.339725085312926e-07,
+      "loss": 0.7811,
+      "step": 4298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.329093192558583e-07,
+      "loss": 0.8249,
+      "step": 4299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.318465966413115e-07,
+      "loss": 0.7672,
+      "step": 4300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.307843410041206e-07,
+      "loss": 0.7208,
+      "step": 4301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.297225526606105e-07,
+      "loss": 0.7452,
+      "step": 4302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.286612319269706e-07,
+      "loss": 0.7648,
+      "step": 4303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.276003791192492e-07,
+      "loss": 0.7523,
+      "step": 4304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.265399945533543e-07,
+      "loss": 0.7858,
+      "step": 4305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.254800785450568e-07,
+      "loss": 0.7128,
+      "step": 4306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.24420631409988e-07,
+      "loss": 0.7909,
+      "step": 4307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.233616534636372e-07,
+      "loss": 0.7165,
+      "step": 4308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.223031450213554e-07,
+      "loss": 0.7022,
+      "step": 4309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.212451063983541e-07,
+      "loss": 0.811,
+      "step": 4310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.201875379097052e-07,
+      "loss": 0.7785,
+      "step": 4311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.191304398703396e-07,
+      "loss": 0.7177,
+      "step": 4312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.180738125950476e-07,
+      "loss": 0.5139,
+      "step": 4313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.170176563984806e-07,
+      "loss": 0.7302,
+      "step": 4314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.159619715951503e-07,
+      "loss": 0.7462,
+      "step": 4315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.149067584994267e-07,
+      "loss": 0.8019,
+      "step": 4316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.138520174255381e-07,
+      "loss": 0.5454,
+      "step": 4317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.127977486875758e-07,
+      "loss": 0.6729,
+      "step": 4318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.117439525994867e-07,
+      "loss": 0.7112,
+      "step": 4319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.106906294750806e-07,
+      "loss": 0.7463,
+      "step": 4320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.096377796280223e-07,
+      "loss": 0.725,
+      "step": 4321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.085854033718402e-07,
+      "loss": 0.6967,
+      "step": 4322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.075335010199176e-07,
+      "loss": 0.7214,
+      "step": 4323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.064820728854998e-07,
+      "loss": 0.7303,
+      "step": 4324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.054311192816889e-07,
+      "loss": 0.6807,
+      "step": 4325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.043806405214456e-07,
+      "loss": 0.8108,
+      "step": 4326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.033306369175906e-07,
+      "loss": 0.7296,
+      "step": 4327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.022811087828032e-07,
+      "loss": 0.7719,
+      "step": 4328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.012320564296198e-07,
+      "loss": 0.803,
+      "step": 4329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.001834801704345e-07,
+      "loss": 0.8152,
+      "step": 4330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.991353803175021e-07,
+      "loss": 0.7528,
+      "step": 4331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.980877571829344e-07,
+      "loss": 0.8012,
+      "step": 4332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.970406110787008e-07,
+      "loss": 0.7206,
+      "step": 4333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.959939423166275e-07,
+      "loss": 0.7015,
+      "step": 4334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.949477512084014e-07,
+      "loss": 0.7725,
+      "step": 4335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.939020380655661e-07,
+      "loss": 0.7393,
+      "step": 4336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.928568031995219e-07,
+      "loss": 0.77,
+      "step": 4337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.918120469215264e-07,
+      "loss": 0.7926,
+      "step": 4338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.907677695426973e-07,
+      "loss": 0.7674,
+      "step": 4339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.897239713740058e-07,
+      "loss": 0.7736,
+      "step": 4340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.886806527262837e-07,
+      "loss": 0.5148,
+      "step": 4341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.876378139102198e-07,
+      "loss": 0.7163,
+      "step": 4342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.865954552363581e-07,
+      "loss": 0.8102,
+      "step": 4343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.855535770150997e-07,
+      "loss": 0.7682,
+      "step": 4344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.845121795567041e-07,
+      "loss": 0.773,
+      "step": 4345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.834712631712883e-07,
+      "loss": 0.7296,
+      "step": 4346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.824308281688234e-07,
+      "loss": 0.7502,
+      "step": 4347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.813908748591382e-07,
+      "loss": 0.7265,
+      "step": 4348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.803514035519189e-07,
+      "loss": 0.5392,
+      "step": 4349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.793124145567089e-07,
+      "loss": 0.747,
+      "step": 4350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.782739081829054e-07,
+      "loss": 0.7492,
+      "step": 4351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.772358847397621e-07,
+      "loss": 0.6933,
+      "step": 4352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.761983445363917e-07,
+      "loss": 0.7598,
+      "step": 4353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.751612878817614e-07,
+      "loss": 0.7652,
+      "step": 4354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.741247150846938e-07,
+      "loss": 0.717,
+      "step": 4355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.730886264538671e-07,
+      "loss": 0.7771,
+      "step": 4356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.720530222978175e-07,
+      "loss": 0.7979,
+      "step": 4357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.710179029249341e-07,
+      "loss": 0.7637,
+      "step": 4358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.699832686434645e-07,
+      "loss": 0.7792,
+      "step": 4359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.689491197615091e-07,
+      "loss": 0.738,
+      "step": 4360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.679154565870265e-07,
+      "loss": 0.7793,
+      "step": 4361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.668822794278279e-07,
+      "loss": 0.6285,
+      "step": 4362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.658495885915825e-07,
+      "loss": 0.7646,
+      "step": 4363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.648173843858124e-07,
+      "loss": 0.5228,
+      "step": 4364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.637856671178951e-07,
+      "loss": 0.7489,
+      "step": 4365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.627544370950644e-07,
+      "loss": 0.7355,
+      "step": 4366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.61723694624409e-07,
+      "loss": 0.7543,
+      "step": 4367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.606934400128711e-07,
+      "loss": 0.7232,
+      "step": 4368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.596636735672473e-07,
+      "loss": 0.7839,
+      "step": 4369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.586343955941903e-07,
+      "loss": 0.7704,
+      "step": 4370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.576056064002081e-07,
+      "loss": 0.711,
+      "step": 4371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.565773062916608e-07,
+      "loss": 0.7096,
+      "step": 4372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.555494955747629e-07,
+      "loss": 0.8061,
+      "step": 4373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.545221745555851e-07,
+      "loss": 0.8065,
+      "step": 4374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.534953435400522e-07,
+      "loss": 0.6292,
+      "step": 4375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.524690028339416e-07,
+      "loss": 0.717,
+      "step": 4376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.514431527428841e-07,
+      "loss": 0.7755,
+      "step": 4377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.504177935723673e-07,
+      "loss": 0.7411,
+      "step": 4378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.493929256277295e-07,
+      "loss": 0.7574,
+      "step": 4379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.483685492141656e-07,
+      "loss": 0.767,
+      "step": 4380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.473446646367209e-07,
+      "loss": 0.7405,
+      "step": 4381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.463212722002978e-07,
+      "loss": 0.7556,
+      "step": 4382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.452983722096489e-07,
+      "loss": 0.661,
+      "step": 4383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.44275964969383e-07,
+      "loss": 0.6974,
+      "step": 4384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.43254050783959e-07,
+      "loss": 0.7439,
+      "step": 4385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.422326299576928e-07,
+      "loss": 0.6863,
+      "step": 4386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.412117027947492e-07,
+      "loss": 0.7866,
+      "step": 4387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.401912695991493e-07,
+      "loss": 0.8416,
+      "step": 4388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.391713306747673e-07,
+      "loss": 0.7813,
+      "step": 4389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.38151886325326e-07,
+      "loss": 0.7382,
+      "step": 4390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.371329368544048e-07,
+      "loss": 0.6695,
+      "step": 4391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.361144825654349e-07,
+      "loss": 0.7881,
+      "step": 4392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.350965237617017e-07,
+      "loss": 0.7555,
+      "step": 4393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.340790607463375e-07,
+      "loss": 0.5735,
+      "step": 4394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.330620938223328e-07,
+      "loss": 0.7543,
+      "step": 4395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.320456232925286e-07,
+      "loss": 0.7681,
+      "step": 4396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.310296494596159e-07,
+      "loss": 0.7344,
+      "step": 4397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.300141726261418e-07,
+      "loss": 0.6954,
+      "step": 4398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.28999193094501e-07,
+      "loss": 0.6906,
+      "step": 4399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.279847111669442e-07,
+      "loss": 0.7524,
+      "step": 4400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.269707271455701e-07,
+      "loss": 0.6352,
+      "step": 4401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.25957241332333e-07,
+      "loss": 0.6899,
+      "step": 4402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.249442540290359e-07,
+      "loss": 0.7097,
+      "step": 4403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.239317655373336e-07,
+      "loss": 0.7568,
+      "step": 4404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.229197761587337e-07,
+      "loss": 0.7103,
+      "step": 4405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.219082861945953e-07,
+      "loss": 0.6652,
+      "step": 4406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.208972959461278e-07,
+      "loss": 0.78,
+      "step": 4407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.198868057143905e-07,
+      "loss": 0.7455,
+      "step": 4408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.188768158002966e-07,
+      "loss": 0.7711,
+      "step": 4409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.178673265046094e-07,
+      "loss": 0.7801,
+      "step": 4410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.168583381279425e-07,
+      "loss": 0.8024,
+      "step": 4411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.158498509707593e-07,
+      "loss": 0.733,
+      "step": 4412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.148418653333764e-07,
+      "loss": 0.7839,
+      "step": 4413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.138343815159605e-07,
+      "loss": 0.7367,
+      "step": 4414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.128273998185274e-07,
+      "loss": 0.7198,
+      "step": 4415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.118209205409436e-07,
+      "loss": 0.7181,
+      "step": 4416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.10814943982928e-07,
+      "loss": 0.6615,
+      "step": 4417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.098094704440468e-07,
+      "loss": 0.7647,
+      "step": 4418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.088045002237199e-07,
+      "loss": 0.8044,
+      "step": 4419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.078000336212136e-07,
+      "loss": 0.7502,
+      "step": 4420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.067960709356479e-07,
+      "loss": 0.6844,
+      "step": 4421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.057926124659887e-07,
+      "loss": 0.7239,
+      "step": 4422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.047896585110562e-07,
+      "loss": 0.6463,
+      "step": 4423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.037872093695162e-07,
+      "loss": 0.7624,
+      "step": 4424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.027852653398876e-07,
+      "loss": 0.7326,
+      "step": 4425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.017838267205364e-07,
+      "loss": 0.7692,
+      "step": 4426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.007828938096795e-07,
+      "loss": 0.6757,
+      "step": 4427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.99782466905383e-07,
+      "loss": 0.7435,
+      "step": 4428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.98782546305561e-07,
+      "loss": 0.7707,
+      "step": 4429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.977831323079784e-07,
+      "loss": 0.5028,
+      "step": 4430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.967842252102501e-07,
+      "loss": 0.5355,
+      "step": 4431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.957858253098371e-07,
+      "loss": 0.7344,
+      "step": 4432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.947879329040511e-07,
+      "loss": 0.7425,
+      "step": 4433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.937905482900526e-07,
+      "loss": 0.7605,
+      "step": 4434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.927936717648521e-07,
+      "loss": 0.778,
+      "step": 4435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.917973036253066e-07,
+      "loss": 0.79,
+      "step": 4436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.908014441681214e-07,
+      "loss": 0.764,
+      "step": 4437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.89806093689853e-07,
+      "loss": 0.7344,
+      "step": 4438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.888112524869055e-07,
+      "loss": 0.7088,
+      "step": 4439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.878169208555297e-07,
+      "loss": 0.7144,
+      "step": 4440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.868230990918249e-07,
+      "loss": 0.7436,
+      "step": 4441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.858297874917414e-07,
+      "loss": 0.7415,
+      "step": 4442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.848369863510732e-07,
+      "loss": 0.6914,
+      "step": 4443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.838446959654661e-07,
+      "loss": 0.631,
+      "step": 4444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.828529166304128e-07,
+      "loss": 0.7416,
+      "step": 4445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.818616486412525e-07,
+      "loss": 0.7633,
+      "step": 4446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.808708922931726e-07,
+      "loss": 0.7233,
+      "step": 4447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.798806478812088e-07,
+      "loss": 0.7454,
+      "step": 4448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.788909157002456e-07,
+      "loss": 0.529,
+      "step": 4449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.779016960450106e-07,
+      "loss": 0.7141,
+      "step": 4450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.769129892100829e-07,
+      "loss": 0.6712,
+      "step": 4451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.75924795489888e-07,
+      "loss": 0.5349,
+      "step": 4452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.749371151786986e-07,
+      "loss": 0.6695,
+      "step": 4453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.739499485706334e-07,
+      "loss": 0.7401,
+      "step": 4454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.72963295959658e-07,
+      "loss": 0.7536,
+      "step": 4455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.719771576395876e-07,
+      "loss": 0.8608,
+      "step": 4456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.709915339040807e-07,
+      "loss": 0.7839,
+      "step": 4457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.700064250466457e-07,
+      "loss": 0.7996,
+      "step": 4458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.690218313606346e-07,
+      "loss": 0.8074,
+      "step": 4459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.680377531392496e-07,
+      "loss": 0.7762,
+      "step": 4460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.670541906755355e-07,
+      "loss": 0.7181,
+      "step": 4461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.660711442623872e-07,
+      "loss": 0.7697,
+      "step": 4462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.650886141925426e-07,
+      "loss": 0.7199,
+      "step": 4463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.64106600758589e-07,
+      "loss": 0.7646,
+      "step": 4464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.631251042529567e-07,
+      "loss": 0.7554,
+      "step": 4465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.621441249679254e-07,
+      "loss": 0.7321,
+      "step": 4466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.611636631956176e-07,
+      "loss": 0.7447,
+      "step": 4467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.601837192280031e-07,
+      "loss": 0.5449,
+      "step": 4468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.592042933568977e-07,
+      "loss": 0.7953,
+      "step": 4469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.582253858739638e-07,
+      "loss": 0.8169,
+      "step": 4470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.572469970707075e-07,
+      "loss": 0.7247,
+      "step": 4471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.562691272384806e-07,
+      "loss": 0.8274,
+      "step": 4472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.55291776668482e-07,
+      "loss": 0.7333,
+      "step": 4473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.543149456517551e-07,
+      "loss": 0.7856,
+      "step": 4474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.533386344791885e-07,
+      "loss": 0.8015,
+      "step": 4475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.523628434415151e-07,
+      "loss": 0.7064,
+      "step": 4476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.513875728293141e-07,
+      "loss": 0.7768,
+      "step": 4477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.504128229330107e-07,
+      "loss": 0.5049,
+      "step": 4478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494385940428731e-07,
+      "loss": 0.6437,
+      "step": 4479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.484648864490138e-07,
+      "loss": 0.7285,
+      "step": 4480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.474917004413929e-07,
+      "loss": 0.6883,
+      "step": 4481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.465190363098123e-07,
+      "loss": 0.7051,
+      "step": 4482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.455468943439212e-07,
+      "loss": 0.7959,
+      "step": 4483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.445752748332102e-07,
+      "loss": 0.7592,
+      "step": 4484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.436041780670178e-07,
+      "loss": 0.8048,
+      "step": 4485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.426336043345231e-07,
+      "loss": 0.7009,
+      "step": 4486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.416635539247533e-07,
+      "loss": 0.7574,
+      "step": 4487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.40694027126577e-07,
+      "loss": 0.7477,
+      "step": 4488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.397250242287065e-07,
+      "loss": 0.7617,
+      "step": 4489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.387565455197004e-07,
+      "loss": 0.722,
+      "step": 4490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.377885912879604e-07,
+      "loss": 0.8418,
+      "step": 4491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.368211618217325e-07,
+      "loss": 0.7499,
+      "step": 4492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.358542574091033e-07,
+      "loss": 0.7106,
+      "step": 4493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.348878783380062e-07,
+      "loss": 0.7194,
+      "step": 4494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.339220248962187e-07,
+      "loss": 0.7882,
+      "step": 4495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.329566973713595e-07,
+      "loss": 0.6647,
+      "step": 4496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.31991896050891e-07,
+      "loss": 0.7613,
+      "step": 4497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.310276212221196e-07,
+      "loss": 0.6844,
+      "step": 4498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.300638731721965e-07,
+      "loss": 0.7604,
+      "step": 4499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.291006521881122e-07,
+      "loss": 0.7757,
+      "step": 4500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.281379585567042e-07,
+      "loss": 0.8139,
+      "step": 4501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.271757925646494e-07,
+      "loss": 0.7188,
+      "step": 4502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.262141544984713e-07,
+      "loss": 0.7299,
+      "step": 4503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.252530446445321e-07,
+      "loss": 0.7782,
+      "step": 4504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.242924632890411e-07,
+      "loss": 0.6979,
+      "step": 4505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.233324107180464e-07,
+      "loss": 0.7932,
+      "step": 4506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.223728872174401e-07,
+      "loss": 0.7058,
+      "step": 4507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.214138930729573e-07,
+      "loss": 0.7939,
+      "step": 4508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.204554285701759e-07,
+      "loss": 0.7972,
+      "step": 4509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.19497493994514e-07,
+      "loss": 0.7078,
+      "step": 4510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.185400896312328e-07,
+      "loss": 0.7404,
+      "step": 4511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.175832157654364e-07,
+      "loss": 0.7703,
+      "step": 4512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.166268726820713e-07,
+      "loss": 0.7094,
+      "step": 4513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.156710606659242e-07,
+      "loss": 0.6795,
+      "step": 4514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.147157800016238e-07,
+      "loss": 0.7717,
+      "step": 4515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.137610309736421e-07,
+      "loss": 0.7567,
+      "step": 4516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.128068138662925e-07,
+      "loss": 0.7152,
+      "step": 4517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.118531289637287e-07,
+      "loss": 0.7234,
+      "step": 4518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.108999765499464e-07,
+      "loss": 0.736,
+      "step": 4519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.099473569087839e-07,
+      "loss": 0.7383,
+      "step": 4520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.089952703239186e-07,
+      "loss": 0.7741,
+      "step": 4521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.080437170788723e-07,
+      "loss": 0.7176,
+      "step": 4522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.070926974570044e-07,
+      "loss": 0.7468,
+      "step": 4523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.061422117415187e-07,
+      "loss": 0.7188,
+      "step": 4524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.051922602154574e-07,
+      "loss": 0.6823,
+      "step": 4525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.042428431617057e-07,
+      "loss": 0.6842,
+      "step": 4526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.03293960862988e-07,
+      "loss": 0.7398,
+      "step": 4527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.023456136018697e-07,
+      "loss": 0.6704,
+      "step": 4528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.013978016607578e-07,
+      "loss": 0.6952,
+      "step": 4529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.004505253218998e-07,
+      "loss": 0.7524,
+      "step": 4530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.995037848673833e-07,
+      "loss": 0.6588,
+      "step": 4531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.98557580579135e-07,
+      "loss": 0.7551,
+      "step": 4532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.976119127389241e-07,
+      "loss": 0.7038,
+      "step": 4533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.966667816283598e-07,
+      "loss": 0.8125,
+      "step": 4534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.957221875288905e-07,
+      "loss": 0.7206,
+      "step": 4535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.947781307218038e-07,
+      "loss": 0.6638,
+      "step": 4536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.938346114882297e-07,
+      "loss": 0.7958,
+      "step": 4537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.928916301091376e-07,
+      "loss": 0.7649,
+      "step": 4538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.919491868653352e-07,
+      "loss": 0.7499,
+      "step": 4539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9100728203747e-07,
+      "loss": 0.7227,
+      "step": 4540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.900659159060311e-07,
+      "loss": 0.6676,
+      "step": 4541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.891250887513468e-07,
+      "loss": 0.7864,
+      "step": 4542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.881848008535832e-07,
+      "loss": 0.7691,
+      "step": 4543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.872450524927463e-07,
+      "loss": 0.7622,
+      "step": 4544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.863058439486831e-07,
+      "loss": 0.8024,
+      "step": 4545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.853671755010773e-07,
+      "loss": 0.7577,
+      "step": 4546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.844290474294544e-07,
+      "loss": 0.7093,
+      "step": 4547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.834914600131779e-07,
+      "loss": 0.7578,
+      "step": 4548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.825544135314496e-07,
+      "loss": 0.6987,
+      "step": 4549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.816179082633101e-07,
+      "loss": 0.7868,
+      "step": 4550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.806819444876398e-07,
+      "loss": 0.7805,
+      "step": 4551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.797465224831596e-07,
+      "loss": 0.7253,
+      "step": 4552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.788116425284236e-07,
+      "loss": 0.7493,
+      "step": 4553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.778773049018298e-07,
+      "loss": 0.7088,
+      "step": 4554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.769435098816121e-07,
+      "loss": 0.8713,
+      "step": 4555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.760102577458449e-07,
+      "loss": 0.6863,
+      "step": 4556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.750775487724384e-07,
+      "loss": 0.7463,
+      "step": 4557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.741453832391418e-07,
+      "loss": 0.6638,
+      "step": 4558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.732137614235442e-07,
+      "loss": 0.6917,
+      "step": 4559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7228268360307e-07,
+      "loss": 0.7697,
+      "step": 4560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.713521500549841e-07,
+      "loss": 0.7337,
+      "step": 4561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.704221610563874e-07,
+      "loss": 0.7144,
+      "step": 4562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.694927168842208e-07,
+      "loss": 0.8225,
+      "step": 4563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.685638178152598e-07,
+      "loss": 0.7887,
+      "step": 4564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.676354641261215e-07,
+      "loss": 0.7186,
+      "step": 4565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.667076560932578e-07,
+      "loss": 0.7933,
+      "step": 4566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.657803939929577e-07,
+      "loss": 0.7344,
+      "step": 4567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.648536781013496e-07,
+      "loss": 0.7113,
+      "step": 4568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.639275086943995e-07,
+      "loss": 0.6072,
+      "step": 4569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.630018860479084e-07,
+      "loss": 0.5205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.620768104375153e-07,
+      "loss": 0.7638,
+      "step": 4571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.61152282138697e-07,
+      "loss": 0.6767,
+      "step": 4572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.602283014267682e-07,
+      "loss": 0.6964,
+      "step": 4573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.593048685768785e-07,
+      "loss": 0.7879,
+      "step": 4574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.58381983864014e-07,
+      "loss": 0.7673,
+      "step": 4575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.574596475630001e-07,
+      "loss": 0.762,
+      "step": 4576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.565378599484978e-07,
+      "loss": 0.7255,
+      "step": 4577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.556166212950039e-07,
+      "loss": 0.7162,
+      "step": 4578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.546959318768514e-07,
+      "loss": 0.7523,
+      "step": 4579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.537757919682114e-07,
+      "loss": 0.7498,
+      "step": 4580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.528562018430915e-07,
+      "loss": 0.7623,
+      "step": 4581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.519371617753337e-07,
+      "loss": 0.7946,
+      "step": 4582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.510186720386166e-07,
+      "loss": 0.7313,
+      "step": 4583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.501007329064566e-07,
+      "loss": 0.5399,
+      "step": 4584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.491833446522039e-07,
+      "loss": 0.7358,
+      "step": 4585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.482665075490471e-07,
+      "loss": 0.7484,
+      "step": 4586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.473502218700081e-07,
+      "loss": 0.7131,
+      "step": 4587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.464344878879469e-07,
+      "loss": 0.7218,
+      "step": 4588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.455193058755568e-07,
+      "loss": 0.7336,
+      "step": 4589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.446046761053698e-07,
+      "loss": 0.7694,
+      "step": 4590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.436905988497505e-07,
+      "loss": 0.594,
+      "step": 4591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.427770743809e-07,
+      "loss": 0.7517,
+      "step": 4592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.418641029708552e-07,
+      "loss": 0.7669,
+      "step": 4593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.409516848914885e-07,
+      "loss": 0.5416,
+      "step": 4594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.400398204145086e-07,
+      "loss": 0.6942,
+      "step": 4595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.391285098114544e-07,
+      "loss": 0.8045,
+      "step": 4596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.382177533537051e-07,
+      "loss": 0.6833,
+      "step": 4597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.37307551312474e-07,
+      "loss": 0.7685,
+      "step": 4598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.363979039588075e-07,
+      "loss": 0.7335,
+      "step": 4599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.354888115635871e-07,
+      "loss": 0.7386,
+      "step": 4600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.345802743975302e-07,
+      "loss": 0.7729,
+      "step": 4601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.33672292731189e-07,
+      "loss": 0.8004,
+      "step": 4602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.327648668349481e-07,
+      "loss": 0.7412,
+      "step": 4603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.3185799697903e-07,
+      "loss": 0.73,
+      "step": 4604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.309516834334886e-07,
+      "loss": 0.7288,
+      "step": 4605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.300459264682127e-07,
+      "loss": 0.7704,
+      "step": 4606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.291407263529264e-07,
+      "loss": 0.8784,
+      "step": 4607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.282360833571885e-07,
+      "loss": 0.8197,
+      "step": 4608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.2733199775039e-07,
+      "loss": 0.8532,
+      "step": 4609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.264284698017562e-07,
+      "loss": 0.6873,
+      "step": 4610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.255254997803473e-07,
+      "loss": 0.76,
+      "step": 4611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.24623087955058e-07,
+      "loss": 0.6797,
+      "step": 4612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.23721234594615e-07,
+      "loss": 0.7272,
+      "step": 4613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.228199399675789e-07,
+      "loss": 0.7608,
+      "step": 4614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.219192043423447e-07,
+      "loss": 0.712,
+      "step": 4615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.21019027987142e-07,
+      "loss": 0.6947,
+      "step": 4616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.201194111700318e-07,
+      "loss": 0.772,
+      "step": 4617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.192203541589079e-07,
+      "loss": 0.7781,
+      "step": 4618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.183218572215002e-07,
+      "loss": 0.6158,
+      "step": 4619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.174239206253707e-07,
+      "loss": 0.7653,
+      "step": 4620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.165265446379134e-07,
+      "loss": 0.7781,
+      "step": 4621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.156297295263558e-07,
+      "loss": 0.6928,
+      "step": 4622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.147334755577597e-07,
+      "loss": 0.6818,
+      "step": 4623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.138377829990178e-07,
+      "loss": 0.799,
+      "step": 4624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.12942652116858e-07,
+      "loss": 0.7559,
+      "step": 4625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.120480831778378e-07,
+      "loss": 0.6588,
+      "step": 4626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.111540764483509e-07,
+      "loss": 0.7449,
+      "step": 4627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.102606321946206e-07,
+      "loss": 0.7849,
+      "step": 4628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.093677506827048e-07,
+      "loss": 0.85,
+      "step": 4629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.084754321784927e-07,
+      "loss": 0.7177,
+      "step": 4630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.075836769477053e-07,
+      "loss": 0.5665,
+      "step": 4631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.06692485255897e-07,
+      "loss": 0.7394,
+      "step": 4632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.058018573684551e-07,
+      "loss": 0.6684,
+      "step": 4633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.049117935505974e-07,
+      "loss": 0.7507,
+      "step": 4634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.040222940673731e-07,
+      "loss": 0.5383,
+      "step": 4635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.031333591836657e-07,
+      "loss": 0.7607,
+      "step": 4636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.022449891641899e-07,
+      "loss": 0.5114,
+      "step": 4637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.013571842734908e-07,
+      "loss": 0.5438,
+      "step": 4638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.004699447759454e-07,
+      "loss": 0.7261,
+      "step": 4639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.995832709357641e-07,
+      "loss": 0.7054,
+      "step": 4640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.986971630169883e-07,
+      "loss": 0.7112,
+      "step": 4641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.978116212834895e-07,
+      "loss": 0.8126,
+      "step": 4642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.969266459989706e-07,
+      "loss": 0.7679,
+      "step": 4643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.960422374269687e-07,
+      "loss": 0.7124,
+      "step": 4644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.951583958308479e-07,
+      "loss": 0.706,
+      "step": 4645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.942751214738079e-07,
+      "loss": 0.7913,
+      "step": 4646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.933924146188754e-07,
+      "loss": 0.7129,
+      "step": 4647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.925102755289113e-07,
+      "loss": 0.5583,
+      "step": 4648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.91628704466605e-07,
+      "loss": 0.5389,
+      "step": 4649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.907477016944779e-07,
+      "loss": 0.6775,
+      "step": 4650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.898672674748843e-07,
+      "loss": 0.7287,
+      "step": 4651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.889874020700037e-07,
+      "loss": 0.7789,
+      "step": 4652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.881081057418511e-07,
+      "loss": 0.5354,
+      "step": 4653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.872293787522706e-07,
+      "loss": 0.7234,
+      "step": 4654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.863512213629374e-07,
+      "loss": 0.7562,
+      "step": 4655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.854736338353542e-07,
+      "loss": 0.7698,
+      "step": 4656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.84596616430857e-07,
+      "loss": 0.779,
+      "step": 4657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.837201694106112e-07,
+      "loss": 0.7376,
+      "step": 4658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.828442930356132e-07,
+      "loss": 0.7367,
+      "step": 4659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.819689875666875e-07,
+      "loss": 0.7177,
+      "step": 4660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.810942532644892e-07,
+      "loss": 0.7297,
+      "step": 4661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.80220090389505e-07,
+      "loss": 0.7355,
+      "step": 4662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.793464992020487e-07,
+      "loss": 0.7971,
+      "step": 4663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.784734799622666e-07,
+      "loss": 0.7112,
+      "step": 4664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.776010329301324e-07,
+      "loss": 0.7346,
+      "step": 4665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.767291583654513e-07,
+      "loss": 0.5365,
+      "step": 4666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.758578565278564e-07,
+      "loss": 0.7007,
+      "step": 4667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.749871276768118e-07,
+      "loss": 0.7512,
+      "step": 4668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.741169720716097e-07,
+      "loss": 0.7898,
+      "step": 4669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.732473899713712e-07,
+      "loss": 0.7221,
+      "step": 4670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.723783816350484e-07,
+      "loss": 0.7441,
+      "step": 4671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.715099473214219e-07,
+      "loss": 0.7152,
+      "step": 4672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.706420872891008e-07,
+      "loss": 0.7188,
+      "step": 4673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.697748017965227e-07,
+      "loss": 0.7517,
+      "step": 4674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.689080911019553e-07,
+      "loss": 0.7506,
+      "step": 4675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.680419554634955e-07,
+      "loss": 0.7555,
+      "step": 4676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.671763951390676e-07,
+      "loss": 0.7085,
+      "step": 4677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.663114103864242e-07,
+      "loss": 0.7403,
+      "step": 4678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.654470014631486e-07,
+      "loss": 0.7762,
+      "step": 4679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.645831686266515e-07,
+      "loss": 0.7003,
+      "step": 4680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.637199121341716e-07,
+      "loss": 0.8509,
+      "step": 4681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.628572322427755e-07,
+      "loss": 0.7499,
+      "step": 4682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.619951292093609e-07,
+      "loss": 0.7199,
+      "step": 4683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.611336032906498e-07,
+      "loss": 0.7287,
+      "step": 4684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.602726547431961e-07,
+      "loss": 0.7416,
+      "step": 4685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.594122838233782e-07,
+      "loss": 0.7676,
+      "step": 4686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.585524907874062e-07,
+      "loss": 0.7156,
+      "step": 4687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.576932758913142e-07,
+      "loss": 0.7373,
+      "step": 4688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.568346393909683e-07,
+      "loss": 0.7423,
+      "step": 4689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.559765815420586e-07,
+      "loss": 0.7842,
+      "step": 4690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.551191026001046e-07,
+      "loss": 0.6745,
+      "step": 4691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.542622028204536e-07,
+      "loss": 0.7581,
+      "step": 4692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.534058824582808e-07,
+      "loss": 0.7466,
+      "step": 4693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.525501417685874e-07,
+      "loss": 0.7689,
+      "step": 4694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.516949810062025e-07,
+      "loss": 0.7169,
+      "step": 4695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.508404004257828e-07,
+      "loss": 0.7536,
+      "step": 4696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.499864002818126e-07,
+      "loss": 0.6542,
+      "step": 4697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.491329808286044e-07,
+      "loss": 0.7722,
+      "step": 4698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.482801423202933e-07,
+      "loss": 0.781,
+      "step": 4699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.474278850108461e-07,
+      "loss": 0.7248,
+      "step": 4700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.465762091540553e-07,
+      "loss": 0.6789,
+      "step": 4701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.457251150035389e-07,
+      "loss": 0.7639,
+      "step": 4702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448746028127425e-07,
+      "loss": 0.8107,
+      "step": 4703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440246728349385e-07,
+      "loss": 0.7343,
+      "step": 4704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.431753253232269e-07,
+      "loss": 0.7278,
+      "step": 4705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.423265605305322e-07,
+      "loss": 0.6891,
+      "step": 4706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.414783787096073e-07,
+      "loss": 0.8053,
+      "step": 4707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.406307801130301e-07,
+      "loss": 0.7242,
+      "step": 4708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.397837649932044e-07,
+      "loss": 0.7216,
+      "step": 4709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.389373336023623e-07,
+      "loss": 0.7226,
+      "step": 4710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.380914861925615e-07,
+      "loss": 0.7264,
+      "step": 4711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.372462230156847e-07,
+      "loss": 0.7823,
+      "step": 4712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.364015443234399e-07,
+      "loss": 0.7755,
+      "step": 4713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.355574503673638e-07,
+      "loss": 0.7387,
+      "step": 4714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.347139413988175e-07,
+      "loss": 0.7888,
+      "step": 4715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.338710176689879e-07,
+      "loss": 0.5289,
+      "step": 4716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.330286794288863e-07,
+      "loss": 0.7712,
+      "step": 4717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.32186926929352e-07,
+      "loss": 0.7735,
+      "step": 4718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.313457604210495e-07,
+      "loss": 0.8355,
+      "step": 4719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.305051801544672e-07,
+      "loss": 0.6745,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.296651863799196e-07,
+      "loss": 0.7579,
+      "step": 4721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.288257793475479e-07,
+      "loss": 0.7462,
+      "step": 4722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.279869593073161e-07,
+      "loss": 0.5224,
+      "step": 4723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.271487265090163e-07,
+      "loss": 0.777,
+      "step": 4724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.263110812022629e-07,
+      "loss": 0.7299,
+      "step": 4725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.254740236364977e-07,
+      "loss": 0.7237,
+      "step": 4726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.246375540609855e-07,
+      "loss": 0.6466,
+      "step": 4727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.238016727248179e-07,
+      "loss": 0.6795,
+      "step": 4728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.229663798769099e-07,
+      "loss": 0.7099,
+      "step": 4729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.221316757660014e-07,
+      "loss": 0.6494,
+      "step": 4730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.212975606406575e-07,
+      "loss": 0.695,
+      "step": 4731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.204640347492684e-07,
+      "loss": 0.7609,
+      "step": 4732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.19631098340048e-07,
+      "loss": 0.7039,
+      "step": 4733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.187987516610335e-07,
+      "loss": 0.7738,
+      "step": 4734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.179669949600887e-07,
+      "loss": 0.7693,
+      "step": 4735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.171358284849015e-07,
+      "loss": 0.6696,
+      "step": 4736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.163052524829831e-07,
+      "loss": 0.705,
+      "step": 4737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.154752672016678e-07,
+      "loss": 0.7364,
+      "step": 4738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.146458728881162e-07,
+      "loss": 0.7161,
+      "step": 4739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.138170697893128e-07,
+      "loss": 0.7338,
+      "step": 4740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.129888581520645e-07,
+      "loss": 0.7179,
+      "step": 4741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.121612382230021e-07,
+      "loss": 0.7238,
+      "step": 4742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.113342102485819e-07,
+      "loss": 0.6662,
+      "step": 4743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.105077744750831e-07,
+      "loss": 0.7344,
+      "step": 4744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.096819311486082e-07,
+      "loss": 0.7856,
+      "step": 4745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.088566805150827e-07,
+      "loss": 0.6765,
+      "step": 4746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.080320228202576e-07,
+      "loss": 0.5333,
+      "step": 4747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.072079583097051e-07,
+      "loss": 0.7348,
+      "step": 4748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.063844872288223e-07,
+      "loss": 0.7264,
+      "step": 4749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.055616098228286e-07,
+      "loss": 0.7753,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.047393263367681e-07,
+      "loss": 0.718,
+      "step": 4751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.039176370155058e-07,
+      "loss": 0.7047,
+      "step": 4752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.030965421037317e-07,
+      "loss": 0.7706,
+      "step": 4753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.022760418459582e-07,
+      "loss": 0.6972,
+      "step": 4754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.014561364865192e-07,
+      "loss": 0.6853,
+      "step": 4755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.006368262695737e-07,
+      "loss": 0.6309,
+      "step": 4756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.998181114391023e-07,
+      "loss": 0.7475,
+      "step": 4757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.989999922389102e-07,
+      "loss": 0.7092,
+      "step": 4758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981824689126205e-07,
+      "loss": 0.7205,
+      "step": 4759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.973655417036835e-07,
+      "loss": 0.6386,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.965492108553707e-07,
+      "loss": 0.54,
+      "step": 4761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.957334766107744e-07,
+      "loss": 0.7742,
+      "step": 4762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.949183392128121e-07,
+      "loss": 0.6132,
+      "step": 4763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.941037989042205e-07,
+      "loss": 0.7764,
+      "step": 4764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.932898559275612e-07,
+      "loss": 0.7116,
+      "step": 4765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.924765105252158e-07,
+      "loss": 0.8008,
+      "step": 4766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.916637629393894e-07,
+      "loss": 0.753,
+      "step": 4767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.908516134121086e-07,
+      "loss": 0.7388,
+      "step": 4768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.900400621852208e-07,
+      "loss": 0.6402,
+      "step": 4769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.89229109500397e-07,
+      "loss": 0.8091,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.884187555991301e-07,
+      "loss": 0.7664,
+      "step": 4771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.876090007227327e-07,
+      "loss": 0.7247,
+      "step": 4772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867998451123399e-07,
+      "loss": 0.675,
+      "step": 4773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.859912890089088e-07,
+      "loss": 0.7399,
+      "step": 4774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.851833326532188e-07,
+      "loss": 0.7407,
+      "step": 4775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.843759762858688e-07,
+      "loss": 0.7839,
+      "step": 4776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.835692201472794e-07,
+      "loss": 0.7562,
+      "step": 4777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.827630644776935e-07,
+      "loss": 0.7799,
+      "step": 4778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.819575095171753e-07,
+      "loss": 0.7562,
+      "step": 4779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.811525555056085e-07,
+      "loss": 0.7296,
+      "step": 4780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.803482026826986e-07,
+      "loss": 0.7145,
+      "step": 4781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.795444512879727e-07,
+      "loss": 0.8154,
+      "step": 4782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.787413015607789e-07,
+      "loss": 0.6486,
+      "step": 4783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.779387537402852e-07,
+      "loss": 0.7091,
+      "step": 4784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7713680806548e-07,
+      "loss": 0.7588,
+      "step": 4785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7633546477517443e-07,
+      "loss": 0.7449,
+      "step": 4786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.755347241079977e-07,
+      "loss": 0.8021,
+      "step": 4787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.747345863024019e-07,
+      "loss": 0.6667,
+      "step": 4788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.739350515966576e-07,
+      "loss": 0.7442,
+      "step": 4789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.731361202288576e-07,
+      "loss": 0.6726,
+      "step": 4790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.72337792436913e-07,
+      "loss": 0.6732,
+      "step": 4791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7154006845855747e-07,
+      "loss": 0.8596,
+      "step": 4792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7074294853134334e-07,
+      "loss": 0.7546,
+      "step": 4793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6994643289264257e-07,
+      "loss": 0.7577,
+      "step": 4794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6915052177964827e-07,
+      "loss": 0.7133,
+      "step": 4795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6835521542937466e-07,
+      "loss": 0.7826,
+      "step": 4796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.675605140786532e-07,
+      "loss": 0.7938,
+      "step": 4797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.667664179641362e-07,
+      "loss": 0.7541,
+      "step": 4798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6597292732229626e-07,
+      "loss": 0.7603,
+      "step": 4799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.651800423894265e-07,
+      "loss": 0.8069,
+      "step": 4800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6438776340163775e-07,
+      "loss": 0.703,
+      "step": 4801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6359609059486054e-07,
+      "loss": 0.7478,
+      "step": 4802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6280502420484624e-07,
+      "loss": 0.7316,
+      "step": 4803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.620145644671656e-07,
+      "loss": 0.7254,
+      "step": 4804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.612247116172075e-07,
+      "loss": 0.7043,
+      "step": 4805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6043546589017993e-07,
+      "loss": 0.7812,
+      "step": 4806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5964682752111245e-07,
+      "loss": 0.6837,
+      "step": 4807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5885879674485043e-07,
+      "loss": 0.7857,
+      "step": 4808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5807137379606155e-07,
+      "loss": 0.6868,
+      "step": 4809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.572845589092295e-07,
+      "loss": 0.6961,
+      "step": 4810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5649835231866004e-07,
+      "loss": 0.7102,
+      "step": 4811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.557127542584741e-07,
+      "loss": 0.7288,
+      "step": 4812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5492776496261493e-07,
+      "loss": 0.723,
+      "step": 4813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.54143384664843e-07,
+      "loss": 0.7362,
+      "step": 4814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.53359613598737e-07,
+      "loss": 0.7555,
+      "step": 4815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5257645199769383e-07,
+      "loss": 0.8029,
+      "step": 4816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5179390009493006e-07,
+      "loss": 0.7292,
+      "step": 4817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5101195812348176e-07,
+      "loss": 0.6652,
+      "step": 4818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.502306263162004e-07,
+      "loss": 0.7655,
+      "step": 4819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4944990490575686e-07,
+      "loss": 0.711,
+      "step": 4820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4866979412464153e-07,
+      "loss": 0.6863,
+      "step": 4821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4789029420516274e-07,
+      "loss": 0.5128,
+      "step": 4822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4711140537944535e-07,
+      "loss": 0.7085,
+      "step": 4823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.463331278794325e-07,
+      "loss": 0.7153,
+      "step": 4824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.455554619368874e-07,
+      "loss": 0.7265,
+      "step": 4825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.447784077833883e-07,
+      "loss": 0.6887,
+      "step": 4826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4400196565033415e-07,
+      "loss": 0.7511,
+      "step": 4827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432261357689385e-07,
+      "loss": 0.727,
+      "step": 4828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424509183702355e-07,
+      "loss": 0.656,
+      "step": 4829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.416763136850749e-07,
+      "loss": 0.7642,
+      "step": 4830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4090232194412525e-07,
+      "loss": 0.5199,
+      "step": 4831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4012894337787213e-07,
+      "loss": 0.4935,
+      "step": 4832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3935617821661723e-07,
+      "loss": 0.7462,
+      "step": 4833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.385840266904817e-07,
+      "loss": 0.7487,
+      "step": 4834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3781248902940385e-07,
+      "loss": 0.7457,
+      "step": 4835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3704156546313714e-07,
+      "loss": 0.7409,
+      "step": 4836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3627125622125345e-07,
+      "loss": 0.78,
+      "step": 4837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3550156153314207e-07,
+      "loss": 0.7924,
+      "step": 4838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3473248162800905e-07,
+      "loss": 0.6647,
+      "step": 4839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3396401673487743e-07,
+      "loss": 0.7177,
+      "step": 4840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3319616708258545e-07,
+      "loss": 0.7773,
+      "step": 4841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3242893289979077e-07,
+      "loss": 0.6697,
+      "step": 4842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3166231441496675e-07,
+      "loss": 0.6693,
+      "step": 4843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.30896311856403e-07,
+      "loss": 0.7118,
+      "step": 4844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3013092545220513e-07,
+      "loss": 0.7365,
+      "step": 4845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.293661554302972e-07,
+      "loss": 0.717,
+      "step": 4846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2860200201841754e-07,
+      "loss": 0.7926,
+      "step": 4847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.278384654441228e-07,
+      "loss": 0.72,
+      "step": 4848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2707554593478415e-07,
+      "loss": 0.7111,
+      "step": 4849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2631324371759125e-07,
+      "loss": 0.7111,
+      "step": 4850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.255515590195472e-07,
+      "loss": 0.5056,
+      "step": 4851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2479049206747366e-07,
+      "loss": 0.7232,
+      "step": 4852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.240300430880062e-07,
+      "loss": 0.7275,
+      "step": 4853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.232702123075988e-07,
+      "loss": 0.7421,
+      "step": 4854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.225109999525184e-07,
+      "loss": 0.7855,
+      "step": 4855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2175240624885094e-07,
+      "loss": 0.8156,
+      "step": 4856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.209944314224959e-07,
+      "loss": 0.725,
+      "step": 4857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.202370756991686e-07,
+      "loss": 0.7538,
+      "step": 4858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1948033930440083e-07,
+      "loss": 0.7298,
+      "step": 4859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187242224635399e-07,
+      "loss": 0.7357,
+      "step": 4860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.179687254017495e-07,
+      "loss": 0.7799,
+      "step": 4861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.172138483440052e-07,
+      "loss": 0.7283,
+      "step": 4862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.164595915151015e-07,
+      "loss": 0.5156,
+      "step": 4863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1570595513964804e-07,
+      "loss": 0.7558,
+      "step": 4864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.149529394420679e-07,
+      "loss": 0.7514,
+      "step": 4865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.142005446465994e-07,
+      "loss": 0.8168,
+      "step": 4866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1344877097729716e-07,
+      "loss": 0.7382,
+      "step": 4867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.126976186580317e-07,
+      "loss": 0.7112,
+      "step": 4868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.119470879124854e-07,
+      "loss": 0.7293,
+      "step": 4869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1119717896415887e-07,
+      "loss": 0.7399,
+      "step": 4870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1044789203636524e-07,
+      "loss": 0.7211,
+      "step": 4871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0969922735223296e-07,
+      "loss": 0.7045,
+      "step": 4872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0895118513470553e-07,
+      "loss": 0.7486,
+      "step": 4873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0820376560654183e-07,
+      "loss": 0.7485,
+      "step": 4874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0745696899031406e-07,
+      "loss": 0.7331,
+      "step": 4875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0671079550840817e-07,
+      "loss": 0.777,
+      "step": 4876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0596524538302707e-07,
+      "loss": 0.7353,
+      "step": 4877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0522031883618665e-07,
+      "loss": 0.7247,
+      "step": 4878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.044760160897168e-07,
+      "loss": 0.7684,
+      "step": 4879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0373233736526155e-07,
+      "loss": 0.6612,
+      "step": 4880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.029892828842799e-07,
+      "loss": 0.6734,
+      "step": 4881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.022468528680451e-07,
+      "loss": 0.7809,
+      "step": 4882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.015050475376436e-07,
+      "loss": 0.7297,
+      "step": 4883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0076386711397574e-07,
+      "loss": 0.6441,
+      "step": 4884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0002331181775684e-07,
+      "loss": 0.6841,
+      "step": 4885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9928338186951436e-07,
+      "loss": 0.7523,
+      "step": 4886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.985440774895924e-07,
+      "loss": 0.6914,
+      "step": 4887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9780539889814505e-07,
+      "loss": 0.7504,
+      "step": 4888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.970673463151434e-07,
+      "loss": 0.682,
+      "step": 4889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.963299199603696e-07,
+      "loss": 0.6793,
+      "step": 4890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9559312005342157e-07,
+      "loss": 0.5004,
+      "step": 4891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.948569468137084e-07,
+      "loss": 0.7346,
+      "step": 4892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9412140046045487e-07,
+      "loss": 0.8,
+      "step": 4893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9338648121269643e-07,
+      "loss": 0.7578,
+      "step": 4894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9265218928928484e-07,
+      "loss": 0.7311,
+      "step": 4895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9191852490888247e-07,
+      "loss": 0.7612,
+      "step": 4896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9118548828996547e-07,
+      "loss": 0.6728,
+      "step": 4897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9045307965082407e-07,
+      "loss": 0.7711,
+      "step": 4898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8972129920956116e-07,
+      "loss": 0.7605,
+      "step": 4899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.889901471840918e-07,
+      "loss": 0.8306,
+      "step": 4900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.882596237921435e-07,
+      "loss": 0.6966,
+      "step": 4901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8752972925125835e-07,
+      "loss": 0.7468,
+      "step": 4902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.868004637787906e-07,
+      "loss": 0.7345,
+      "step": 4903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.860718275919062e-07,
+      "loss": 0.7102,
+      "step": 4904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.853438209075838e-07,
+      "loss": 0.7705,
+      "step": 4905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.846164439426156e-07,
+      "loss": 0.6747,
+      "step": 4906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.838896969136066e-07,
+      "loss": 0.758,
+      "step": 4907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8316358003697237e-07,
+      "loss": 0.6676,
+      "step": 4908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.824380935289418e-07,
+      "loss": 0.822,
+      "step": 4909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8171323760555654e-07,
+      "loss": 0.7659,
+      "step": 4910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8098901248266954e-07,
+      "loss": 0.7467,
+      "step": 4911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.802654183759474e-07,
+      "loss": 0.7339,
+      "step": 4912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7954245550086644e-07,
+      "loss": 0.6753,
+      "step": 4913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.788201240727177e-07,
+      "loss": 0.7946,
+      "step": 4914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7809842430660134e-07,
+      "loss": 0.5333,
+      "step": 4915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7737735641743213e-07,
+      "loss": 0.6745,
+      "step": 4916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.766569206199361e-07,
+      "loss": 0.7073,
+      "step": 4917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7593711712864857e-07,
+      "loss": 0.8317,
+      "step": 4918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.752179461579192e-07,
+      "loss": 0.7106,
+      "step": 4919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.744994079219083e-07,
+      "loss": 0.7008,
+      "step": 4920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.737815026345898e-07,
+      "loss": 0.7802,
+      "step": 4921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.730642305097448e-07,
+      "loss": 0.716,
+      "step": 4922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7234759176096923e-07,
+      "loss": 0.7814,
+      "step": 4923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7163158660167035e-07,
+      "loss": 0.7029,
+      "step": 4924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.709162152450646e-07,
+      "loss": 0.8375,
+      "step": 4925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7020147790418266e-07,
+      "loss": 0.6581,
+      "step": 4926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6948737479186316e-07,
+      "loss": 0.7091,
+      "step": 4927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.687739061207585e-07,
+      "loss": 0.712,
+      "step": 4928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6806107210333034e-07,
+      "loss": 0.8021,
+      "step": 4929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6734887295185343e-07,
+      "loss": 0.7115,
+      "step": 4930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.666373088784106e-07,
+      "loss": 0.7047,
+      "step": 4931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.659263800948984e-07,
+      "loss": 0.7257,
+      "step": 4932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6521608681302195e-07,
+      "loss": 0.7159,
+      "step": 4933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.64506429244299e-07,
+      "loss": 0.7719,
+      "step": 4934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6379740760005697e-07,
+      "loss": 0.7744,
+      "step": 4935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630890220914329e-07,
+      "loss": 0.8714,
+      "step": 4936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623812729293763e-07,
+      "loss": 0.7681,
+      "step": 4937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616741603246474e-07,
+      "loss": 0.659,
+      "step": 4938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.609676844878146e-07,
+      "loss": 0.7744,
+      "step": 4939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.602618456292578e-07,
+      "loss": 0.7368,
+      "step": 4940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.595566439591683e-07,
+      "loss": 0.8087,
+      "step": 4941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5885207968754663e-07,
+      "loss": 0.7222,
+      "step": 4942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.581481530242034e-07,
+      "loss": 0.7994,
+      "step": 4943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5744486417875896e-07,
+      "loss": 0.7174,
+      "step": 4944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5674221336064464e-07,
+      "loss": 0.7371,
+      "step": 4945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.560402007791025e-07,
+      "loss": 0.7736,
+      "step": 4946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5533882664318286e-07,
+      "loss": 0.7412,
+      "step": 4947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5463809116174556e-07,
+      "loss": 0.734,
+      "step": 4948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.539379945434629e-07,
+      "loss": 0.6975,
+      "step": 4949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.532385369968136e-07,
+      "loss": 0.7531,
+      "step": 4950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.525397187300894e-07,
+      "loss": 0.7289,
+      "step": 4951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5184153995138887e-07,
+      "loss": 0.7035,
+      "step": 4952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5114400086862203e-07,
+      "loss": 0.6819,
+      "step": 4953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50447101689507e-07,
+      "loss": 0.7477,
+      "step": 4954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.497508426215729e-07,
+      "loss": 0.733,
+      "step": 4955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4905522387215675e-07,
+      "loss": 0.6833,
+      "step": 4956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4836024564840533e-07,
+      "loss": 0.7247,
+      "step": 4957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4766590815727503e-07,
+      "loss": 0.7657,
+      "step": 4958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.469722116055321e-07,
+      "loss": 0.7138,
+      "step": 4959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.462791561997503e-07,
+      "loss": 0.7409,
+      "step": 4960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.455867421463127e-07,
+      "loss": 0.7798,
+      "step": 4961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.448949696514126e-07,
+      "loss": 0.5506,
+      "step": 4962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4420383892105227e-07,
+      "loss": 0.7484,
+      "step": 4963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.435133501610413e-07,
+      "loss": 0.6817,
+      "step": 4964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4282350357699863e-07,
+      "loss": 0.7716,
+      "step": 4965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4213429937435287e-07,
+      "loss": 0.7192,
+      "step": 4966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.414457377583413e-07,
+      "loss": 0.7265,
+      "step": 4967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.407578189340091e-07,
+      "loss": 0.8024,
+      "step": 4968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.400705431062093e-07,
+      "loss": 0.749,
+      "step": 4969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.393839104796051e-07,
+      "loss": 0.7083,
+      "step": 4970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386979212586683e-07,
+      "loss": 0.785,
+      "step": 4971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.380125756476768e-07,
+      "loss": 0.7354,
+      "step": 4972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373278738507199e-07,
+      "loss": 0.7551,
+      "step": 4973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3664381607169255e-07,
+      "loss": 0.7551,
+      "step": 4974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35960402514299e-07,
+      "loss": 0.7371,
+      "step": 4975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.352776333820518e-07,
+      "loss": 0.715,
+      "step": 4976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.345955088782718e-07,
+      "loss": 0.7791,
+      "step": 4977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3391402920608746e-07,
+      "loss": 0.7921,
+      "step": 4978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.332331945684347e-07,
+      "loss": 0.8178,
+      "step": 4979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.325530051680581e-07,
+      "loss": 0.7039,
+      "step": 4980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3187346120751087e-07,
+      "loss": 0.7256,
+      "step": 4981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.311945628891525e-07,
+      "loss": 0.6685,
+      "step": 4982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.305163104151499e-07,
+      "loss": 0.752,
+      "step": 4983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2983870398747953e-07,
+      "loss": 0.6759,
+      "step": 4984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2916174380792474e-07,
+      "loss": 0.803,
+      "step": 4985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2848543007807573e-07,
+      "loss": 0.7097,
+      "step": 4986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2780976299933016e-07,
+      "loss": 0.6704,
+      "step": 4987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.271347427728949e-07,
+      "loss": 0.6689,
+      "step": 4988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.264603695997814e-07,
+      "loss": 0.7829,
+      "step": 4989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.257866436808113e-07,
+      "loss": 0.7914,
+      "step": 4990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.25113565216611e-07,
+      "loss": 0.723,
+      "step": 4991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2444113440761623e-07,
+      "loss": 0.7446,
+      "step": 4992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.237693514540677e-07,
+      "loss": 0.7442,
+      "step": 4993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2309821655601563e-07,
+      "loss": 0.7706,
+      "step": 4994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.224277299133155e-07,
+      "loss": 0.736,
+      "step": 4995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.217578917256295e-07,
+      "loss": 0.7183,
+      "step": 4996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2108870219242807e-07,
+      "loss": 0.6893,
+      "step": 4997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2042016151298815e-07,
+      "loss": 0.7922,
+      "step": 4998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.197522698863931e-07,
+      "loss": 0.8535,
+      "step": 4999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1908502751153203e-07,
+      "loss": 0.8536,
+      "step": 5000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1841843458710273e-07,
+      "loss": 0.7977,
+      "step": 5001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1775249131160895e-07,
+      "loss": 0.7769,
+      "step": 5002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1708719788336006e-07,
+      "loss": 0.7723,
+      "step": 5003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1642255450047203e-07,
+      "loss": 0.7631,
+      "step": 5004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1575856136086826e-07,
+      "loss": 0.7125,
+      "step": 5005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1509521866227825e-07,
+      "loss": 0.739,
+      "step": 5006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1443252660223746e-07,
+      "loss": 0.7032,
+      "step": 5007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.137704853780871e-07,
+      "loss": 0.763,
+      "step": 5008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1310909518697556e-07,
+      "loss": 0.7572,
+      "step": 5009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1244835622585677e-07,
+      "loss": 0.7765,
+      "step": 5010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1178826869149137e-07,
+      "loss": 0.5102,
+      "step": 5011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1112883278044457e-07,
+      "loss": 0.739,
+      "step": 5012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1047004868908964e-07,
+      "loss": 0.7281,
+      "step": 5013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0981191661360375e-07,
+      "loss": 0.6337,
+      "step": 5014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.091544367499713e-07,
+      "loss": 0.7164,
+      "step": 5015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0849760929398147e-07,
+      "loss": 0.739,
+      "step": 5016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.078414344412303e-07,
+      "loss": 0.7634,
+      "step": 5017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.071859123871182e-07,
+      "loss": 0.7377,
+      "step": 5018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0653104332685175e-07,
+      "loss": 0.7767,
+      "step": 5019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0587682745544465e-07,
+      "loss": 0.7524,
+      "step": 5020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.052232649677123e-07,
+      "loss": 0.7464,
+      "step": 5021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0457035605827857e-07,
+      "loss": 0.7631,
+      "step": 5022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0391810092157246e-07,
+      "loss": 0.7268,
+      "step": 5023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0326649975182866e-07,
+      "loss": 0.5113,
+      "step": 5024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.026155527430838e-07,
+      "loss": 0.6884,
+      "step": 5025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.019652600891834e-07,
+      "loss": 0.7699,
+      "step": 5026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0131562198377763e-07,
+      "loss": 0.7385,
+      "step": 5027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0066663862031943e-07,
+      "loss": 0.698,
+      "step": 5028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.000183101920695e-07,
+      "loss": 0.8001,
+      "step": 5029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.993706368920915e-07,
+      "loss": 0.647,
+      "step": 5030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.987236189132556e-07,
+      "loss": 0.6999,
+      "step": 5031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.980772564482351e-07,
+      "loss": 0.4761,
+      "step": 5032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9743154968951e-07,
+      "loss": 0.7323,
+      "step": 5033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.967864988293637e-07,
+      "loss": 0.5469,
+      "step": 5034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.961421040598836e-07,
+      "loss": 0.8168,
+      "step": 5035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.95498365572964e-07,
+      "loss": 0.7625,
+      "step": 5036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9485528356030247e-07,
+      "loss": 0.7318,
+      "step": 5037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9421285821340103e-07,
+      "loss": 0.6605,
+      "step": 5038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9357108972356554e-07,
+      "loss": 0.7555,
+      "step": 5039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9292997828190785e-07,
+      "loss": 0.7759,
+      "step": 5040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.922895240793433e-07,
+      "loss": 0.7309,
+      "step": 5041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9164972730659105e-07,
+      "loss": 0.8009,
+      "step": 5042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.910105881541747e-07,
+      "loss": 0.7968,
+      "step": 5043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9037210681242296e-07,
+      "loss": 0.6987,
+      "step": 5044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.897342834714678e-07,
+      "loss": 0.7553,
+      "step": 5045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8909711832124544e-07,
+      "loss": 0.7536,
+      "step": 5046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8846061155149513e-07,
+      "loss": 0.7224,
+      "step": 5047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.878247633517625e-07,
+      "loss": 0.7706,
+      "step": 5048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8718957391139417e-07,
+      "loss": 0.7854,
+      "step": 5049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.86555043419543e-07,
+      "loss": 0.7189,
+      "step": 5050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8592117206516365e-07,
+      "loss": 0.7742,
+      "step": 5051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.852879600370165e-07,
+      "loss": 0.7284,
+      "step": 5052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.846554075236635e-07,
+      "loss": 0.5407,
+      "step": 5053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8402351471347206e-07,
+      "loss": 0.6034,
+      "step": 5054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8339228179461163e-07,
+      "loss": 0.7993,
+      "step": 5055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.827617089550566e-07,
+      "loss": 0.7148,
+      "step": 5056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8213179638258287e-07,
+      "loss": 0.7531,
+      "step": 5057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.815025442647723e-07,
+      "loss": 0.6689,
+      "step": 5058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.808739527890081e-07,
+      "loss": 0.7284,
+      "step": 5059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.802460221424766e-07,
+      "loss": 0.6487,
+      "step": 5060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.796187525121685e-07,
+      "loss": 0.721,
+      "step": 5061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.789921440848778e-07,
+      "loss": 0.7382,
+      "step": 5062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7836619704720043e-07,
+      "loss": 0.7495,
+      "step": 5063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7774091158553563e-07,
+      "loss": 0.7161,
+      "step": 5064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.771162878860864e-07,
+      "loss": 0.7576,
+      "step": 5065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7649232613485874e-07,
+      "loss": 0.7169,
+      "step": 5066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.758690265176603e-07,
+      "loss": 0.7248,
+      "step": 5067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7524638922010184e-07,
+      "loss": 0.7268,
+      "step": 5068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.746244144275981e-07,
+      "loss": 0.7774,
+      "step": 5069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.740031023253656e-07,
+      "loss": 0.7731,
+      "step": 5070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.733824530984239e-07,
+      "loss": 0.7821,
+      "step": 5071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7276246693159395e-07,
+      "loss": 0.7475,
+      "step": 5072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7214314400950156e-07,
+      "loss": 0.6852,
+      "step": 5073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7152448451657236e-07,
+      "loss": 0.5041,
+      "step": 5074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7090648863703633e-07,
+      "loss": 0.5314,
+      "step": 5075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7028915655492594e-07,
+      "loss": 0.7432,
+      "step": 5076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.696724884540747e-07,
+      "loss": 0.7145,
+      "step": 5077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6905648451811847e-07,
+      "loss": 0.7962,
+      "step": 5078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.684411449304966e-07,
+      "loss": 0.6844,
+      "step": 5079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6782646987445065e-07,
+      "loss": 0.7601,
+      "step": 5080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6721245953302136e-07,
+      "loss": 0.5181,
+      "step": 5081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.66599114089055e-07,
+      "loss": 0.743,
+      "step": 5082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6598643372519826e-07,
+      "loss": 0.739,
+      "step": 5083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6537441862390074e-07,
+      "loss": 0.7049,
+      "step": 5084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.647630689674127e-07,
+      "loss": 0.7556,
+      "step": 5085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6415238493778613e-07,
+      "loss": 0.6255,
+      "step": 5086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6354236671687627e-07,
+      "loss": 0.6874,
+      "step": 5087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.62933014486339e-07,
+      "loss": 0.7877,
+      "step": 5088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.623243284276322e-07,
+      "loss": 0.7656,
+      "step": 5089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.617163087220151e-07,
+      "loss": 0.7282,
+      "step": 5090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6110895555054914e-07,
+      "loss": 0.7573,
+      "step": 5091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6050226909409616e-07,
+      "loss": 0.7229,
+      "step": 5092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.59896249533321e-07,
+      "loss": 0.6958,
+      "step": 5093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.592908970486882e-07,
+      "loss": 0.6965,
+      "step": 5094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.586862118204653e-07,
+      "loss": 0.7876,
+      "step": 5095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5808219402871924e-07,
+      "loss": 0.8158,
+      "step": 5096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.574788438533207e-07,
+      "loss": 0.6658,
+      "step": 5097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5687616147393956e-07,
+      "loss": 0.7467,
+      "step": 5098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5627414707004667e-07,
+      "loss": 0.7894,
+      "step": 5099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.556728008209153e-07,
+      "loss": 0.6776,
+      "step": 5100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5507212290561993e-07,
+      "loss": 0.7581,
+      "step": 5101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.544721135030345e-07,
+      "loss": 0.7518,
+      "step": 5102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.538727727918341e-07,
+      "loss": 0.7526,
+      "step": 5103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.532741009504963e-07,
+      "loss": 0.7805,
+      "step": 5104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5267609815729814e-07,
+      "loss": 0.7558,
+      "step": 5105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.520787645903178e-07,
+      "loss": 0.671,
+      "step": 5106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.514821004274332e-07,
+      "loss": 0.6876,
+      "step": 5107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.508861058463244e-07,
+      "loss": 0.7111,
+      "step": 5108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.502907810244723e-07,
+      "loss": 0.6957,
+      "step": 5109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4969612613915644e-07,
+      "loss": 0.7673,
+      "step": 5110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.49102141367458e-07,
+      "loss": 0.7283,
+      "step": 5111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.485088268862593e-07,
+      "loss": 0.8448,
+      "step": 5112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.479161828722412e-07,
+      "loss": 0.6561,
+      "step": 5113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4732420950188713e-07,
+      "loss": 0.7562,
+      "step": 5114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.467329069514787e-07,
+      "loss": 0.7083,
+      "step": 5115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4614227539709963e-07,
+      "loss": 0.7008,
+      "step": 5116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.455523150146322e-07,
+      "loss": 0.7206,
+      "step": 5117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.449630259797603e-07,
+      "loss": 0.7355,
+      "step": 5118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.44374408467967e-07,
+      "loss": 0.6982,
+      "step": 5119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.437864626545344e-07,
+      "loss": 0.6935,
+      "step": 5120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.431991887145471e-07,
+      "loss": 0.6889,
+      "step": 5121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.426125868228876e-07,
+      "loss": 0.7208,
+      "step": 5122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.420266571542401e-07,
+      "loss": 0.7082,
+      "step": 5123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4144139988308586e-07,
+      "loss": 0.6697,
+      "step": 5124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.408568151837079e-07,
+      "loss": 0.8307,
+      "step": 5125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4027290323018953e-07,
+      "loss": 0.7478,
+      "step": 5126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3968966419641195e-07,
+      "loss": 0.7552,
+      "step": 5127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3910709825605645e-07,
+      "loss": 0.6852,
+      "step": 5128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3852520558260454e-07,
+      "loss": 0.7373,
+      "step": 5129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3794398634933767e-07,
+      "loss": 0.6569,
+      "step": 5130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3736344072933487e-07,
+      "loss": 0.7253,
+      "step": 5131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.367835688954767e-07,
+      "loss": 0.6984,
+      "step": 5132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3620437102044093e-07,
+      "loss": 0.7762,
+      "step": 5133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3562584727670673e-07,
+      "loss": 0.7767,
+      "step": 5134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.350479978365511e-07,
+      "loss": 0.7219,
+      "step": 5135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3447082287205102e-07,
+      "loss": 0.789,
+      "step": 5136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3389432255508215e-07,
+      "loss": 0.6185,
+      "step": 5137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.333184970573188e-07,
+      "loss": 0.7146,
+      "step": 5138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3274334655023545e-07,
+      "loss": 0.7141,
+      "step": 5139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.321688712051054e-07,
+      "loss": 0.8024,
+      "step": 5140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.315950711930004e-07,
+      "loss": 0.7323,
+      "step": 5141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3102194668479073e-07,
+      "loss": 0.692,
+      "step": 5142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3044949785114622e-07,
+      "loss": 0.7134,
+      "step": 5143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2987772486253584e-07,
+      "loss": 0.7216,
+      "step": 5144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2930662788922615e-07,
+      "loss": 0.6405,
+      "step": 5145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.28736207101283e-07,
+      "loss": 0.6612,
+      "step": 5146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.281664626685709e-07,
+      "loss": 0.7396,
+      "step": 5147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2759739476075377e-07,
+      "loss": 0.724,
+      "step": 5148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2702900354729247e-07,
+      "loss": 0.746,
+      "step": 5149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2646128919744687e-07,
+      "loss": 0.7069,
+      "step": 5150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.258942518802762e-07,
+      "loss": 0.7133,
+      "step": 5151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2532789176463675e-07,
+      "loss": 0.617,
+      "step": 5152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.247622090191845e-07,
+      "loss": 0.7755,
+      "step": 5153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.241972038123724e-07,
+      "loss": 0.847,
+      "step": 5154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2363287631245262e-07,
+      "loss": 0.7591,
+      "step": 5155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2306922668747483e-07,
+      "loss": 0.7484,
+      "step": 5156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2250625510528783e-07,
+      "loss": 0.7451,
+      "step": 5157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2194396173353727e-07,
+      "loss": 0.7977,
+      "step": 5158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2138234673966697e-07,
+      "loss": 0.6842,
+      "step": 5159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.208214102909198e-07,
+      "loss": 0.7115,
+      "step": 5160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2026115255433607e-07,
+      "loss": 0.7385,
+      "step": 5161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1970157369675393e-07,
+      "loss": 0.6985,
+      "step": 5162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.191426738848082e-07,
+      "loss": 0.7277,
+      "step": 5163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.185844532849335e-07,
+      "loss": 0.6776,
+      "step": 5164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.180269120633613e-07,
+      "loss": 0.7324,
+      "step": 5165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1747005038612073e-07,
+      "loss": 0.7614,
+      "step": 5166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1691386841903776e-07,
+      "loss": 0.7528,
+      "step": 5167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1635836632773744e-07,
+      "loss": 0.7246,
+      "step": 5168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1580354427764184e-07,
+      "loss": 0.768,
+      "step": 5169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1524940243396992e-07,
+      "loss": 0.7102,
+      "step": 5170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1469594096173828e-07,
+      "loss": 0.7436,
+      "step": 5171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1414316002576125e-07,
+      "loss": 0.7941,
+      "step": 5172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1359105979065126e-07,
+      "loss": 0.5713,
+      "step": 5173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1303964042081626e-07,
+      "loss": 0.7502,
+      "step": 5174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1248890208046213e-07,
+      "loss": 0.7481,
+      "step": 5175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1193884493359322e-07,
+      "loss": 0.7284,
+      "step": 5176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.113894691440091e-07,
+      "loss": 0.723,
+      "step": 5177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1084077487530717e-07,
+      "loss": 0.7468,
+      "step": 5178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1029276229088318e-07,
+      "loss": 0.7498,
+      "step": 5179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0974543155392795e-07,
+      "loss": 0.7821,
+      "step": 5180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0919878282742972e-07,
+      "loss": 0.7856,
+      "step": 5181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0865281627417438e-07,
+      "loss": 0.7538,
+      "step": 5182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0810753205674493e-07,
+      "loss": 0.7648,
+      "step": 5183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0756293033751902e-07,
+      "loss": 0.727,
+      "step": 5184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0701901127867296e-07,
+      "loss": 0.7321,
+      "step": 5185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0647577504218002e-07,
+      "loss": 0.6514,
+      "step": 5186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.059332217898094e-07,
+      "loss": 0.749,
+      "step": 5187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0539135168312685e-07,
+      "loss": 0.8091,
+      "step": 5188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0485016488349414e-07,
+      "loss": 0.7407,
+      "step": 5189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0430966155207122e-07,
+      "loss": 0.6461,
+      "step": 5190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0376984184981242e-07,
+      "loss": 0.6604,
+      "step": 5191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0323070593747107e-07,
+      "loss": 0.6944,
+      "step": 5192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0269225397559377e-07,
+      "loss": 0.7362,
+      "step": 5193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0215448612452643e-07,
+      "loss": 0.7783,
+      "step": 5194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.01617402544409e-07,
+      "loss": 0.6743,
+      "step": 5195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0108100339517934e-07,
+      "loss": 0.6188,
+      "step": 5196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0054528883657025e-07,
+      "loss": 0.6495,
+      "step": 5197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0001025902811084e-07,
+      "loss": 0.744,
+      "step": 5198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9947591412912693e-07,
+      "loss": 0.7006,
+      "step": 5199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9894225429874013e-07,
+      "loss": 0.7735,
+      "step": 5200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.984092796958681e-07,
+      "loss": 0.718,
+      "step": 5201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9787699047922383e-07,
+      "loss": 0.7007,
+      "step": 5202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9734538680731663e-07,
+      "loss": 0.733,
+      "step": 5203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9681446883845267e-07,
+      "loss": 0.7182,
+      "step": 5204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9628423673073217e-07,
+      "loss": 0.7292,
+      "step": 5205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9575469064205183e-07,
+      "loss": 0.718,
+      "step": 5206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9522583073010444e-07,
+      "loss": 0.726,
+      "step": 5207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9469765715237844e-07,
+      "loss": 0.673,
+      "step": 5208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9417017006615746e-07,
+      "loss": 0.7407,
+      "step": 5209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9364336962852004e-07,
+      "loss": 0.673,
+      "step": 5210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9311725599634206e-07,
+      "loss": 0.7183,
+      "step": 5211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.925918293262938e-07,
+      "loss": 0.7293,
+      "step": 5212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9206708977484083e-07,
+      "loss": 0.7714,
+      "step": 5213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9154303749824382e-07,
+      "loss": 0.7396,
+      "step": 5214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.910196726525601e-07,
+      "loss": 0.7697,
+      "step": 5215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9049699539364059e-07,
+      "loss": 0.7533,
+      "step": 5216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8997500587713353e-07,
+      "loss": 0.8156,
+      "step": 5217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8945370425848013e-07,
+      "loss": 0.7669,
+      "step": 5218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889330906929182e-07,
+      "loss": 0.7372,
+      "step": 5219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.884131653354801e-07,
+      "loss": 0.7073,
+      "step": 5220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.878939283409939e-07,
+      "loss": 0.738,
+      "step": 5221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.873753798640815e-07,
+      "loss": 0.8523,
+      "step": 5222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8685752005916046e-07,
+      "loss": 0.7749,
+      "step": 5223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8634034908044357e-07,
+      "loss": 0.7564,
+      "step": 5224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.858238670819379e-07,
+      "loss": 0.6865,
+      "step": 5225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.853080742174468e-07,
+      "loss": 0.7506,
+      "step": 5226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8479297064056518e-07,
+      "loss": 0.7497,
+      "step": 5227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.842785565046859e-07,
+      "loss": 0.8298,
+      "step": 5228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.837648319629956e-07,
+      "loss": 0.6863,
+      "step": 5229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8325179716847523e-07,
+      "loss": 0.7338,
+      "step": 5230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8273945227389954e-07,
+      "loss": 0.7394,
+      "step": 5231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8222779743183927e-07,
+      "loss": 0.7546,
+      "step": 5232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8171683279465946e-07,
+      "loss": 0.7774,
+      "step": 5233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8120655851451868e-07,
+      "loss": 0.8148,
+      "step": 5234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8069697474337123e-07,
+      "loss": 0.7486,
+      "step": 5235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8018808163296485e-07,
+      "loss": 0.7399,
+      "step": 5236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7967987933484117e-07,
+      "loss": 0.6923,
+      "step": 5237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7917236800033712e-07,
+      "loss": 0.696,
+      "step": 5238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7866554778058402e-07,
+      "loss": 0.627,
+      "step": 5239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7815941882650651e-07,
+      "loss": 0.7913,
+      "step": 5240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7765398128882338e-07,
+      "loss": 0.7095,
+      "step": 5241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7714923531804817e-07,
+      "loss": 0.6847,
+      "step": 5242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.766451810644887e-07,
+      "loss": 0.7628,
+      "step": 5243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7614181867824598e-07,
+      "loss": 0.8113,
+      "step": 5244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7563914830921487e-07,
+      "loss": 0.7017,
+      "step": 5245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7513717010708476e-07,
+      "loss": 0.7733,
+      "step": 5246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7463588422133943e-07,
+      "loss": 0.7786,
+      "step": 5247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.741352908012553e-07,
+      "loss": 0.7699,
+      "step": 5248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7363538999590284e-07,
+      "loss": 0.7285,
+      "step": 5249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.731361819541466e-07,
+      "loss": 0.4987,
+      "step": 5250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7263766682464573e-07,
+      "loss": 0.518,
+      "step": 5251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7213984475585143e-07,
+      "loss": 0.7655,
+      "step": 5252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.716427158960085e-07,
+      "loss": 0.5115,
+      "step": 5253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7114628039315712e-07,
+      "loss": 0.7062,
+      "step": 5254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7065053839512901e-07,
+      "loss": 0.7442,
+      "step": 5255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7015549004955085e-07,
+      "loss": 0.6391,
+      "step": 5256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6966113550384134e-07,
+      "loss": 0.8051,
+      "step": 5257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.691674749052144e-07,
+      "loss": 0.7048,
+      "step": 5258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6867450840067523e-07,
+      "loss": 0.7488,
+      "step": 5259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6818223613702418e-07,
+      "loss": 0.7643,
+      "step": 5260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.676906582608537e-07,
+      "loss": 0.8041,
+      "step": 5261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6719977491854944e-07,
+      "loss": 0.6966,
+      "step": 5262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.667095862562909e-07,
+      "loss": 0.7274,
+      "step": 5263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6622009242005106e-07,
+      "loss": 0.776,
+      "step": 5264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6573129355559463e-07,
+      "loss": 0.8302,
+      "step": 5265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524318980847997e-07,
+      "loss": 0.706,
+      "step": 5266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475578132405855e-07,
+      "loss": 0.7381,
+      "step": 5267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.642690682474757e-07,
+      "loss": 0.7936,
+      "step": 5268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.637830507236679e-07,
+      "loss": 0.7846,
+      "step": 5269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.63297728897365e-07,
+      "loss": 0.7458,
+      "step": 5270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.628131029130908e-07,
+      "loss": 0.7374,
+      "step": 5271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.623291729151613e-07,
+      "loss": 0.6931,
+      "step": 5272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6184593904768454e-07,
+      "loss": 0.7579,
+      "step": 5273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6136340145456175e-07,
+      "loss": 0.7722,
+      "step": 5274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6088156027948748e-07,
+      "loss": 0.7394,
+      "step": 5275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.604004156659472e-07,
+      "loss": 0.8005,
+      "step": 5276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5991996775722136e-07,
+      "loss": 0.7794,
+      "step": 5277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5944021669638043e-07,
+      "loss": 0.7905,
+      "step": 5278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5896116262628935e-07,
+      "loss": 0.6919,
+      "step": 5279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.584828056896043e-07,
+      "loss": 0.7581,
+      "step": 5280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5800514602877404e-07,
+      "loss": 0.741,
+      "step": 5281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5752818378604124e-07,
+      "loss": 0.7352,
+      "step": 5282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.570519191034378e-07,
+      "loss": 0.7164,
+      "step": 5283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.565763521227903e-07,
+      "loss": 0.7347,
+      "step": 5284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5610148298571715e-07,
+      "loss": 0.7681,
+      "step": 5285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5562731183362906e-07,
+      "loss": 0.7083,
+      "step": 5286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5515383880772756e-07,
+      "loss": 0.5271,
+      "step": 5287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5468106404900763e-07,
+      "loss": 0.7409,
+      "step": 5288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.542089876982561e-07,
+      "loss": 0.7015,
+      "step": 5289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5373760989605218e-07,
+      "loss": 0.732,
+      "step": 5290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5326693078276583e-07,
+      "loss": 0.7093,
+      "step": 5291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5279695049855964e-07,
+      "loss": 0.7199,
+      "step": 5292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.523276691833886e-07,
+      "loss": 0.753,
+      "step": 5293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.518590869769984e-07,
+      "loss": 0.7364,
+      "step": 5294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5139120401892804e-07,
+      "loss": 0.7744,
+      "step": 5295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5092402044850657e-07,
+      "loss": 0.7499,
+      "step": 5296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.504575364048569e-07,
+      "loss": 0.7431,
+      "step": 5297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4999175202689092e-07,
+      "loss": 0.7434,
+      "step": 5298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4952666745331522e-07,
+      "loss": 0.7293,
+      "step": 5299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.490622828226254e-07,
+      "loss": 0.705,
+      "step": 5300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4859859827310946e-07,
+      "loss": 0.708,
+      "step": 5301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4813561394284775e-07,
+      "loss": 0.7619,
+      "step": 5302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.476733299697114e-07,
+      "loss": 0.8097,
+      "step": 5303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4721174649136332e-07,
+      "loss": 0.7815,
+      "step": 5304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4675086364525638e-07,
+      "loss": 0.6582,
+      "step": 5305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4629068156863712e-07,
+      "loss": 0.7208,
+      "step": 5306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.458312003985421e-07,
+      "loss": 0.7568,
+      "step": 5307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4537242027179932e-07,
+      "loss": 0.7488,
+      "step": 5308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4491434132502728e-07,
+      "loss": 0.6012,
+      "step": 5309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.444569636946372e-07,
+      "loss": 0.764,
+      "step": 5310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.440002875168306e-07,
+      "loss": 0.7287,
+      "step": 5311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4354431292760013e-07,
+      "loss": 0.801,
+      "step": 5312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4308904006272912e-07,
+      "loss": 0.6755,
+      "step": 5313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4263446905779326e-07,
+      "loss": 0.8065,
+      "step": 5314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4218060004815763e-07,
+      "loss": 0.8041,
+      "step": 5315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.417274331689794e-07,
+      "loss": 0.7675,
+      "step": 5316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.412749685552059e-07,
+      "loss": 0.7933,
+      "step": 5317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4082320634157625e-07,
+      "loss": 0.7497,
+      "step": 5318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.40372146662619e-07,
+      "loss": 0.7051,
+      "step": 5319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.399217896526553e-07,
+      "loss": 0.7571,
+      "step": 5320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3947213544579535e-07,
+      "loss": 0.7577,
+      "step": 5321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3902318417594096e-07,
+      "loss": 0.7665,
+      "step": 5322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3857493597678461e-07,
+      "loss": 0.8154,
+      "step": 5323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3812739098180956e-07,
+      "loss": 0.7135,
+      "step": 5324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3768054932428887e-07,
+      "loss": 0.7618,
+      "step": 5325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3723441113728668e-07,
+      "loss": 0.6642,
+      "step": 5326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.367889765536576e-07,
+      "loss": 0.6922,
+      "step": 5327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.363442457060468e-07,
+      "loss": 0.7232,
+      "step": 5328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3590021872689042e-07,
+      "loss": 0.7914,
+      "step": 5329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3545689574841341e-07,
+      "loss": 0.6558,
+      "step": 5330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3501427690263247e-07,
+      "loss": 0.8227,
+      "step": 5331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3457236232135462e-07,
+      "loss": 0.6188,
+      "step": 5332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.341311521361763e-07,
+      "loss": 0.6725,
+      "step": 5333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.336906464784843e-07,
+      "loss": 0.7599,
+      "step": 5334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3325084547945622e-07,
+      "loss": 0.7089,
+      "step": 5335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3281174927006007e-07,
+      "loss": 0.8553,
+      "step": 5336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3237335798105266e-07,
+      "loss": 0.5094,
+      "step": 5337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.319356717429826e-07,
+      "loss": 0.7319,
+      "step": 5338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3149869068618705e-07,
+      "loss": 0.7545,
+      "step": 5339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3106241494079325e-07,
+      "loss": 0.7902,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3062684463671981e-07,
+      "loss": 0.7339,
+      "step": 5341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.301919799036741e-07,
+      "loss": 0.672,
+      "step": 5342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2975782087115335e-07,
+      "loss": 0.7604,
+      "step": 5343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2932436766844498e-07,
+      "loss": 0.6847,
+      "step": 5344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2889162042462655e-07,
+      "loss": 0.804,
+      "step": 5345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2845957926856472e-07,
+      "loss": 0.7453,
+      "step": 5346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2802824432891658e-07,
+      "loss": 0.7025,
+      "step": 5347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.275976157341277e-07,
+      "loss": 0.7806,
+      "step": 5348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.271676936124347e-07,
+      "loss": 0.7421,
+      "step": 5349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2673847809186347e-07,
+      "loss": 0.7106,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.26309969300229e-07,
+      "loss": 0.8206,
+      "step": 5351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2588216736513588e-07,
+      "loss": 0.6241,
+      "step": 5352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2545507241397886e-07,
+      "loss": 0.7508,
+      "step": 5353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2502868457394097e-07,
+      "loss": 0.7272,
+      "step": 5354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2460300397199644e-07,
+      "loss": 0.5719,
+      "step": 5355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2417803073490693e-07,
+      "loss": 0.7359,
+      "step": 5356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2375376498922538e-07,
+      "loss": 0.5296,
+      "step": 5357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2333020686129183e-07,
+      "loss": 0.6793,
+      "step": 5358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.229073564772379e-07,
+      "loss": 0.6911,
+      "step": 5359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2248521396298313e-07,
+      "loss": 0.7537,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2206377944423587e-07,
+      "loss": 0.8112,
+      "step": 5361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2164305304649486e-07,
+      "loss": 0.7441,
+      "step": 5362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2122303489504772e-07,
+      "loss": 0.7641,
+      "step": 5363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2080372511497017e-07,
+      "loss": 0.7853,
+      "step": 5364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2038512383112794e-07,
+      "loss": 0.789,
+      "step": 5365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1996723116817516e-07,
+      "loss": 0.6866,
+      "step": 5366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1955004725055618e-07,
+      "loss": 0.6862,
+      "step": 5367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1913357220250277e-07,
+      "loss": 0.7478,
+      "step": 5368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1871780614803569e-07,
+      "loss": 0.6327,
+      "step": 5369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.183027492109659e-07,
+      "loss": 0.7827,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1788840151489234e-07,
+      "loss": 0.7532,
+      "step": 5371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1747476318320266e-07,
+      "loss": 0.7093,
+      "step": 5372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1706183433907309e-07,
+      "loss": 0.4907,
+      "step": 5373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1664961510546912e-07,
+      "loss": 0.7455,
+      "step": 5374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1623810560514537e-07,
+      "loss": 0.8101,
+      "step": 5375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1582730596064379e-07,
+      "loss": 0.7785,
+      "step": 5376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1541721629429515e-07,
+      "loss": 0.7371,
+      "step": 5377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1500783672822036e-07,
+      "loss": 0.7501,
+      "step": 5378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1459916738432691e-07,
+      "loss": 0.6456,
+      "step": 5379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.141912083843122e-07,
+      "loss": 0.7655,
+      "step": 5380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1378395984966095e-07,
+      "loss": 0.7629,
+      "step": 5381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1337742190164758e-07,
+      "loss": 0.6425,
+      "step": 5382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1297159466133356e-07,
+      "loss": 0.6627,
+      "step": 5383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1256647824957002e-07,
+      "loss": 0.6088,
+      "step": 5384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1216207278699599e-07,
+      "loss": 0.7765,
+      "step": 5385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1175837839403735e-07,
+      "loss": 0.68,
+      "step": 5386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1135539519091021e-07,
+      "loss": 0.6377,
+      "step": 5387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1095312329761826e-07,
+      "loss": 0.779,
+      "step": 5388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1055156283395401e-07,
+      "loss": 0.8112,
+      "step": 5389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1015071391949573e-07,
+      "loss": 0.7858,
+      "step": 5390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0975057667361233e-07,
+      "loss": 0.7451,
+      "step": 5391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0935115121545992e-07,
+      "loss": 0.6729,
+      "step": 5392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.089524376639825e-07,
+      "loss": 0.7547,
+      "step": 5393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0855443613791261e-07,
+      "loss": 0.705,
+      "step": 5394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0815714675576932e-07,
+      "loss": 0.695,
+      "step": 5395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.077605696358619e-07,
+      "loss": 0.6586,
+      "step": 5396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0736470489628531e-07,
+      "loss": 0.7406,
+      "step": 5397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0696955265492414e-07,
+      "loss": 0.753,
+      "step": 5398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0657511302944956e-07,
+      "loss": 0.6383,
+      "step": 5399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0618138613732065e-07,
+      "loss": 0.7543,
+      "step": 5400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0578837209578502e-07,
+      "loss": 0.6929,
+      "step": 5401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0539607102187765e-07,
+      "loss": 0.754,
+      "step": 5402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0500448303242123e-07,
+      "loss": 0.7417,
+      "step": 5403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0461360824402522e-07,
+      "loss": 0.52,
+      "step": 5404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0422344677308793e-07,
+      "loss": 0.6892,
+      "step": 5405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0383399873579503e-07,
+      "loss": 0.7557,
+      "step": 5406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0344526424811956e-07,
+      "loss": 0.7687,
+      "step": 5407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0305724342582119e-07,
+      "loss": 0.7999,
+      "step": 5408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0266993638444856e-07,
+      "loss": 0.6774,
+      "step": 5409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.022833432393372e-07,
+      "loss": 0.7549,
+      "step": 5410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0189746410560974e-07,
+      "loss": 0.8038,
+      "step": 5411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0151229909817567e-07,
+      "loss": 0.8326,
+      "step": 5412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.011278483317335e-07,
+      "loss": 0.8112,
+      "step": 5413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0074411192076805e-07,
+      "loss": 0.7257,
+      "step": 5414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0036108997955096e-07,
+      "loss": 0.7482,
+      "step": 5415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.997878262214156e-08,
+      "loss": 0.7445,
+      "step": 5416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.959718996238682e-08,
+      "loss": 0.6823,
+      "step": 5417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.921631211392029e-08,
+      "loss": 0.7874,
+      "step": 5418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.883614919016293e-08,
+      "loss": 0.6986,
+      "step": 5419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.845670130432222e-08,
+      "loss": 0.7814,
+      "step": 5420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.80779685693939e-08,
+      "loss": 0.7488,
+      "step": 5421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.769995109815966e-08,
+      "loss": 0.6841,
+      "step": 5422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.732264900318866e-08,
+      "loss": 0.6961,
+      "step": 5423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.69460623968374e-08,
+      "loss": 0.6794,
+      "step": 5424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.657019139124784e-08,
+      "loss": 0.675,
+      "step": 5425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.619503609835074e-08,
+      "loss": 0.693,
+      "step": 5426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.582059662986282e-08,
+      "loss": 0.7718,
+      "step": 5427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.54468730972874e-08,
+      "loss": 0.785,
+      "step": 5428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.507386561191462e-08,
+      "loss": 0.7141,
+      "step": 5429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.470157428482229e-08,
+      "loss": 0.7042,
+      "step": 5430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.432999922687397e-08,
+      "loss": 0.8161,
+      "step": 5431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.395914054872058e-08,
+      "loss": 0.7869,
+      "step": 5432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.358899836079904e-08,
+      "loss": 0.8554,
+      "step": 5433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.321957277333316e-08,
+      "loss": 0.7278,
+      "step": 5434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.285086389633379e-08,
+      "loss": 0.7221,
+      "step": 5435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.248287183959842e-08,
+      "loss": 0.7093,
+      "step": 5436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.211559671270965e-08,
+      "loss": 0.7091,
+      "step": 5437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.17490386250383e-08,
+      "loss": 0.8163,
+      "step": 5438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.1383197685741e-08,
+      "loss": 0.6308,
+      "step": 5439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.101807400376028e-08,
+      "loss": 0.7334,
+      "step": 5440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.065366768782668e-08,
+      "loss": 0.7092,
+      "step": 5441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.028997884645535e-08,
+      "loss": 0.7809,
+      "step": 5442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.992700758794798e-08,
+      "loss": 0.7455,
+      "step": 5443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.956475402039394e-08,
+      "loss": 0.7622,
+      "step": 5444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.920321825166777e-08,
+      "loss": 0.7369,
+      "step": 5445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.884240038943031e-08,
+      "loss": 0.5465,
+      "step": 5446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.848230054112894e-08,
+      "loss": 0.7689,
+      "step": 5447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.812291881399704e-08,
+      "loss": 0.6905,
+      "step": 5448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.776425531505428e-08,
+      "loss": 0.7636,
+      "step": 5449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.740631015110662e-08,
+      "loss": 0.8267,
+      "step": 5450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.704908342874518e-08,
+      "loss": 0.6342,
+      "step": 5451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.66925752543482e-08,
+      "loss": 0.7019,
+      "step": 5452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.633678573407994e-08,
+      "loss": 0.6097,
+      "step": 5453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.598171497388951e-08,
+      "loss": 0.712,
+      "step": 5454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.562736307951319e-08,
+      "loss": 0.6988,
+      "step": 5455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.527373015647294e-08,
+      "loss": 0.5052,
+      "step": 5456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.492081631007564e-08,
+      "loss": 0.6747,
+      "step": 5457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.456862164541585e-08,
+      "loss": 0.7126,
+      "step": 5458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.421714626737187e-08,
+      "loss": 0.7654,
+      "step": 5459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.386639028061e-08,
+      "loss": 0.6734,
+      "step": 5460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.351635378958e-08,
+      "loss": 0.7434,
+      "step": 5461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.316703689851962e-08,
+      "loss": 0.7162,
+      "step": 5462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.281843971145092e-08,
+      "loss": 0.8053,
+      "step": 5463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.247056233218142e-08,
+      "loss": 0.7379,
+      "step": 5464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.21234048643052e-08,
+      "loss": 0.7913,
+      "step": 5465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.177696741120206e-08,
+      "loss": 0.7406,
+      "step": 5466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.143125007603642e-08,
+      "loss": 0.6928,
+      "step": 5467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.10862529617587e-08,
+      "loss": 0.554,
+      "step": 5468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.074197617110508e-08,
+      "loss": 0.7217,
+      "step": 5469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.039841980659769e-08,
+      "loss": 0.6973,
+      "step": 5470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.005558397054275e-08,
+      "loss": 0.7255,
+      "step": 5471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.971346876503249e-08,
+      "loss": 0.7145,
+      "step": 5472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.937207429194544e-08,
+      "loss": 0.6705,
+      "step": 5473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.90314006529444e-08,
+      "loss": 0.748,
+      "step": 5474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.869144794947825e-08,
+      "loss": 0.7194,
+      "step": 5475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.835221628278016e-08,
+      "loss": 0.8048,
+      "step": 5476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.801370575386991e-08,
+      "loss": 0.7439,
+      "step": 5477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.767591646355155e-08,
+      "loss": 0.7002,
+      "step": 5478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.733884851241518e-08,
+      "loss": 0.5425,
+      "step": 5479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.70025020008347e-08,
+      "loss": 0.742,
+      "step": 5480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.666687702897108e-08,
+      "loss": 0.6693,
+      "step": 5481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.633197369676882e-08,
+      "loss": 0.7128,
+      "step": 5482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.599779210395841e-08,
+      "loss": 0.7229,
+      "step": 5483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.566433235005472e-08,
+      "loss": 0.5128,
+      "step": 5484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.533159453435885e-08,
+      "loss": 0.6745,
+      "step": 5485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.499957875595515e-08,
+      "loss": 0.8234,
+      "step": 5486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.466828511371455e-08,
+      "loss": 0.7741,
+      "step": 5487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.433771370629311e-08,
+      "loss": 0.6683,
+      "step": 5488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.400786463212934e-08,
+      "loss": 0.6446,
+      "step": 5489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.367873798944935e-08,
+      "loss": 0.5889,
+      "step": 5490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.335033387626311e-08,
+      "loss": 0.7722,
+      "step": 5491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.302265239036571e-08,
+      "loss": 0.6367,
+      "step": 5492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.269569362933577e-08,
+      "loss": 0.6821,
+      "step": 5493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.236945769053843e-08,
+      "loss": 0.7504,
+      "step": 5494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.204394467112325e-08,
+      "loss": 0.7303,
+      "step": 5495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.17191546680232e-08,
+      "loss": 0.7643,
+      "step": 5496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.139508777795762e-08,
+      "loss": 0.6937,
+      "step": 5497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.107174409742901e-08,
+      "loss": 0.6835,
+      "step": 5498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.07491237227262e-08,
+      "loss": 0.6725,
+      "step": 5499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.042722674992092e-08,
+      "loss": 0.7245,
+      "step": 5500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.010605327487041e-08,
+      "loss": 0.6833,
+      "step": 5501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.978560339321677e-08,
+      "loss": 0.8033,
+      "step": 5502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.946587720038534e-08,
+      "loss": 0.752,
+      "step": 5503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.914687479158716e-08,
+      "loss": 0.7173,
+      "step": 5504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.882859626181765e-08,
+      "loss": 0.8276,
+      "step": 5505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.851104170585598e-08,
+      "loss": 0.6804,
+      "step": 5506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.819421121826597e-08,
+      "loss": 0.6942,
+      "step": 5507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.78781048933963e-08,
+      "loss": 0.7438,
+      "step": 5508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.756272282537945e-08,
+      "loss": 0.6968,
+      "step": 5509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.724806510813281e-08,
+      "loss": 0.7505,
+      "step": 5510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.693413183535724e-08,
+      "loss": 0.7089,
+      "step": 5511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.662092310053825e-08,
+      "loss": 0.7741,
+      "step": 5512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.630843899694623e-08,
+      "loss": 0.7439,
+      "step": 5513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.599667961763478e-08,
+      "loss": 0.7401,
+      "step": 5514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.56856450554419e-08,
+      "loss": 0.5232,
+      "step": 5515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.537533540299068e-08,
+      "loss": 0.7734,
+      "step": 5516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.506575075268695e-08,
+      "loss": 0.7813,
+      "step": 5517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.475689119672169e-08,
+      "loss": 0.7415,
+      "step": 5518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.444875682706936e-08,
+      "loss": 0.7443,
+      "step": 5519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.414134773548908e-08,
+      "loss": 0.7692,
+      "step": 5520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.383466401352317e-08,
+      "loss": 0.7824,
+      "step": 5521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.352870575249887e-08,
+      "loss": 0.7068,
+      "step": 5522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.322347304352605e-08,
+      "loss": 0.7857,
+      "step": 5523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.291896597750036e-08,
+      "loss": 0.6933,
+      "step": 5524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.26151846450998e-08,
+      "loss": 0.7585,
+      "step": 5525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.231212913678674e-08,
+      "loss": 0.7403,
+      "step": 5526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.200979954280817e-08,
+      "loss": 0.6864,
+      "step": 5527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.170819595319317e-08,
+      "loss": 0.5442,
+      "step": 5528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.140731845775628e-08,
+      "loss": 0.6329,
+      "step": 5529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.110716714609554e-08,
+      "loss": 0.6956,
+      "step": 5530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.080774210759227e-08,
+      "loss": 0.6833,
+      "step": 5531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.050904343141095e-08,
+      "loss": 0.6779,
+      "step": 5532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.021107120650127e-08,
+      "loss": 0.7591,
+      "step": 5533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.991382552159559e-08,
+      "loss": 0.7612,
+      "step": 5534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.96173064652103e-08,
+      "loss": 0.694,
+      "step": 5535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.932151412564452e-08,
+      "loss": 0.7021,
+      "step": 5536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9026448590982497e-08,
+      "loss": 0.713,
+      "step": 5537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.87321099490909e-08,
+      "loss": 0.7791,
+      "step": 5538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.843849828762016e-08,
+      "loss": 0.7368,
+      "step": 5539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.814561369400451e-08,
+      "loss": 0.7175,
+      "step": 5540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7853456255461406e-08,
+      "loss": 0.7063,
+      "step": 5541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7562026058991806e-08,
+      "loss": 0.7253,
+      "step": 5542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.72713231913799e-08,
+      "loss": 0.7698,
+      "step": 5543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6981347739194225e-08,
+      "loss": 0.6921,
+      "step": 5544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.669209978878543e-08,
+      "loss": 0.7484,
+      "step": 5545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.640357942628821e-08,
+      "loss": 0.74,
+      "step": 5546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6115786737620526e-08,
+      "loss": 0.7725,
+      "step": 5547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5828721808484084e-08,
+      "loss": 0.703,
+      "step": 5548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.554238472436219e-08,
+      "loss": 0.7477,
+      "step": 5549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5256775570523556e-08,
+      "loss": 0.711,
+      "step": 5550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.497189443201906e-08,
+      "loss": 0.6157,
+      "step": 5551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.468774139368277e-08,
+      "loss": 0.5716,
+      "step": 5552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.440431654013229e-08,
+      "loss": 0.7244,
+      "step": 5553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.412161995576759e-08,
+      "loss": 0.702,
+      "step": 5554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.3839651724773015e-08,
+      "loss": 0.7304,
+      "step": 5555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.3558411931114716e-08,
+      "loss": 0.7075,
+      "step": 5556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.327790065854321e-08,
+      "loss": 0.7303,
+      "step": 5557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.299811799059085e-08,
+      "loss": 0.7386,
+      "step": 5558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2719064010574026e-08,
+      "loss": 0.7822,
+      "step": 5559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.244073880159101e-08,
+      "loss": 0.7568,
+      "step": 5560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216314244652437e-08,
+      "loss": 0.6753,
+      "step": 5561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1886275028038555e-08,
+      "loss": 0.6772,
+      "step": 5562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.161013662858205e-08,
+      "loss": 0.816,
+      "step": 5563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.133472733038464e-08,
+      "loss": 0.7483,
+      "step": 5564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1060047215460704e-08,
+      "loss": 0.8041,
+      "step": 5565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.078609636560622e-08,
+      "loss": 0.7705,
+      "step": 5566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.051287486240064e-08,
+      "loss": 0.7085,
+      "step": 5567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.024038278720611e-08,
+      "loss": 0.7734,
+      "step": 5568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.996862022116744e-08,
+      "loss": 0.7539,
+      "step": 5569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9697587245212395e-08,
+      "loss": 0.6958,
+      "step": 5570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.942728394005114e-08,
+      "loss": 0.731,
+      "step": 5571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.915771038617678e-08,
+      "loss": 0.7361,
+      "step": 5572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8888866663865387e-08,
+      "loss": 0.5342,
+      "step": 5573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8620752853175415e-08,
+      "loss": 0.7592,
+      "step": 5574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8353369033947715e-08,
+      "loss": 0.6586,
+      "step": 5575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.808671528580583e-08,
+      "loss": 0.7415,
+      "step": 5576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.782079168815651e-08,
+      "loss": 0.7477,
+      "step": 5577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.755559832018836e-08,
+      "loss": 0.6259,
+      "step": 5578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7291135260872645e-08,
+      "loss": 0.5865,
+      "step": 5579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7027402588963876e-08,
+      "loss": 0.6785,
+      "step": 5580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676440038299784e-08,
+      "loss": 0.6579,
+      "step": 5581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.650212872129412e-08,
+      "loss": 0.7761,
+      "step": 5582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6240587681953274e-08,
+      "loss": 0.7469,
+      "step": 5583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.597977734285969e-08,
+      "loss": 0.7043,
+      "step": 5584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.571969778167928e-08,
+      "loss": 0.7192,
+      "step": 5585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5460349075860946e-08,
+      "loss": 0.8493,
+      "step": 5586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.520173130263567e-08,
+      "loss": 0.7015,
+      "step": 5587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.494384453901601e-08,
+      "loss": 0.8167,
+      "step": 5588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4686688861798334e-08,
+      "loss": 0.7196,
+      "step": 5589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4430264347560256e-08,
+      "loss": 0.7056,
+      "step": 5590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4174571072662095e-08,
+      "loss": 0.6833,
+      "step": 5591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3919609113246e-08,
+      "loss": 0.702,
+      "step": 5592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.36653785452365e-08,
+      "loss": 0.7915,
+      "step": 5593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3411879444340835e-08,
+      "loss": 0.7107,
+      "step": 5594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.315911188604777e-08,
+      "loss": 0.7641,
+      "step": 5595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2907075945628485e-08,
+      "loss": 0.6492,
+      "step": 5596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2655771698136286e-08,
+      "loss": 0.803,
+      "step": 5597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.240519921840658e-08,
+      "loss": 0.7449,
+      "step": 5598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2155358581056903e-08,
+      "loss": 0.7439,
+      "step": 5599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.190624986048664e-08,
+      "loss": 0.7954,
+      "step": 5600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1657873130877546e-08,
+      "loss": 0.823,
+      "step": 5601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.141022846619325e-08,
+      "loss": 0.6778,
+      "step": 5602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1163315940179474e-08,
+      "loss": 0.6263,
+      "step": 5603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0917135626363515e-08,
+      "loss": 0.7521,
+      "step": 5604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.067168759805534e-08,
+      "loss": 0.6817,
+      "step": 5605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0426971928345936e-08,
+      "loss": 0.7122,
+      "step": 5606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.018298869010895e-08,
+      "loss": 0.826,
+      "step": 5607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.993973795599987e-08,
+      "loss": 0.7727,
+      "step": 5608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9697219798455755e-08,
+      "loss": 0.7412,
+      "step": 5609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9455434289695225e-08,
+      "loss": 0.7598,
+      "step": 5610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.921438150171958e-08,
+      "loss": 0.6948,
+      "step": 5611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.897406150631139e-08,
+      "loss": 0.6613,
+      "step": 5612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.873447437503508e-08,
+      "loss": 0.7357,
+      "step": 5613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.849562017923664e-08,
+      "loss": 0.7424,
+      "step": 5614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.825749899004416e-08,
+      "loss": 0.7459,
+      "step": 5615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8020110878367324e-08,
+      "loss": 0.6846,
+      "step": 5616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.778345591489735e-08,
+      "loss": 0.6775,
+      "step": 5617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7547534170107314e-08,
+      "loss": 0.7733,
+      "step": 5618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7312345714252126e-08,
+      "loss": 0.6915,
+      "step": 5619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.707789061736744e-08,
+      "loss": 0.7943,
+      "step": 5620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.684416894927184e-08,
+      "loss": 0.7689,
+      "step": 5621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6611180779564656e-08,
+      "loss": 0.7345,
+      "step": 5622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.637892617762678e-08,
+      "loss": 0.7881,
+      "step": 5623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6147405212620946e-08,
+      "loss": 0.7484,
+      "step": 5624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.591661795349172e-08,
+      "loss": 0.6283,
+      "step": 5625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5686564468964404e-08,
+      "loss": 0.7266,
+      "step": 5626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.545724482754614e-08,
+      "loss": 0.7099,
+      "step": 5627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.522865909752565e-08,
+      "loss": 0.7616,
+      "step": 5628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5000807346973195e-08,
+      "loss": 0.7427,
+      "step": 5629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4773689643740336e-08,
+      "loss": 0.7441,
+      "step": 5630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.454730605545964e-08,
+      "loss": 0.773,
+      "step": 5631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432165664954551e-08,
+      "loss": 0.7441,
+      "step": 5632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4096741493194196e-08,
+      "loss": 0.7084,
+      "step": 5633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3872560653382125e-08,
+      "loss": 0.7321,
+      "step": 5634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.364911419686784e-08,
+      "loss": 0.6531,
+      "step": 5635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3426402190191163e-08,
+      "loss": 0.6887,
+      "step": 5636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3204424699672945e-08,
+      "loss": 0.656,
+      "step": 5637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.298318179141557e-08,
+      "loss": 0.714,
+      "step": 5638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.276267353130219e-08,
+      "loss": 0.7159,
+      "step": 5639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2542899984998044e-08,
+      "loss": 0.783,
+      "step": 5640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.232386121794856e-08,
+      "loss": 0.7182,
+      "step": 5641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.210555729538156e-08,
+      "loss": 0.8126,
+      "step": 5642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.188798828230449e-08,
+      "loss": 0.7943,
+      "step": 5643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.16711542435072e-08,
+      "loss": 0.6748,
+      "step": 5644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.145505524356024e-08,
+      "loss": 0.7937,
+      "step": 5645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.123969134681548e-08,
+      "loss": 0.7608,
+      "step": 5646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.102506261740579e-08,
+      "loss": 0.7086,
+      "step": 5647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.081116911924476e-08,
+      "loss": 0.7785,
+      "step": 5648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0598010916027265e-08,
+      "loss": 0.7167,
+      "step": 5649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.038558807122921e-08,
+      "loss": 0.7336,
+      "step": 5650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.017390064810832e-08,
+      "loss": 0.697,
+      "step": 5651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.996294870970168e-08,
+      "loss": 0.6537,
+      "step": 5652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.975273231882875e-08,
+      "loss": 0.8293,
+      "step": 5653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.954325153808918e-08,
+      "loss": 0.6983,
+      "step": 5654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9334506429864184e-08,
+      "loss": 0.6845,
+      "step": 5655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.912649705631515e-08,
+      "loss": 0.7479,
+      "step": 5656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8919223479385305e-08,
+      "loss": 0.7675,
+      "step": 5657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8712685760797776e-08,
+      "loss": 0.7223,
+      "step": 5658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8506883962056975e-08,
+      "loss": 0.6722,
+      "step": 5659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830181814444888e-08,
+      "loss": 0.7491,
+      "step": 5660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8097488369038805e-08,
+      "loss": 0.668,
+      "step": 5661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7893894696674472e-08,
+      "loss": 0.5669,
+      "step": 5662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7691037187983218e-08,
+      "loss": 0.8105,
+      "step": 5663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.748891590337366e-08,
+      "loss": 0.7762,
+      "step": 5664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7287530903035154e-08,
+      "loss": 0.728,
+      "step": 5665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7086882246937505e-08,
+      "loss": 0.6656,
+      "step": 5666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6886969994831536e-08,
+      "loss": 0.6855,
+      "step": 5667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6687794206249063e-08,
+      "loss": 0.6997,
+      "step": 5668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6489354940502365e-08,
+      "loss": 0.7721,
+      "step": 5669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.629165225668362e-08,
+      "loss": 0.7117,
+      "step": 5670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6094686213666565e-08,
+      "loss": 0.7022,
+      "step": 5671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5898456870105947e-08,
+      "loss": 0.6847,
+      "step": 5672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.570296428443586e-08,
+      "loss": 0.7522,
+      "step": 5673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.550820851487168e-08,
+      "loss": 0.7575,
+      "step": 5674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5314189619409514e-08,
+      "loss": 0.7387,
+      "step": 5675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.512090765582592e-08,
+      "loss": 0.7035,
+      "step": 5676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.492836268167792e-08,
+      "loss": 0.7901,
+      "step": 5677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4736554754302978e-08,
+      "loss": 0.7857,
+      "step": 5678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4545483930819568e-08,
+      "loss": 0.7265,
+      "step": 5679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4355150268125795e-08,
+      "loss": 0.8001,
+      "step": 5680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4165553822901312e-08,
+      "loss": 0.7245,
+      "step": 5681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3976694651605127e-08,
+      "loss": 0.7371,
+      "step": 5682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3788572810477528e-08,
+      "loss": 0.7443,
+      "step": 5683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3601188355538706e-08,
+      "loss": 0.5996,
+      "step": 5684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3414541342590136e-08,
+      "loss": 0.6929,
+      "step": 5685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3228631827212357e-08,
+      "loss": 0.6728,
+      "step": 5686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.304345986476747e-08,
+      "loss": 0.6589,
+      "step": 5687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2859025510397203e-08,
+      "loss": 0.6316,
+      "step": 5688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.267532881902401e-08,
+      "loss": 0.7642,
+      "step": 5689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2492369845350527e-08,
+      "loss": 0.7164,
+      "step": 5690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.231014864385983e-08,
+      "loss": 0.8031,
+      "step": 5691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2128665268815185e-08,
+      "loss": 0.5695,
+      "step": 5692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.194791977426003e-08,
+      "loss": 0.7312,
+      "step": 5693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1767912214018526e-08,
+      "loss": 0.7089,
+      "step": 5694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.158864264169447e-08,
+      "loss": 0.729,
+      "step": 5695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1410111110672372e-08,
+      "loss": 0.7495,
+      "step": 5696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.123231767411693e-08,
+      "loss": 0.7197,
+      "step": 5697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1055262384972464e-08,
+      "loss": 0.6607,
+      "step": 5698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0878945295964294e-08,
+      "loss": 0.7369,
+      "step": 5699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0703366459597362e-08,
+      "loss": 0.8063,
+      "step": 5700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0528525928157073e-08,
+      "loss": 0.7721,
+      "step": 5701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0354423753708717e-08,
+      "loss": 0.7349,
+      "step": 5702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0181059988098052e-08,
+      "loss": 0.7236,
+      "step": 5703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0008434682950717e-08,
+      "loss": 0.7316,
+      "step": 5704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9836547889671986e-08,
+      "loss": 0.7775,
+      "step": 5705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9665399659448137e-08,
+      "loss": 0.6453,
+      "step": 5706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9494990043245066e-08,
+      "loss": 0.7466,
+      "step": 5707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9325319091808847e-08,
+      "loss": 0.7664,
+      "step": 5708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9156386855664898e-08,
+      "loss": 0.7005,
+      "step": 5709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8988193385119646e-08,
+      "loss": 0.7867,
+      "step": 5710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8820738730259413e-08,
+      "loss": 0.7283,
+      "step": 5711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865402294094959e-08,
+      "loss": 0.8227,
+      "step": 5712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8488046066836297e-08,
+      "loss": 0.6828,
+      "step": 5713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.832280815734555e-08,
+      "loss": 0.7601,
+      "step": 5714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815830926168327e-08,
+      "loss": 0.7318,
+      "step": 5715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7994549428835273e-08,
+      "loss": 0.8007,
+      "step": 5716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7831528707566993e-08,
+      "loss": 0.7905,
+      "step": 5717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7669247146424594e-08,
+      "loss": 0.7701,
+      "step": 5718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7507704793733037e-08,
+      "loss": 0.7906,
+      "step": 5719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7346901697598006e-08,
+      "loss": 0.8282,
+      "step": 5720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.718683790590481e-08,
+      "loss": 0.7626,
+      "step": 5721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7027513466318368e-08,
+      "loss": 0.6562,
+      "step": 5722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6868928426283516e-08,
+      "loss": 0.701,
+      "step": 5723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.671108283302497e-08,
+      "loss": 0.7436,
+      "step": 5724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6553976733547638e-08,
+      "loss": 0.7881,
+      "step": 5725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.639761017463548e-08,
+      "loss": 0.7294,
+      "step": 5726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.624198320285264e-08,
+      "loss": 0.6922,
+      "step": 5727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6087095864543435e-08,
+      "loss": 0.7415,
+      "step": 5728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5932948205830688e-08,
+      "loss": 0.7438,
+      "step": 5729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779540272618244e-08,
+      "loss": 0.7947,
+      "step": 5730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.562687211058872e-08,
+      "loss": 0.6875,
+      "step": 5731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5474943765205475e-08,
+      "loss": 0.7441,
+      "step": 5732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.532375528171065e-08,
+      "loss": 0.6795,
+      "step": 5733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.517330670512629e-08,
+      "loss": 0.7257,
+      "step": 5734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.502359808025433e-08,
+      "loss": 0.6374,
+      "step": 5735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.487462945167606e-08,
+      "loss": 0.724,
+      "step": 5736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4726400863752654e-08,
+      "loss": 0.7831,
+      "step": 5737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4578912360624919e-08,
+      "loss": 0.7294,
+      "step": 5738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4432163986213e-08,
+      "loss": 0.7755,
+      "step": 5739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4286155784216659e-08,
+      "loss": 0.6771,
+      "step": 5740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4140887798115843e-08,
+      "loss": 0.6931,
+      "step": 5741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3996360071169279e-08,
+      "loss": 0.6762,
+      "step": 5742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3852572646415873e-08,
+      "loss": 0.7642,
+      "step": 5743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3709525566673321e-08,
+      "loss": 0.691,
+      "step": 5744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3567218874540044e-08,
+      "loss": 0.6724,
+      "step": 5745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3425652612392425e-08,
+      "loss": 0.7392,
+      "step": 5746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3284826822387852e-08,
+      "loss": 0.7749,
+      "step": 5747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3144741546462503e-08,
+      "loss": 0.7793,
+      "step": 5748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3005396826331618e-08,
+      "loss": 0.7264,
+      "step": 5749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.286679270349117e-08,
+      "loss": 0.6484,
+      "step": 5750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2728929219215091e-08,
+      "loss": 0.7379,
+      "step": 5751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.259180641455776e-08,
+      "loss": 0.7331,
+      "step": 5752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2455424330352905e-08,
+      "loss": 0.667,
+      "step": 5753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2319783007213315e-08,
+      "loss": 0.7654,
+      "step": 5754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2184882485531124e-08,
+      "loss": 0.7086,
+      "step": 5755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2050722805478365e-08,
+      "loss": 0.7953,
+      "step": 5756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1917304007006136e-08,
+      "loss": 0.6987,
+      "step": 5757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1784626129844878e-08,
+      "loss": 0.7918,
+      "step": 5758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1652689213504932e-08,
+      "loss": 0.6398,
+      "step": 5759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1521493297274588e-08,
+      "loss": 0.5638,
+      "step": 5760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1391038420223433e-08,
+      "loss": 0.7188,
+      "step": 5761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1261324621198722e-08,
+      "loss": 0.7159,
+      "step": 5762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1132351938828167e-08,
+      "loss": 0.7661,
+      "step": 5763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1004120411517717e-08,
+      "loss": 0.773,
+      "step": 5764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0876630077453487e-08,
+      "loss": 0.7384,
+      "step": 5765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0749880974600669e-08,
+      "loss": 0.741,
+      "step": 5766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.062387314070351e-08,
+      "loss": 0.7313,
+      "step": 5767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.049860661328589e-08,
+      "loss": 0.6431,
+      "step": 5768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0374081429650185e-08,
+      "loss": 0.7762,
+      "step": 5769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.025029762687868e-08,
+      "loss": 0.7646,
+      "step": 5770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0127255241833e-08,
+      "loss": 0.7176,
+      "step": 5771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0004954311153559e-08,
+      "loss": 0.7525,
+      "step": 5772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.883394871260111e-09,
+      "loss": 0.4875,
+      "step": 5773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.762576958351478e-09,
+      "loss": 0.7339,
+      "step": 5774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.642500608405825e-09,
+      "loss": 0.6716,
+      "step": 5775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.52316585718066e-09,
+      "loss": 0.7322,
+      "step": 5776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.404572740212004e-09,
+      "loss": 0.7403,
+      "step": 5777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.286721292816048e-09,
+      "loss": 0.5178,
+      "step": 5778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.169611550086943e-09,
+      "loss": 0.6876,
+      "step": 5779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.053243546899293e-09,
+      "loss": 0.7062,
+      "step": 5780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.937617317905379e-09,
+      "loss": 0.7103,
+      "step": 5781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.822732897537655e-09,
+      "loss": 0.7099,
+      "step": 5782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.70859032000737e-09,
+      "loss": 0.7385,
+      "step": 5783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.595189619304833e-09,
+      "loss": 0.7361,
+      "step": 5784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.482530829199143e-09,
+      "loss": 0.6933,
+      "step": 5785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.370613983239018e-09,
+      "loss": 0.7695,
+      "step": 5786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.259439114751966e-09,
+      "loss": 0.6812,
+      "step": 5787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.149006256844283e-09,
+      "loss": 0.7837,
+      "step": 5788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.039315442401885e-09,
+      "loss": 0.6903,
+      "step": 5789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.930366704089199e-09,
+      "loss": 0.7531,
+      "step": 5790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.822160074349717e-09,
+      "loss": 0.7132,
+      "step": 5791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.714695585406273e-09,
+      "loss": 0.7429,
+      "step": 5792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.607973269260494e-09,
+      "loss": 0.7528,
+      "step": 5793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.501993157693066e-09,
+      "loss": 0.7237,
+      "step": 5794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.396755282263468e-09,
+      "loss": 0.7405,
+      "step": 5795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.292259674310243e-09,
+      "loss": 0.7288,
+      "step": 5796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.188506364951275e-09,
+      "loss": 0.7082,
+      "step": 5797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.085495385082685e-09,
+      "loss": 0.7004,
+      "step": 5798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.983226765379936e-09,
+      "loss": 0.6715,
+      "step": 5799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.8817005362975535e-09,
+      "loss": 0.7608,
+      "step": 5800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.780916728069131e-09,
+      "loss": 0.7019,
+      "step": 5801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.680875370706219e-09,
+      "loss": 0.775,
+      "step": 5802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.5815764940002634e-09,
+      "loss": 0.6634,
+      "step": 5803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.483020127521222e-09,
+      "loss": 0.5307,
+      "step": 5804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.385206300618118e-09,
+      "loss": 0.7123,
+      "step": 5805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.288135042418764e-09,
+      "loss": 0.7309,
+      "step": 5806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.1918063818297565e-09,
+      "loss": 0.7003,
+      "step": 5807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.096220347536485e-09,
+      "loss": 0.7351,
+      "step": 5808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.001376968003681e-09,
+      "loss": 0.6873,
+      "step": 5809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.907276271474305e-09,
+      "loss": 0.6855,
+      "step": 5810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.813918285970388e-09,
+      "loss": 0.7488,
+      "step": 5811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.721303039292747e-09,
+      "loss": 0.7896,
+      "step": 5812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.629430559021265e-09,
+      "loss": 0.7332,
+      "step": 5813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.538300872514613e-09,
+      "loss": 0.7911,
+      "step": 5814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.447914006909971e-09,
+      "loss": 0.7111,
+      "step": 5815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.35826998912331e-09,
+      "loss": 0.7069,
+      "step": 5816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.269368845849942e-09,
+      "loss": 0.7027,
+      "step": 5817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.181210603563136e-09,
+      "loss": 0.7892,
+      "step": 5818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.093795288515501e-09,
+      "loss": 0.7091,
+      "step": 5819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.007122926738439e-09,
+      "loss": 0.7623,
+      "step": 5820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.92119354404158e-09,
+      "loss": 0.8235,
+      "step": 5821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836007166014179e-09,
+      "loss": 0.7134,
+      "step": 5822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.751563818023442e-09,
+      "loss": 0.7579,
+      "step": 5823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6678635252153655e-09,
+      "loss": 0.8257,
+      "step": 5824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5849063125152916e-09,
+      "loss": 0.7461,
+      "step": 5825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.502692204626791e-09,
+      "loss": 0.7445,
+      "step": 5826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.421221226031947e-09,
+      "loss": 0.7285,
+      "step": 5827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3404934009924625e-09,
+      "loss": 0.6827,
+      "step": 5828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.260508753547443e-09,
+      "loss": 0.7272,
+      "step": 5829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.181267307515613e-09,
+      "loss": 0.6827,
+      "step": 5830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.102769086494207e-09,
+      "loss": 0.7447,
+      "step": 5831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025014113859249e-09,
+      "loss": 0.7415,
+      "step": 5832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.948002412764718e-09,
+      "loss": 0.698,
+      "step": 5833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.871734006144212e-09,
+      "loss": 0.7537,
+      "step": 5834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.796208916709565e-09,
+      "loss": 0.6876,
+      "step": 5835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.721427166950842e-09,
+      "loss": 0.7262,
+      "step": 5836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.647388779137451e-09,
+      "loss": 0.7578,
+      "step": 5837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5740937753170335e-09,
+      "loss": 0.7419,
+      "step": 5838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5015421773162974e-09,
+      "loss": 0.6895,
+      "step": 5839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4297340067399044e-09,
+      "loss": 0.7169,
+      "step": 5840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3586692849713055e-09,
+      "loss": 0.7222,
+      "step": 5841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2883480331730167e-09,
+      "loss": 0.686,
+      "step": 5842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.218770272286065e-09,
+      "loss": 0.7665,
+      "step": 5843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.149936023029432e-09,
+      "loss": 0.7676,
+      "step": 5844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0818453059011654e-09,
+      "loss": 0.7635,
+      "step": 5845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.014498141178379e-09,
+      "loss": 0.6887,
+      "step": 5846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9478945489155865e-09,
+      "loss": 0.7206,
+      "step": 5847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8820345489469214e-09,
+      "loss": 0.7532,
+      "step": 5848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.816918160884752e-09,
+      "loss": 0.7389,
+      "step": 5849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7525454041196777e-09,
+      "loss": 0.7111,
+      "step": 5850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.688916297821642e-09,
+      "loss": 0.7268,
+      "step": 5851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6260308609379893e-09,
+      "loss": 0.7535,
+      "step": 5852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5638891121956834e-09,
+      "loss": 0.733,
+      "step": 5853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.502491070099644e-09,
+      "loss": 0.5081,
+      "step": 5854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4418367529335796e-09,
+      "loss": 0.7633,
+      "step": 5855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3819261787594307e-09,
+      "loss": 0.6595,
+      "step": 5856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.322759365417926e-09,
+      "loss": 0.6691,
+      "step": 5857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.264336330528305e-09,
+      "loss": 0.6768,
+      "step": 5858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.206657091488318e-09,
+      "loss": 0.7484,
+      "step": 5859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1497216654739474e-09,
+      "loss": 0.6902,
+      "step": 5860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.093530069439964e-09,
+      "loss": 0.7166,
+      "step": 5861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.038082320119927e-09,
+      "loss": 0.7892,
+      "step": 5862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9833784340247962e-09,
+      "loss": 0.6903,
+      "step": 5863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.929418427445151e-09,
+      "loss": 0.7175,
+      "step": 5864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8762023164498043e-09,
+      "loss": 0.7896,
+      "step": 5865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8237301168855248e-09,
+      "loss": 0.7255,
+      "step": 5866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7720018443778685e-09,
+      "loss": 0.7145,
+      "step": 5867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7210175143314579e-09,
+      "loss": 0.7614,
+      "step": 5868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6707771419280372e-09,
+      "loss": 0.6473,
+      "step": 5869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6212807421292497e-09,
+      "loss": 0.5474,
+      "step": 5870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.572528329674139e-09,
+      "loss": 0.697,
+      "step": 5871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5245199190808136e-09,
+      "loss": 0.7721,
+      "step": 5872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4772555246456155e-09,
+      "loss": 0.6605,
+      "step": 5873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4307351604431197e-09,
+      "loss": 0.7883,
+      "step": 5874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.384958840326689e-09,
+      "loss": 0.6064,
+      "step": 5875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3399265779276417e-09,
+      "loss": 0.7612,
+      "step": 5876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.295638386656639e-09,
+      "loss": 0.7449,
+      "step": 5877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2520942797017433e-09,
+      "loss": 0.7276,
+      "step": 5878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2092942700298038e-09,
+      "loss": 0.6877,
+      "step": 5879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1672383703864588e-09,
+      "loss": 0.7851,
+      "step": 5880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1259265932953012e-09,
+      "loss": 0.7026,
+      "step": 5881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0853589510587125e-09,
+      "loss": 0.711,
+      "step": 5882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.045535455756752e-09,
+      "loss": 0.5349,
+      "step": 5883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0064561192488221e-09,
+      "loss": 0.7034,
+      "step": 5884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.681209531720026e-10,
+      "loss": 0.7427,
+      "step": 5885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.305299689424396e-10,
+      "loss": 0.7089,
+      "step": 5886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.936831777536792e-10,
+      "loss": 0.7044,
+      "step": 5887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.575805905788881e-10,
+      "loss": 0.7639,
+      "step": 5888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.222222181683559e-10,
+      "loss": 0.6661,
+      "step": 5889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.876080710519929e-10,
+      "loss": 0.7447,
+      "step": 5890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.537381595371096e-10,
+      "loss": 0.7867,
+      "step": 5891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.206124937100823e-10,
+      "loss": 0.6714,
+      "step": 5892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.882310834346872e-10,
+      "loss": 0.5353,
+      "step": 5893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.565939383545994e-10,
+      "loss": 0.7442,
+      "step": 5894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.25701067890061e-10,
+      "loss": 0.797,
+      "step": 5895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.955524812414903e-10,
+      "loss": 0.7184,
+      "step": 5896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.661481873861508e-10,
+      "loss": 0.7455,
+      "step": 5897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.374881950803712e-10,
+      "loss": 0.7663,
+      "step": 5898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.095725128589912e-10,
+      "loss": 0.6927,
+      "step": 5899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.824011490350832e-10,
+      "loss": 0.5174,
+      "step": 5900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5597411169939763e-10,
+      "loss": 0.6742,
+      "step": 5901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.302914087220278e-10,
+      "loss": 0.681,
+      "step": 5902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0535304775102257e-10,
+      "loss": 0.6879,
+      "step": 5903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8115903621238627e-10,
+      "loss": 0.7609,
+      "step": 5904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5770938131118873e-10,
+      "loss": 0.7186,
+      "step": 5905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350040900299001e-10,
+      "loss": 0.6871,
+      "step": 5906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.130431691306113e-10,
+      "loss": 0.6717,
+      "step": 5907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.918266251525359e-10,
+      "loss": 0.7735,
+      "step": 5908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7135446441367565e-10,
+      "loss": 0.6854,
+      "step": 5909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.516266930108202e-10,
+      "loss": 0.6444,
+      "step": 5910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3264331681815965e-10,
+      "loss": 0.6726,
+      "step": 5911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.144043414889496e-10,
+      "loss": 0.7394,
+      "step": 5912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9690977245467869e-10,
+      "loss": 0.6503,
+      "step": 5913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8015961492479084e-10,
+      "loss": 0.7369,
+      "step": 5914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6415387388751814e-10,
+      "loss": 0.6973,
+      "step": 5915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4889255410877045e-10,
+      "loss": 0.7224,
+      "step": 5916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3437566013380087e-10,
+      "loss": 0.6963,
+      "step": 5917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.206031962849852e-10,
+      "loss": 0.6432,
+      "step": 5918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0757516666376477e-10,
+      "loss": 0.7634,
+      "step": 5919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.529157515009157e-11,
+      "loss": 0.7299,
+      "step": 5920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.375242540131778e-11,
+      "loss": 0.7469,
+      "step": 5921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.295772085386121e-11,
+      "loss": 0.7515,
+      "step": 5922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.290746472237263e-11,
+      "loss": 0.6873,
+      "step": 5923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3601659999735764e-11,
+      "loss": 0.7469,
+      "step": 5924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504030945706728e-11,
+      "loss": 0.696,
+      "step": 5925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.722341564371679e-11,
+      "loss": 0.6817,
+      "step": 5926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0150980887544426e-11,
+      "loss": 0.635,
+      "step": 5927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3823007294643265e-11,
+      "loss": 0.7171,
+      "step": 5928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8239496749339336e-11,
+      "loss": 0.7686,
+      "step": 5929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3400450914469176e-11,
+      "loss": 0.7662,
+      "step": 5930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.305871230824715e-12,
+      "loss": 0.7375,
+      "step": 5931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.955758917985943e-12,
+      "loss": 0.6988,
+      "step": 5932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3501149732106942e-12,
+      "loss": 0.7444,
+      "step": 5933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4889401728224174e-12,
+      "loss": 0.7007,
+      "step": 5934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.722350708224021e-13,
+      "loss": 0.6976,
+      "step": 5935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5245,
+      "step": 5936
+    },
+    {
+      "epoch": 1.0,
+      "step": 5936,
+      "total_flos": 2157678509735936.0,
+      "train_loss": 0.17528112896409317,
+      "train_runtime": 16210.4198,
+      "train_samples_per_second": 46.871,
+      "train_steps_per_second": 0.366
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5936,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1500,
+  "total_flos": 2157678509735936.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}