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": 1.11731843575419e-07,
+      "loss": 1.5674,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 1.8301,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 1.7702,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 1.6435,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 0.2379,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.6605,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 1.5997,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 1.5196,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.7877,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 1.6397,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 1.5391,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 1.6533,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 1.585,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 1.5306,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 1.4156,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 0.2259,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 1.4321,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.3653,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 0.2069,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 1.3641,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 1.3171,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 1.3648,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 1.3339,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 1.3171,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 0.2476,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 1.2155,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 1.3489,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 1.2946,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 1.2961,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 1.1689,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 0.2169,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 1.1682,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 1.1408,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 1.1032,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 1.2036,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 1.1565,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 1.1858,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 1.2295,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 1.173,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 1.0546,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 1.0657,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 1.1598,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.0322,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 1.1488,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 1.096,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 1.0702,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 1.0675,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 1.0862,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 1.0776,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 1.0282,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 1.0928,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 1.1759,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 1.108,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 1.098,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 1.0288,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 1.1422,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 1.0665,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 1.0801,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 1.0783,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 1.0997,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 1.0607,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 0.2135,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 1.0188,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 0.9777,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 1.0919,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 1.1095,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 1.1058,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 1.057,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 1.0441,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 0.9872,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 0.9741,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 1.0505,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 0.2171,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 1.0804,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 1.0167,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 1.0233,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 0.9129,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 1.0462,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 1.0505,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 1.0698,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 1.0455,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 0.2142,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 0.9956,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 1.0248,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 0.962,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 0.2526,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 1.0136,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 0.9981,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.9949,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.005586592178771e-05,
+      "loss": 1.0123,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0167597765363129e-05,
+      "loss": 1.0163,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0279329608938548e-05,
+      "loss": 1.0412,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0391061452513967e-05,
+      "loss": 0.9385,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 0.227,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0614525139664806e-05,
+      "loss": 1.0155,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0726256983240225e-05,
+      "loss": 0.99,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0837988826815644e-05,
+      "loss": 0.8994,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0949720670391063e-05,
+      "loss": 1.0445,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 0.9696,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.11731843575419e-05,
+      "loss": 0.2057,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1284916201117319e-05,
+      "loss": 1.0262,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1396648044692738e-05,
+      "loss": 0.9077,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1508379888268157e-05,
+      "loss": 0.982,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 0.9806,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1731843575418994e-05,
+      "loss": 0.9862,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1843575418994415e-05,
+      "loss": 0.9866,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1955307262569834e-05,
+      "loss": 1.002,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2067039106145253e-05,
+      "loss": 1.0165,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 1.0051,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.229050279329609e-05,
+      "loss": 1.0131,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2402234636871509e-05,
+      "loss": 0.9648,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2513966480446928e-05,
+      "loss": 1.0191,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625698324022347e-05,
+      "loss": 1.0083,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 0.9142,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2849162011173186e-05,
+      "loss": 0.9712,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2960893854748603e-05,
+      "loss": 1.0077,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3072625698324024e-05,
+      "loss": 0.9705,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3184357541899443e-05,
+      "loss": 1.0275,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 1.0301,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.340782122905028e-05,
+      "loss": 1.0233,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3519553072625699e-05,
+      "loss": 0.9611,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3631284916201118e-05,
+      "loss": 0.9757,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3743016759776537e-05,
+      "loss": 0.8996,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 0.9764,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3966480446927374e-05,
+      "loss": 1.0181,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4078212290502795e-05,
+      "loss": 0.9832,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4189944134078212e-05,
+      "loss": 0.8543,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4301675977653633e-05,
+      "loss": 1.0326,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 0.9824,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.452513966480447e-05,
+      "loss": 1.0121,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.463687150837989e-05,
+      "loss": 1.113,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4748603351955308e-05,
+      "loss": 0.9711,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4860335195530729e-05,
+      "loss": 0.9681,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 0.9891,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083798882681566e-05,
+      "loss": 0.9343,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5195530726256983e-05,
+      "loss": 1.0039,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5307262569832404e-05,
+      "loss": 0.9733,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5418994413407823e-05,
+      "loss": 0.9799,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 0.9143,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564245810055866e-05,
+      "loss": 1.0471,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575418994413408e-05,
+      "loss": 0.9547,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5865921787709498e-05,
+      "loss": 0.9952,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5977653631284917e-05,
+      "loss": 0.9373,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 0.9773,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6201117318435755e-05,
+      "loss": 0.9202,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6312849162011177e-05,
+      "loss": 0.967,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6424581005586592e-05,
+      "loss": 0.9688,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6536312849162014e-05,
+      "loss": 0.8971,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 0.9278,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6759776536312852e-05,
+      "loss": 0.935,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.687150837988827e-05,
+      "loss": 0.8913,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698324022346369e-05,
+      "loss": 0.998,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.709497206703911e-05,
+      "loss": 0.9799,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 1.0135,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7318435754189946e-05,
+      "loss": 0.9647,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7430167597765365e-05,
+      "loss": 0.9132,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7541899441340784e-05,
+      "loss": 0.9594,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7653631284916203e-05,
+      "loss": 0.9725,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 0.9203,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.787709497206704e-05,
+      "loss": 1.0047,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.798882681564246e-05,
+      "loss": 0.8371,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8100558659217878e-05,
+      "loss": 0.9514,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8212290502793297e-05,
+      "loss": 1.005,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 0.927,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8435754189944135e-05,
+      "loss": 0.9495,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8547486033519553e-05,
+      "loss": 0.9416,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8659217877094972e-05,
+      "loss": 0.9062,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8770949720670394e-05,
+      "loss": 0.9633,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 0.8724,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8994413407821232e-05,
+      "loss": 0.9148,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.910614525139665e-05,
+      "loss": 1.0373,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.921787709497207e-05,
+      "loss": 0.9289,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.932960893854749e-05,
+      "loss": 0.9153,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 0.9067,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9553072625698326e-05,
+      "loss": 0.9755,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9664804469273745e-05,
+      "loss": 0.9831,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776536312849164e-05,
+      "loss": 0.9861,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9888268156424583e-05,
+      "loss": 0.2682,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9925,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999851105972e-05,
+      "loss": 0.9235,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999404423931e-05,
+      "loss": 1.0182,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986599540108e-05,
+      "loss": 0.932,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976176964328e-05,
+      "loss": 0.9557,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999962776515077e-05,
+      "loss": 0.993,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999946398196345e-05,
+      "loss": 0.9797,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999927042013002e-05,
+      "loss": 0.2372,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904707970824e-05,
+      "loss": 0.2344,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999879396076452e-05,
+      "loss": 0.8939,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999851106337428e-05,
+      "loss": 0.9363,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999819838762172e-05,
+      "loss": 0.9731,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999785593360003e-05,
+      "loss": 0.8761,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999974837014111e-05,
+      "loss": 0.9599,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970816911659e-05,
+      "loss": 0.964,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999664990298395e-05,
+      "loss": 0.953,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999618833699402e-05,
+      "loss": 0.9161,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956969933335e-05,
+      "loss": 0.9795,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999951758721486e-05,
+      "loss": 0.9686,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999462497359468e-05,
+      "loss": 1.0378,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999404429783567e-05,
+      "loss": 0.9662,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999934338450445e-05,
+      "loss": 1.0186,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999279361540302e-05,
+      "loss": 0.8895,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999212360910184e-05,
+      "loss": 0.8763,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999142382634047e-05,
+      "loss": 0.9662,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999906942673273e-05,
+      "loss": 0.2722,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999899349322796e-05,
+      "loss": 0.9984,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998914582142347e-05,
+      "loss": 0.896,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999883269349939e-05,
+      "loss": 0.9705,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999874782732348e-05,
+      "loss": 0.9495,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998659983639883e-05,
+      "loss": 0.9329,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998569162474757e-05,
+      "loss": 0.8785,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998475363855152e-05,
+      "loss": 0.9544,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998378587808998e-05,
+      "loss": 0.9602,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998278834365113e-05,
+      "loss": 0.9597,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998176103553205e-05,
+      "loss": 0.9721,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998070395403862e-05,
+      "loss": 0.8802,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997961709948566e-05,
+      "loss": 0.9672,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999785004721968e-05,
+      "loss": 0.9917,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997735407250455e-05,
+      "loss": 0.9266,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997617790075036e-05,
+      "loss": 0.9234,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999749719572844e-05,
+      "loss": 0.9826,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997373624246584e-05,
+      "loss": 0.8771,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997247075666264e-05,
+      "loss": 0.9583,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997117550025163e-05,
+      "loss": 0.9955,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996985047361855e-05,
+      "loss": 0.9145,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999684956771579e-05,
+      "loss": 0.9846,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996711111127326e-05,
+      "loss": 0.9386,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996569677637687e-05,
+      "loss": 0.9402,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996425267288987e-05,
+      "loss": 0.9071,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999627788012423e-05,
+      "loss": 0.8982,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996127516187313e-05,
+      "loss": 0.9826,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995974175523006e-05,
+      "loss": 0.9218,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995817858176973e-05,
+      "loss": 0.9274,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999565856419577e-05,
+      "loss": 0.924,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999549629362682e-05,
+      "loss": 0.9355,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995331046518456e-05,
+      "loss": 0.9369,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999516282291988e-05,
+      "loss": 0.9347,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994991622881195e-05,
+      "loss": 0.9365,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994817446453374e-05,
+      "loss": 0.9917,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999464029368829e-05,
+      "loss": 0.9268,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994460164638692e-05,
+      "loss": 0.9696,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999427705935823e-05,
+      "loss": 0.9764,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999409097790142e-05,
+      "loss": 0.9715,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390192032368e-05,
+      "loss": 0.9325,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993709886681305e-05,
+      "loss": 0.9666,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993514877031486e-05,
+      "loss": 0.8393,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999331689143229e-05,
+      "loss": 0.9205,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993115929942678e-05,
+      "loss": 0.9195,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999291199262249e-05,
+      "loss": 0.9463,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999270507953246e-05,
+      "loss": 0.8782,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992495190734203e-05,
+      "loss": 0.9438,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999228232629022e-05,
+      "loss": 0.9481,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99920664862639e-05,
+      "loss": 0.8559,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999184767071952e-05,
+      "loss": 0.9386,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999162587972224e-05,
+      "loss": 0.9312,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991401113338103e-05,
+      "loss": 0.8937,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999117337163405e-05,
+      "loss": 0.9075,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990942654677887e-05,
+      "loss": 0.9498,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999070896253833e-05,
+      "loss": 1.0044,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990472295284963e-05,
+      "loss": 0.9414,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990232652988268e-05,
+      "loss": 0.8639,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989990035719602e-05,
+      "loss": 0.9502,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998974444355122e-05,
+      "loss": 0.8866,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998949587655625e-05,
+      "loss": 0.9229,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989244334808714e-05,
+      "loss": 0.8962,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988989818383523e-05,
+      "loss": 0.9594,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998873232735646e-05,
+      "loss": 0.9917,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988471861804217e-05,
+      "loss": 0.8707,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988208421804337e-05,
+      "loss": 0.9285,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998794200743529e-05,
+      "loss": 0.9638,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987672618776398e-05,
+      "loss": 0.9513,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987400255907886e-05,
+      "loss": 0.9092,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987124918910858e-05,
+      "loss": 0.9807,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998684660786731e-05,
+      "loss": 0.8766,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986565322860117e-05,
+      "loss": 0.8339,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986281063973044e-05,
+      "loss": 0.8735,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985993831290734e-05,
+      "loss": 0.8462,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985703624898732e-05,
+      "loss": 0.9449,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998541044488345e-05,
+      "loss": 0.947,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99851142913322e-05,
+      "loss": 0.2572,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984815164333163e-05,
+      "loss": 0.9266,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984513063975423e-05,
+      "loss": 0.8342,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984207990348943e-05,
+      "loss": 0.9191,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983899943544563e-05,
+      "loss": 0.896,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983588923654026e-05,
+      "loss": 0.2656,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983274930769938e-05,
+      "loss": 0.9466,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982957964985813e-05,
+      "loss": 0.2217,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998263802639603e-05,
+      "loss": 0.8629,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982315115095875e-05,
+      "loss": 0.9972,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981989231181494e-05,
+      "loss": 0.9765,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998166037474994e-05,
+      "loss": 0.9844,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981328545899143e-05,
+      "loss": 0.9499,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980993744727906e-05,
+      "loss": 0.943,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980655971335944e-05,
+      "loss": 0.9015,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980315225823836e-05,
+      "loss": 0.8696,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997997150829305e-05,
+      "loss": 0.9383,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997962481884594e-05,
+      "loss": 0.867,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997927515758575e-05,
+      "loss": 0.946,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99789225246166e-05,
+      "loss": 0.9533,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997856692004351e-05,
+      "loss": 0.9521,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997820834397236e-05,
+      "loss": 0.8602,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977846796509942e-05,
+      "loss": 0.8908,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977482277763917e-05,
+      "loss": 0.756,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997711478784283e-05,
+      "loss": 0.8738,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976744326856118e-05,
+      "loss": 0.9695,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976370894914106e-05,
+      "loss": 0.973,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997599449212799e-05,
+      "loss": 0.8755,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975615118609854e-05,
+      "loss": 0.8677,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975232774472682e-05,
+      "loss": 0.2048,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974847459830327e-05,
+      "loss": 0.8879,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997445917479753e-05,
+      "loss": 0.911,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974067919489916e-05,
+      "loss": 0.8973,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973673694024002e-05,
+      "loss": 0.8917,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973276498517177e-05,
+      "loss": 0.9319,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972876333087728e-05,
+      "loss": 0.2395,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972473197854813e-05,
+      "loss": 0.8725,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972067092938482e-05,
+      "loss": 0.8783,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997165801845967e-05,
+      "loss": 0.9867,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971245974540197e-05,
+      "loss": 0.9201,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997083096130276e-05,
+      "loss": 0.8946,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970412978870946e-05,
+      "loss": 0.9123,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996999202736923e-05,
+      "loss": 0.9073,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996956810692296e-05,
+      "loss": 0.8971,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969141217658377e-05,
+      "loss": 0.9436,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99687113597026e-05,
+      "loss": 0.8924,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996827853318365e-05,
+      "loss": 0.9564,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967842738230393e-05,
+      "loss": 0.8529,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967403974972625e-05,
+      "loss": 0.2444,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966962243540996e-05,
+      "loss": 0.9224,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966517544067045e-05,
+      "loss": 0.9707,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966069876683204e-05,
+      "loss": 0.9342,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996561924152278e-05,
+      "loss": 0.9396,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965165638719974e-05,
+      "loss": 0.9248,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996470906840985e-05,
+      "loss": 0.9636,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996424953072838e-05,
+      "loss": 0.9546,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963787025812405e-05,
+      "loss": 0.9015,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996332155379965e-05,
+      "loss": 0.8003,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962853114828736e-05,
+      "loss": 0.884,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962381709039153e-05,
+      "loss": 0.923,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996190733657128e-05,
+      "loss": 0.8943,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996142999756638e-05,
+      "loss": 0.8313,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960949692166594e-05,
+      "loss": 0.9513,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996046642051496e-05,
+      "loss": 0.8878,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995998018275539e-05,
+      "loss": 0.9207,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959490979032668e-05,
+      "loss": 0.82,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958998809492485e-05,
+      "loss": 0.9169,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99585036742814e-05,
+      "loss": 0.9302,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958005573546857e-05,
+      "loss": 0.9915,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957504507437185e-05,
+      "loss": 0.9298,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99570004761016e-05,
+      "loss": 0.8706,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995649347969019e-05,
+      "loss": 0.9472,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995598351835393e-05,
+      "loss": 0.8676,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955470592244688e-05,
+      "loss": 0.7891,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954954701515206e-05,
+      "loss": 0.8941,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954435846319108e-05,
+      "loss": 0.8948,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953914026810904e-05,
+      "loss": 0.878,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953389243145982e-05,
+      "loss": 0.891,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952861495480623e-05,
+      "loss": 0.903,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952330783971975e-05,
+      "loss": 0.9231,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951797108778088e-05,
+      "loss": 0.8826,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995126047005788e-05,
+      "loss": 0.8668,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995072086797115e-05,
+      "loss": 0.971,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950178302678588e-05,
+      "loss": 0.9363,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994963277434177e-05,
+      "loss": 0.9332,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949084283123142e-05,
+      "loss": 0.8699,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994853282918604e-05,
+      "loss": 0.9338,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947978412694675e-05,
+      "loss": 0.8626,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994742103381415e-05,
+      "loss": 0.2513,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994686069271045e-05,
+      "loss": 0.846,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946297389550433e-05,
+      "loss": 0.9206,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945731124501845e-05,
+      "loss": 0.98,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994516189773331e-05,
+      "loss": 0.9064,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994458970941434e-05,
+      "loss": 0.9108,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944014559715325e-05,
+      "loss": 0.8791,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943436448807537e-05,
+      "loss": 0.9266,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942855376863134e-05,
+      "loss": 0.9124,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994227134405515e-05,
+      "loss": 0.8852,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941684350557503e-05,
+      "loss": 0.9441,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994109439654499e-05,
+      "loss": 0.8064,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940501482193298e-05,
+      "loss": 0.8339,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939905607678986e-05,
+      "loss": 0.8786,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939306773179498e-05,
+      "loss": 0.9169,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938704978873158e-05,
+      "loss": 0.9103,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993810022493918e-05,
+      "loss": 0.8269,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937492511557646e-05,
+      "loss": 0.9186,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993688183890953e-05,
+      "loss": 0.8585,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993626820717668e-05,
+      "loss": 0.8774,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993565161654183e-05,
+      "loss": 0.8824,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993503206718859e-05,
+      "loss": 0.8753,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993440955930146e-05,
+      "loss": 0.9533,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933784093065812e-05,
+      "loss": 0.9702,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99331556686679e-05,
+      "loss": 0.248,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932524286294868e-05,
+      "loss": 0.8899,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993188994613473e-05,
+      "loss": 0.9106,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993125264837638e-05,
+      "loss": 0.8513,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993061239320961e-05,
+      "loss": 0.97,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929969180825067e-05,
+      "loss": 0.863,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929323011414303e-05,
+      "loss": 0.8501,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928673885169732e-05,
+      "loss": 0.923,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992802180228466e-05,
+      "loss": 0.9249,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927366762953267e-05,
+      "loss": 0.9927,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992670876737062e-05,
+      "loss": 0.9233,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926047815732658e-05,
+      "loss": 0.9337,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925383908236205e-05,
+      "loss": 0.9731,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924717045078964e-05,
+      "loss": 0.8719,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924047226459523e-05,
+      "loss": 0.8934,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923374452577343e-05,
+      "loss": 0.849,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992269872363277e-05,
+      "loss": 0.83,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922020039827023e-05,
+      "loss": 0.9414,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921338401362207e-05,
+      "loss": 0.9293,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992065380844131e-05,
+      "loss": 0.8828,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99199662612682e-05,
+      "loss": 0.8803,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991927576004761e-05,
+      "loss": 0.932,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918582304985167e-05,
+      "loss": 0.9384,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991788589628737e-05,
+      "loss": 0.9192,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991718653416161e-05,
+      "loss": 0.8867,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916484218816143e-05,
+      "loss": 0.9007,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915778950460112e-05,
+      "loss": 0.9474,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915070729303533e-05,
+      "loss": 0.8766,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991435955555731e-05,
+      "loss": 0.9347,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913645429433224e-05,
+      "loss": 0.9485,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991292835114393e-05,
+      "loss": 0.9241,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912208320902963e-05,
+      "loss": 0.9065,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991148533892474e-05,
+      "loss": 0.9356,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910759405424562e-05,
+      "loss": 0.9284,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99100305206186e-05,
+      "loss": 0.8228,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909298684723905e-05,
+      "loss": 0.876,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908563897958414e-05,
+      "loss": 0.893,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990782616054093e-05,
+      "loss": 0.8977,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990708547269115e-05,
+      "loss": 0.9738,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906341834629642e-05,
+      "loss": 0.8241,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905595246577848e-05,
+      "loss": 0.91,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990484570875809e-05,
+      "loss": 0.829,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990409322139358e-05,
+      "loss": 0.8775,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903337784708396e-05,
+      "loss": 0.8069,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902579398927498e-05,
+      "loss": 0.8779,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901818064276725e-05,
+      "loss": 0.8016,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901053780982788e-05,
+      "loss": 0.2441,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900286549273292e-05,
+      "loss": 0.9241,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899516369376698e-05,
+      "loss": 0.8378,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898743241522362e-05,
+      "loss": 0.9109,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897967165940515e-05,
+      "loss": 0.9257,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989718814286226e-05,
+      "loss": 0.9299,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989640617251958e-05,
+      "loss": 0.8993,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895621255145335e-05,
+      "loss": 0.8446,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894833390973266e-05,
+      "loss": 0.9114,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894042580237994e-05,
+      "loss": 0.9062,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893248823175003e-05,
+      "loss": 0.8764,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892452120020674e-05,
+      "loss": 0.831,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891652471012254e-05,
+      "loss": 0.246,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989084987638786e-05,
+      "loss": 0.8891,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989004433638651e-05,
+      "loss": 0.8166,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889235851248066e-05,
+      "loss": 0.9217,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888424421213303e-05,
+      "loss": 0.8269,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887610046523847e-05,
+      "loss": 0.9409,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886792727422207e-05,
+      "loss": 0.8907,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885972464151774e-05,
+      "loss": 0.8835,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988514925695681e-05,
+      "loss": 0.8765,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988432310608246e-05,
+      "loss": 0.8493,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988349401177474e-05,
+      "loss": 0.8674,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882661974280544e-05,
+      "loss": 0.8792,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881826993847644e-05,
+      "loss": 0.8436,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880989070724688e-05,
+      "loss": 0.8799,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880148205161193e-05,
+      "loss": 0.8969,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987930439740757e-05,
+      "loss": 0.8907,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878457647715085e-05,
+      "loss": 0.9538,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987760795633589e-05,
+      "loss": 0.9019,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876755323523024e-05,
+      "loss": 0.832,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987589974953038e-05,
+      "loss": 0.9383,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875041234612738e-05,
+      "loss": 0.8325,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874179779025762e-05,
+      "loss": 0.9102,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873315383025974e-05,
+      "loss": 0.9675,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872448046870786e-05,
+      "loss": 0.8471,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871577770818476e-05,
+      "loss": 0.9217,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870704555128206e-05,
+      "loss": 0.7781,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986982840006001e-05,
+      "loss": 0.8702,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986894930587479e-05,
+      "loss": 0.8882,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986806727283434e-05,
+      "loss": 0.8852,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986718230120131e-05,
+      "loss": 0.8489,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986629439123924e-05,
+      "loss": 0.8965,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865403543212532e-05,
+      "loss": 0.2492,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864509757386474e-05,
+      "loss": 0.8414,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.8864,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986271337340182e-05,
+      "loss": 0.9393,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986181077577816e-05,
+      "loss": 0.9451,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860905241425043e-05,
+      "loss": 0.856,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985999677061211e-05,
+      "loss": 0.8948,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98590853636099e-05,
+      "loss": 0.8161,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985817102068982e-05,
+      "loss": 0.8778,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857253742124144e-05,
+      "loss": 0.8817,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856333528186034e-05,
+      "loss": 0.9278,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985541037914952e-05,
+      "loss": 0.8782,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854484295289493e-05,
+      "loss": 0.9204,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853555276881744e-05,
+      "loss": 0.867,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852623324202914e-05,
+      "loss": 0.9457,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985168843753053e-05,
+      "loss": 0.9082,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850750617142994e-05,
+      "loss": 0.8599,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849809863319574e-05,
+      "loss": 0.8316,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848866176340414e-05,
+      "loss": 0.9093,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847919556486534e-05,
+      "loss": 0.927,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846970004039825e-05,
+      "loss": 0.8644,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846017519283055e-05,
+      "loss": 0.907,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984506210249986e-05,
+      "loss": 0.9209,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844103753974753e-05,
+      "loss": 0.9295,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843142473993123e-05,
+      "loss": 0.836,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984217826284122e-05,
+      "loss": 0.8993,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984121112080618e-05,
+      "loss": 0.8664,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840241048176004e-05,
+      "loss": 0.8503,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839268045239566e-05,
+      "loss": 0.7703,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983829211228662e-05,
+      "loss": 0.779,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983731324960778e-05,
+      "loss": 0.2667,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836331457494546e-05,
+      "loss": 0.8757,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835346736239285e-05,
+      "loss": 0.9244,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983435908613523e-05,
+      "loss": 0.8261,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983336850747649e-05,
+      "loss": 0.7938,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832375000558054e-05,
+      "loss": 0.891,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831378565675773e-05,
+      "loss": 0.8323,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830379203126376e-05,
+      "loss": 0.8787,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829376913207455e-05,
+      "loss": 0.9166,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828371696217486e-05,
+      "loss": 0.8784,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982736355245581e-05,
+      "loss": 0.8622,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982635248222264e-05,
+      "loss": 0.7639,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825338485819056e-05,
+      "loss": 0.9034,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982432156354702e-05,
+      "loss": 0.8617,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823301715709355e-05,
+      "loss": 0.8298,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982227894260976e-05,
+      "loss": 0.9298,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982125324455281e-05,
+      "loss": 0.8241,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982022462184394e-05,
+      "loss": 0.2484,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981919307478946e-05,
+      "loss": 0.8413,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818158603696558e-05,
+      "loss": 0.9324,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817121208873284e-05,
+      "loss": 0.7985,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816080890628564e-05,
+      "loss": 0.9078,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815037649272188e-05,
+      "loss": 0.2511,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813991485114826e-05,
+      "loss": 0.9312,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812942398468008e-05,
+      "loss": 0.8463,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811890389644146e-05,
+      "loss": 0.8165,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981083545895651e-05,
+      "loss": 0.901,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980977760671925e-05,
+      "loss": 0.8927,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808716833247374e-05,
+      "loss": 0.9148,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980765313885678e-05,
+      "loss": 0.8931,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806586523864212e-05,
+      "loss": 0.8879,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98055169885873e-05,
+      "loss": 0.7861,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980444453334454e-05,
+      "loss": 0.8504,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803369158455293e-05,
+      "loss": 0.8621,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802290864239796e-05,
+      "loss": 0.9103,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801209651019153e-05,
+      "loss": 0.8627,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980012551911533e-05,
+      "loss": 0.8755,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799038468851177e-05,
+      "loss": 0.8937,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97979485005504e-05,
+      "loss": 0.8582,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979685561453758e-05,
+      "loss": 0.8412,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795759811138156e-05,
+      "loss": 0.8736,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794661090678464e-05,
+      "loss": 0.8539,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793559453485676e-05,
+      "loss": 0.9357,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792454899887845e-05,
+      "loss": 0.7946,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791347430213906e-05,
+      "loss": 0.8188,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790237044793637e-05,
+      "loss": 0.827,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789123743957707e-05,
+      "loss": 0.2624,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978800752803764e-05,
+      "loss": 0.852,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786888397365828e-05,
+      "loss": 0.2592,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978576635227554e-05,
+      "loss": 0.8927,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978464139310091e-05,
+      "loss": 0.9457,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783513520176933e-05,
+      "loss": 0.9449,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782382733839472e-05,
+      "loss": 0.8295,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978124903442527e-05,
+      "loss": 0.8826,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780112422271928e-05,
+      "loss": 0.9408,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778972897717906e-05,
+      "loss": 0.9272,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777830461102555e-05,
+      "loss": 0.8737,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776685112766064e-05,
+      "loss": 0.8371,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977553685304952e-05,
+      "loss": 0.941,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774385682294848e-05,
+      "loss": 0.9244,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977323160084486e-05,
+      "loss": 0.8802,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772074609043224e-05,
+      "loss": 0.8729,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977091470723448e-05,
+      "loss": 0.876,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769751895764034e-05,
+      "loss": 0.841,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768586174978156e-05,
+      "loss": 0.8927,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767417545223985e-05,
+      "loss": 0.8896,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976624600684952e-05,
+      "loss": 0.8177,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765071560203636e-05,
+      "loss": 0.9056,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976389420563607e-05,
+      "loss": 0.8588,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762713943497425e-05,
+      "loss": 0.8984,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976153077413916e-05,
+      "loss": 0.9241,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976034469791362e-05,
+      "loss": 0.8336,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759155715173996e-05,
+      "loss": 0.8868,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757963826274357e-05,
+      "loss": 0.8959,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756769031569635e-05,
+      "loss": 0.8657,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975557133141562e-05,
+      "loss": 0.8856,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754370726168975e-05,
+      "loss": 0.8217,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975316721618723e-05,
+      "loss": 0.8996,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751960801828768e-05,
+      "loss": 0.8576,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750751483452854e-05,
+      "loss": 0.8683,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749539261419604e-05,
+      "loss": 0.8455,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974832413609e-05,
+      "loss": 0.9154,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747106107825897e-05,
+      "loss": 0.9381,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745885176990007e-05,
+      "loss": 0.8937,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744661343945908e-05,
+      "loss": 0.8941,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743434609058046e-05,
+      "loss": 0.8533,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742204972691725e-05,
+      "loss": 0.2667,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740972435213114e-05,
+      "loss": 0.8195,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739736996989253e-05,
+      "loss": 0.964,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973849865838804e-05,
+      "loss": 0.8128,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737257419778233e-05,
+      "loss": 0.7721,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736013281529463e-05,
+      "loss": 0.9425,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734766244012218e-05,
+      "loss": 0.8928,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973351630759785e-05,
+      "loss": 0.8368,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732263472658574e-05,
+      "loss": 0.843,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973100773956747e-05,
+      "loss": 0.9398,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729749108698484e-05,
+      "loss": 0.7759,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728487580426417e-05,
+      "loss": 0.9316,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727223155126938e-05,
+      "loss": 0.9493,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972595583317658e-05,
+      "loss": 0.8742,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724685614952733e-05,
+      "loss": 0.8171,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723412500833656e-05,
+      "loss": 0.8795,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972213649119846e-05,
+      "loss": 0.8053,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720857586427135e-05,
+      "loss": 0.8518,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971957578690052e-05,
+      "loss": 0.8499,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971829109300032e-05,
+      "loss": 0.8451,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717003505109097e-05,
+      "loss": 0.8189,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715713023610285e-05,
+      "loss": 0.8339,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971441964888817e-05,
+      "loss": 0.916,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971312338132791e-05,
+      "loss": 0.8631,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971182422131551e-05,
+      "loss": 0.2646,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710522169237848e-05,
+      "loss": 0.8865,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709217225482658e-05,
+      "loss": 0.8959,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970790939043854e-05,
+      "loss": 0.8625,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706598664494948e-05,
+      "loss": 0.9006,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705285048042203e-05,
+      "loss": 0.886,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703968541471485e-05,
+      "loss": 0.8666,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970264914517483e-05,
+      "loss": 0.7806,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701326859545143e-05,
+      "loss": 0.7882,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970000168497618e-05,
+      "loss": 0.8819,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698673621862568e-05,
+      "loss": 0.7836,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697342670599783e-05,
+      "loss": 0.8866,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696008831584172e-05,
+      "loss": 0.888,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694672105212927e-05,
+      "loss": 0.8782,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969333249188412e-05,
+      "loss": 0.8723,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691989991996663e-05,
+      "loss": 0.8386,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690644605950343e-05,
+      "loss": 0.8851,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689296334145794e-05,
+      "loss": 0.8576,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968794517698452e-05,
+      "loss": 0.242,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686591134868883e-05,
+      "loss": 0.914,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968523420820209e-05,
+      "loss": 0.8423,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683874397388223e-05,
+      "loss": 0.8396,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968251170283222e-05,
+      "loss": 0.9492,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681146124939873e-05,
+      "loss": 0.8804,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967977766411783e-05,
+      "loss": 0.8632,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967840632077361e-05,
+      "loss": 0.8219,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967703209531558e-05,
+      "loss": 0.8336,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967565498815297e-05,
+      "loss": 0.8897,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674274999695858e-05,
+      "loss": 0.8729,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672892130355193e-05,
+      "loss": 0.8369,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967150638054278e-05,
+      "loss": 0.8535,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670117750671277e-05,
+      "loss": 0.9152,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668726241154198e-05,
+      "loss": 0.843,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667331852405923e-05,
+      "loss": 0.8499,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966593458484168e-05,
+      "loss": 0.8819,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664534438877563e-05,
+      "loss": 0.9348,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663131414930514e-05,
+      "loss": 0.8413,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966172551341834e-05,
+      "loss": 0.8852,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96603167347597e-05,
+      "loss": 0.7513,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658905079374112e-05,
+      "loss": 0.855,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657490547681948e-05,
+      "loss": 0.8925,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656073140104443e-05,
+      "loss": 0.8211,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965465285706368e-05,
+      "loss": 0.804,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965322969898261e-05,
+      "loss": 0.8864,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651803666285022e-05,
+      "loss": 0.8558,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650374759395577e-05,
+      "loss": 0.8449,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648942978739783e-05,
+      "loss": 0.8397,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647508324744012e-05,
+      "loss": 0.8383,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646070797835488e-05,
+      "loss": 0.8129,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964463039844228e-05,
+      "loss": 0.8346,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643187126993332e-05,
+      "loss": 0.8303,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641740983918426e-05,
+      "loss": 0.8171,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964029196964821e-05,
+      "loss": 0.8321,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638840084614182e-05,
+      "loss": 0.837,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637385329248695e-05,
+      "loss": 0.8266,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963592770398496e-05,
+      "loss": 0.8333,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963446720925704e-05,
+      "loss": 0.8326,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633003845499846e-05,
+      "loss": 0.8836,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631537613149164e-05,
+      "loss": 0.8636,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963006851264161e-05,
+      "loss": 0.8153,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628596544414668e-05,
+      "loss": 0.8983,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962712170890667e-05,
+      "loss": 0.2796,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962564400655681e-05,
+      "loss": 0.8565,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962416343780512e-05,
+      "loss": 0.9181,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622680003092503e-05,
+      "loss": 0.8649,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962119370286071e-05,
+      "loss": 0.8839,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619704537552345e-05,
+      "loss": 0.8748,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618212507610854e-05,
+      "loss": 0.8339,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961671761348055e-05,
+      "loss": 0.8802,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615219855606598e-05,
+      "loss": 0.8195,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961371923443501e-05,
+      "loss": 0.879,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612215750412654e-05,
+      "loss": 0.8354,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610709403987248e-05,
+      "loss": 0.8179,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609200195607363e-05,
+      "loss": 0.8666,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607688125722426e-05,
+      "loss": 0.8482,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606173194782712e-05,
+      "loss": 0.832,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960465540323935e-05,
+      "loss": 0.881,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603134751544323e-05,
+      "loss": 0.8293,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601611240150455e-05,
+      "loss": 0.7898,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600084869511435e-05,
+      "loss": 0.8215,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95985556400818e-05,
+      "loss": 0.8007,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959702355231693e-05,
+      "loss": 0.9187,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595488606673067e-05,
+      "loss": 0.7721,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593950803607302e-05,
+      "loss": 0.9099,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592410143577566e-05,
+      "loss": 0.2666,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590866627042654e-05,
+      "loss": 0.9039,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589320254462206e-05,
+      "loss": 0.9112,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587771026296717e-05,
+      "loss": 0.8852,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586218943007525e-05,
+      "loss": 0.8239,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958466400505682e-05,
+      "loss": 0.9354,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958310621290765e-05,
+      "loss": 0.8847,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95815455670239e-05,
+      "loss": 0.8644,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579982067870317e-05,
+      "loss": 0.8624,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957841571591249e-05,
+      "loss": 0.8551,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957684651161686e-05,
+      "loss": 0.891,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575274455450718e-05,
+      "loss": 0.8084,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573699547882204e-05,
+      "loss": 0.8821,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572121789380306e-05,
+      "loss": 0.8623,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957054118041486e-05,
+      "loss": 0.9542,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568957721456553e-05,
+      "loss": 0.8228,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567371412976923e-05,
+      "loss": 0.2312,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565782255448354e-05,
+      "loss": 0.8195,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956419024934407e-05,
+      "loss": 0.8751,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956259539513816e-05,
+      "loss": 0.8538,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560997693305554e-05,
+      "loss": 0.8598,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955939714432202e-05,
+      "loss": 0.8683,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557793748664187e-05,
+      "loss": 0.8432,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556187506809528e-05,
+      "loss": 0.9015,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955457841923636e-05,
+      "loss": 0.8232,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552966486423853e-05,
+      "loss": 0.7635,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955135170885202e-05,
+      "loss": 0.8734,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549734087001722e-05,
+      "loss": 0.9254,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548113621354668e-05,
+      "loss": 0.9459,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954649031239341e-05,
+      "loss": 0.907,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544864160601358e-05,
+      "loss": 0.8265,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543235166462752e-05,
+      "loss": 0.8129,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541603330462692e-05,
+      "loss": 0.8184,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953996865308712e-05,
+      "loss": 0.8875,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953833113482282e-05,
+      "loss": 0.904,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536690776157426e-05,
+      "loss": 0.913,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953504757757942e-05,
+      "loss": 0.8896,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533401539578122e-05,
+      "loss": 0.8751,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953175266264371e-05,
+      "loss": 0.8499,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530100947267192e-05,
+      "loss": 0.81,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952844639394043e-05,
+      "loss": 0.8867,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952678900315614e-05,
+      "loss": 0.8231,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952512877540786e-05,
+      "loss": 0.865,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523465711189992e-05,
+      "loss": 0.8469,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521799810997778e-05,
+      "loss": 0.8361,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95201310753273e-05,
+      "loss": 0.8223,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518459504675492e-05,
+      "loss": 0.8679,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516785099540122e-05,
+      "loss": 0.8607,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515107860419813e-05,
+      "loss": 0.8776,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951342778781402e-05,
+      "loss": 0.8005,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951174488222306e-05,
+      "loss": 0.9056,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951005914414807e-05,
+      "loss": 0.8293,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950837057409105e-05,
+      "loss": 0.8672,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506679172554835e-05,
+      "loss": 0.8988,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95049849400431e-05,
+      "loss": 0.8616,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950328787706037e-05,
+      "loss": 0.8314,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501587984112017e-05,
+      "loss": 0.9289,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499885261704238e-05,
+      "loss": 0.848,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498179710344088e-05,
+      "loss": 0.2709,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949647133053946e-05,
+      "loss": 0.8392,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494760122799085e-05,
+      "loss": 0.8218,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949304608763255e-05,
+      "loss": 0.791,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491329225550264e-05,
+      "loss": 0.8638,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489609537063492e-05,
+      "loss": 0.9304,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487887022684336e-05,
+      "loss": 0.846,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948616168292574e-05,
+      "loss": 0.9284,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484433518301496e-05,
+      "loss": 0.8274,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948270252932622e-05,
+      "loss": 0.7529,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480968716515386e-05,
+      "loss": 0.873,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479232080385306e-05,
+      "loss": 0.8938,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947749262145312e-05,
+      "loss": 0.8753,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947575034023683e-05,
+      "loss": 0.9002,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474005237255254e-05,
+      "loss": 0.8421,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472257313028072e-05,
+      "loss": 0.9253,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947050656807579e-05,
+      "loss": 0.8504,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468753002919764e-05,
+      "loss": 0.2704,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466996618082175e-05,
+      "loss": 0.8267,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465237414086065e-05,
+      "loss": 0.8666,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94634753914553e-05,
+      "loss": 0.8431,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461710550714582e-05,
+      "loss": 0.8288,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459942892389472e-05,
+      "loss": 0.9915,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.8778,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945639912509244e-05,
+      "loss": 0.8449,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454623017175814e-05,
+      "loss": 0.9061,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945284409378537e-05,
+      "loss": 0.8197,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451062355450854e-05,
+      "loss": 0.9054,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944927780270285e-05,
+      "loss": 0.2389,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447490436072764e-05,
+      "loss": 0.9125,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445700256092862e-05,
+      "loss": 0.8965,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944390726329624e-05,
+      "loss": 0.863,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944211145821682e-05,
+      "loss": 0.8596,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944031284138938e-05,
+      "loss": 0.8983,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438511413349524e-05,
+      "loss": 0.8138,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436707174633696e-05,
+      "loss": 0.8718,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434900125779178e-05,
+      "loss": 0.9237,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433090267324087e-05,
+      "loss": 0.8862,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431277599807373e-05,
+      "loss": 0.8243,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942946212376883e-05,
+      "loss": 0.7766,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427643839749085e-05,
+      "loss": 0.8255,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425822748289603e-05,
+      "loss": 0.8215,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942399884993268e-05,
+      "loss": 0.8264,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422172145221453e-05,
+      "loss": 0.8575,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420342634699893e-05,
+      "loss": 0.8863,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418510318912805e-05,
+      "loss": 0.8674,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416675198405835e-05,
+      "loss": 0.8859,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414837273725453e-05,
+      "loss": 0.8212,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412996545418976e-05,
+      "loss": 0.8283,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411153014034553e-05,
+      "loss": 0.8542,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409306680121155e-05,
+      "loss": 0.8083,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940745754422861e-05,
+      "loss": 0.8502,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405605606907563e-05,
+      "loss": 0.8921,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94037508687095e-05,
+      "loss": 0.8014,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401893330186744e-05,
+      "loss": 0.8448,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400032991892435e-05,
+      "loss": 0.7581,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398169854380574e-05,
+      "loss": 0.9158,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396303918205976e-05,
+      "loss": 0.8389,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394435183924288e-05,
+      "loss": 0.8946,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392563652092008e-05,
+      "loss": 0.8813,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939068932326645e-05,
+      "loss": 0.8189,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388812198005765e-05,
+      "loss": 0.8431,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938693227686894e-05,
+      "loss": 0.7661,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385049560415794e-05,
+      "loss": 0.8685,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383164049206976e-05,
+      "loss": 0.8607,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938127574380397e-05,
+      "loss": 0.8909,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937938464476909e-05,
+      "loss": 0.9228,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377490752665486e-05,
+      "loss": 0.8586,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375594068057134e-05,
+      "loss": 0.8208,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937369459150884e-05,
+      "loss": 0.908,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371792323586252e-05,
+      "loss": 0.8273,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936988726485584e-05,
+      "loss": 0.8284,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367979415884904e-05,
+      "loss": 0.7906,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366068777241586e-05,
+      "loss": 0.8502,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364155349494844e-05,
+      "loss": 0.8562,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936223913321448e-05,
+      "loss": 0.8517,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360320128971117e-05,
+      "loss": 0.8203,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358398337336213e-05,
+      "loss": 0.8928,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356473758882052e-05,
+      "loss": 0.819,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354546394181756e-05,
+      "loss": 0.8804,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352616243809264e-05,
+      "loss": 0.2722,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350683308339357e-05,
+      "loss": 0.9002,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348747588347637e-05,
+      "loss": 0.8296,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346809084410544e-05,
+      "loss": 0.8671,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344867797105334e-05,
+      "loss": 0.8822,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93429237270101e-05,
+      "loss": 0.885,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340976874703766e-05,
+      "loss": 0.8985,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339027240766082e-05,
+      "loss": 0.2583,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337074825777625e-05,
+      "loss": 0.8442,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335119630319797e-05,
+      "loss": 0.8647,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333161654974837e-05,
+      "loss": 0.8674,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331200900325803e-05,
+      "loss": 0.2612,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932923736695659e-05,
+      "loss": 0.848,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327271055451907e-05,
+      "loss": 0.7664,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93253019663973e-05,
+      "loss": 0.7635,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323330100379144e-05,
+      "loss": 0.7854,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321355457984638e-05,
+      "loss": 0.8716,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93193780398018e-05,
+      "loss": 0.8462,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317397846419486e-05,
+      "loss": 0.8886,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315414878427373e-05,
+      "loss": 0.8497,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313429136415967e-05,
+      "loss": 0.873,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311440620976597e-05,
+      "loss": 0.8825,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930944933270142e-05,
+      "loss": 0.8491,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307455272183414e-05,
+      "loss": 0.8669,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930545844001639e-05,
+      "loss": 0.8432,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303458836794983e-05,
+      "loss": 0.8681,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301456463114645e-05,
+      "loss": 0.8438,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299451319571663e-05,
+      "loss": 0.8895,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297443406763145e-05,
+      "loss": 0.8866,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295432725287017e-05,
+      "loss": 0.8667,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293419275742048e-05,
+      "loss": 0.8689,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291403058727807e-05,
+      "loss": 0.8305,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289384074844707e-05,
+      "loss": 0.8101,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287362324693977e-05,
+      "loss": 0.8432,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285337808877668e-05,
+      "loss": 0.8669,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283310527998655e-05,
+      "loss": 0.8232,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281280482660637e-05,
+      "loss": 0.9002,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279247673468143e-05,
+      "loss": 0.2818,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277212101026516e-05,
+      "loss": 0.7697,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275173765941927e-05,
+      "loss": 0.8651,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273132668821363e-05,
+      "loss": 0.8904,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271088810272645e-05,
+      "loss": 0.8847,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269042190904407e-05,
+      "loss": 0.8777,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266992811326102e-05,
+      "loss": 0.8126,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264940672148018e-05,
+      "loss": 0.8579,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926288577398126e-05,
+      "loss": 0.7763,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926082811743774e-05,
+      "loss": 0.7856,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258767703130217e-05,
+      "loss": 0.8771,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925670453167225e-05,
+      "loss": 0.2539,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254638603678228e-05,
+      "loss": 0.8951,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925256991976336e-05,
+      "loss": 0.8948,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250498480543678e-05,
+      "loss": 0.2446,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248424286636026e-05,
+      "loss": 0.8583,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924634733865808e-05,
+      "loss": 0.7876,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924426763722833e-05,
+      "loss": 0.8165,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242185182966082e-05,
+      "loss": 0.8009,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924009997649147e-05,
+      "loss": 0.7971,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238012018425437e-05,
+      "loss": 0.8676,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235921309389762e-05,
+      "loss": 0.8729,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233827850007028e-05,
+      "loss": 0.8867,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231731640900638e-05,
+      "loss": 0.9338,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229632682694824e-05,
+      "loss": 0.7302,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227530976014632e-05,
+      "loss": 0.8786,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225426521485922e-05,
+      "loss": 0.909,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223319319735373e-05,
+      "loss": 0.823,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922120937139049e-05,
+      "loss": 0.8172,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219096677079584e-05,
+      "loss": 0.2675,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216981237431796e-05,
+      "loss": 0.8535,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214863053077077e-05,
+      "loss": 0.7783,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212742124646192e-05,
+      "loss": 0.9028,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210618452770733e-05,
+      "loss": 0.841,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920849203808311e-05,
+      "loss": 0.8894,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920636288121653e-05,
+      "loss": 0.8555,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920423098280504e-05,
+      "loss": 0.8552,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202096343483494e-05,
+      "loss": 0.2106,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919995896388756e-05,
+      "loss": 0.8564,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919781884465372e-05,
+      "loss": 0.8202,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195675986419284e-05,
+      "loss": 0.8366,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193530389822364e-05,
+      "loss": 0.7743,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191382055501896e-05,
+      "loss": 0.8578,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189230984097628e-05,
+      "loss": 0.7733,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918707717625012e-05,
+      "loss": 0.7736,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918492063260075e-05,
+      "loss": 0.8822,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182761353791715e-05,
+      "loss": 0.871,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180599340466023e-05,
+      "loss": 0.8979,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178434593267496e-05,
+      "loss": 0.7453,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176267112840765e-05,
+      "loss": 0.8614,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917409689983128e-05,
+      "loss": 0.8623,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171923954885308e-05,
+      "loss": 0.8689,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169748278649928e-05,
+      "loss": 0.8833,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167569871773026e-05,
+      "loss": 0.9116,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165388734903307e-05,
+      "loss": 0.8466,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163204868690283e-05,
+      "loss": 0.8356,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161018273784292e-05,
+      "loss": 0.8898,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915882895083647e-05,
+      "loss": 0.8761,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156636900498772e-05,
+      "loss": 0.9008,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154442123423965e-05,
+      "loss": 0.8746,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915224462026563e-05,
+      "loss": 0.8526,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150044391678153e-05,
+      "loss": 0.8798,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914784143831674e-05,
+      "loss": 0.762,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145635760837398e-05,
+      "loss": 0.8203,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143427359896955e-05,
+      "loss": 0.2656,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914121623615305e-05,
+      "loss": 0.8064,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913900239026412e-05,
+      "loss": 0.8833,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136785822889434e-05,
+      "loss": 0.8345,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913456653468905e-05,
+      "loss": 0.8847,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913234452632385e-05,
+      "loss": 0.8155,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913011979845552e-05,
+      "loss": 0.888,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127892351746554e-05,
+      "loss": 0.8343,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125662186860267e-05,
+      "loss": 0.7694,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123429304460768e-05,
+      "loss": 0.8017,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121193705212988e-05,
+      "loss": 0.8417,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118955389782656e-05,
+      "loss": 0.7329,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911671435883632e-05,
+      "loss": 0.8432,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114470613041335e-05,
+      "loss": 0.8078,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112224153065856e-05,
+      "loss": 0.916,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109974979578852e-05,
+      "loss": 0.8058,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107723093250104e-05,
+      "loss": 0.259,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910546849475019e-05,
+      "loss": 0.8281,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910321118475051e-05,
+      "loss": 0.8174,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100951163923266e-05,
+      "loss": 0.2249,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098688432941454e-05,
+      "loss": 0.8443,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096422992478898e-05,
+      "loss": 0.8493,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909415484321021e-05,
+      "loss": 0.8551,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909188398581083e-05,
+      "loss": 0.9091,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908961042095699e-05,
+      "loss": 0.822,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087334149325717e-05,
+      "loss": 0.8951,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908505517159487e-05,
+      "loss": 0.9024,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082773488443097e-05,
+      "loss": 0.8353,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908048910054986e-05,
+      "loss": 0.8308,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078202008595418e-05,
+      "loss": 0.8106,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907591221326084e-05,
+      "loss": 0.9077,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073619715228e-05,
+      "loss": 0.9083,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907132451517958e-05,
+      "loss": 0.8992,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906902661379906e-05,
+      "loss": 0.8249,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066726011770725e-05,
+      "loss": 0.8384,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064422709779674e-05,
+      "loss": 0.7847,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90621167085118e-05,
+      "loss": 0.9226,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90598080086538e-05,
+      "loss": 0.7439,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057496610893173e-05,
+      "loss": 0.8762,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055182515918238e-05,
+      "loss": 0.8132,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052865724418094e-05,
+      "loss": 0.9476,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905054623708266e-05,
+      "loss": 0.8387,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904822405460265e-05,
+      "loss": 0.8152,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045899177669583e-05,
+      "loss": 0.936,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043571606975776e-05,
+      "loss": 0.8462,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041241343214354e-05,
+      "loss": 0.814,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038908387079244e-05,
+      "loss": 0.2298,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036572739265172e-05,
+      "loss": 0.299,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034234400467658e-05,
+      "loss": 0.8017,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031893371383043e-05,
+      "loss": 0.7782,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902954965270845e-05,
+      "loss": 0.8753,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027203245141816e-05,
+      "loss": 0.8398,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024854149381865e-05,
+      "loss": 0.808,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022502366128136e-05,
+      "loss": 0.8202,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020147896080962e-05,
+      "loss": 0.8338,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017790739941473e-05,
+      "loss": 0.8434,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015430898411604e-05,
+      "loss": 0.7829,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013068372194085e-05,
+      "loss": 0.8256,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010703161992455e-05,
+      "loss": 0.8763,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008335268511038e-05,
+      "loss": 0.758,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005964692454966e-05,
+      "loss": 0.8,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900359143453017e-05,
+      "loss": 0.8531,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001215495443375e-05,
+      "loss": 0.8793,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899883687590211e-05,
+      "loss": 0.858,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996455576614704e-05,
+      "loss": 0.853,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899407159829027e-05,
+      "loss": 0.8021,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991684941638734e-05,
+      "loss": 0.8523,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898929560737081e-05,
+      "loss": 0.886,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986903596198015e-05,
+      "loss": 0.8868,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984508908832664e-05,
+      "loss": 0.8318,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982111545987865e-05,
+      "loss": 0.7826,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979711508377524e-05,
+      "loss": 0.8176,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897730879671634e-05,
+      "loss": 0.3306,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974903411719817e-05,
+      "loss": 0.8261,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972495354104246e-05,
+      "loss": 0.8076,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897008462458672e-05,
+      "loss": 0.8498,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967671223885125e-05,
+      "loss": 0.8252,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965255152718143e-05,
+      "loss": 0.8233,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962836411805246e-05,
+      "loss": 0.7927,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960415001866718e-05,
+      "loss": 0.8157,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957990923623618e-05,
+      "loss": 0.9061,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955564177797806e-05,
+      "loss": 0.8927,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953134765111942e-05,
+      "loss": 0.8454,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950702686289472e-05,
+      "loss": 0.8645,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948267942054647e-05,
+      "loss": 0.8214,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945830533132498e-05,
+      "loss": 0.8085,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894339046024886e-05,
+      "loss": 0.7992,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940947724130353e-05,
+      "loss": 0.8404,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89385023255044e-05,
+      "loss": 0.823,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893605426509921e-05,
+      "loss": 0.858,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933603543643783e-05,
+      "loss": 0.7844,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931150161867917e-05,
+      "loss": 0.8856,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89286941205022e-05,
+      "loss": 0.8902,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926235420278013e-05,
+      "loss": 0.7974,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923774061927526e-05,
+      "loss": 0.8182,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921310046183704e-05,
+      "loss": 0.8633,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89188433737803e-05,
+      "loss": 0.8152,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916374045451855e-05,
+      "loss": 0.3075,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891390206193371e-05,
+      "loss": 0.8649,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911427423961996e-05,
+      "loss": 0.8641,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908950132273625e-05,
+      "loss": 0.7666,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906470187606306e-05,
+      "loss": 0.9311,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890398759069854e-05,
+      "loss": 0.2863,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890150234228961e-05,
+      "loss": 0.8236,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899014443119596e-05,
+      "loss": 0.7897,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889652389392936e-05,
+      "loss": 0.8672,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894030695460566e-05,
+      "loss": 0.8293,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891534848455654e-05,
+      "loss": 0.8119,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889036353657858e-05,
+      "loss": 0.8775,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886535211811202e-05,
+      "loss": 0.7978,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884031423660492e-05,
+      "loss": 0.8633,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881524989951327e-05,
+      "loss": 0.9463,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879015911430095e-05,
+      "loss": 0.8217,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876504188843972e-05,
+      "loss": 0.8349,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873989822940914e-05,
+      "loss": 0.7946,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871472814469667e-05,
+      "loss": 0.8375,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868953164179776e-05,
+      "loss": 0.8615,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866430872821555e-05,
+      "loss": 0.8113,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863905941146115e-05,
+      "loss": 0.7767,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861378369905348e-05,
+      "loss": 0.8193,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858848159851937e-05,
+      "loss": 0.841,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885631531173935e-05,
+      "loss": 0.9219,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885377982632183e-05,
+      "loss": 0.8425,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851241704354427e-05,
+      "loss": 0.9065,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848700946592954e-05,
+      "loss": 0.8964,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884615755379402e-05,
+      "loss": 0.8422,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884361152671502e-05,
+      "loss": 0.8606,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884106286611413e-05,
+      "loss": 0.8028,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883851157275031e-05,
+      "loss": 0.8843,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835957647383304e-05,
+      "loss": 0.7473,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883340109077364e-05,
+      "loss": 0.8761,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830841903682627e-05,
+      "loss": 0.8901,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882828008687237e-05,
+      "loss": 0.8312,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882571564110574e-05,
+      "loss": 0.8775,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823148567146398e-05,
+      "loss": 0.8462,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820578865758793e-05,
+      "loss": 0.8485,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818006537708148e-05,
+      "loss": 0.8448,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815431583760468e-05,
+      "loss": 0.8169,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881285400468255e-05,
+      "loss": 0.8471,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881027380124196e-05,
+      "loss": 0.7819,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807690974207062e-05,
+      "loss": 0.8355,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880510552434698e-05,
+      "loss": 0.8199,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802517452431635e-05,
+      "loss": 0.7151,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799926759231725e-05,
+      "loss": 0.8763,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797333445518724e-05,
+      "loss": 0.8253,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.2786,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879213895964327e-05,
+      "loss": 0.8543,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878953778902767e-05,
+      "loss": 0.8276,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878693400099269e-05,
+      "loss": 0.8028,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878432759631371e-05,
+      "loss": 0.8307,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781718575766888e-05,
+      "loss": 0.8339,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779106940129154e-05,
+      "loss": 0.8951,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776492690178228e-05,
+      "loss": 0.834,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773875826692594e-05,
+      "loss": 0.2847,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877125635045153e-05,
+      "loss": 0.8575,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876863426223508e-05,
+      "loss": 0.843,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766009562824073e-05,
+      "loss": 0.2425,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763382253000117e-05,
+      "loss": 0.8232,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760752333545586e-05,
+      "loss": 0.9056,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758119805243643e-05,
+      "loss": 0.9045,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755484668878225e-05,
+      "loss": 0.8139,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752846925234038e-05,
+      "loss": 0.8971,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750206575096576e-05,
+      "loss": 0.8422,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747563619252102e-05,
+      "loss": 0.7959,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744918058487655e-05,
+      "loss": 0.8232,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742269893591052e-05,
+      "loss": 0.8831,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873961912535089e-05,
+      "loss": 0.8708,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736965754556527e-05,
+      "loss": 0.8649,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734309781998116e-05,
+      "loss": 0.7822,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731651208466563e-05,
+      "loss": 0.7784,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728990034753568e-05,
+      "loss": 0.7868,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872632626165159e-05,
+      "loss": 0.8125,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723659889953875e-05,
+      "loss": 0.7426,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720990920454432e-05,
+      "loss": 0.8327,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871831935394805e-05,
+      "loss": 0.8577,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871564519123029e-05,
+      "loss": 0.8272,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712968433097485e-05,
+      "loss": 0.8328,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710289080346743e-05,
+      "loss": 0.7978,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870760713377594e-05,
+      "loss": 0.8327,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870492259418373e-05,
+      "loss": 0.865,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702235462369537e-05,
+      "loss": 0.8151,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699545739133556e-05,
+      "loss": 0.9383,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869685342527676e-05,
+      "loss": 0.8156,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694158521600876e-05,
+      "loss": 0.8015,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691461028908424e-05,
+      "loss": 0.8794,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868876094800268e-05,
+      "loss": 0.8155,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86860582796877e-05,
+      "loss": 0.8563,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683353024768303e-05,
+      "loss": 0.8161,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680645184050083e-05,
+      "loss": 0.8383,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86779347583394e-05,
+      "loss": 0.8022,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675221748443392e-05,
+      "loss": 0.8553,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867250615516996e-05,
+      "loss": 0.8154,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866978797932777e-05,
+      "loss": 0.8882,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667067221726265e-05,
+      "loss": 0.9043,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664343883175655e-05,
+      "loss": 0.8019,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661617964486914e-05,
+      "loss": 0.8305,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658889466471793e-05,
+      "loss": 0.8708,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656158389942806e-05,
+      "loss": 0.8941,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653424735713234e-05,
+      "loss": 0.8067,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865068850459712e-05,
+      "loss": 0.2845,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647949697409292e-05,
+      "loss": 0.9277,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864520831496533e-05,
+      "loss": 0.813,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864246435808158e-05,
+      "loss": 0.8464,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639717827575164e-05,
+      "loss": 0.9276,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636968724263965e-05,
+      "loss": 0.8109,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634217048966638e-05,
+      "loss": 0.7716,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863146280250259e-05,
+      "loss": 0.7971,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628705985692007e-05,
+      "loss": 0.8294,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625946599355838e-05,
+      "loss": 0.8244,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623184644315792e-05,
+      "loss": 0.8076,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862042012139435e-05,
+      "loss": 0.8068,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617653031414752e-05,
+      "loss": 0.895,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614883375201003e-05,
+      "loss": 0.8417,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612111153577876e-05,
+      "loss": 0.8715,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86093363673709e-05,
+      "loss": 0.8232,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860655901740638e-05,
+      "loss": 0.8435,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603779104511376e-05,
+      "loss": 0.297,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600996629513713e-05,
+      "loss": 0.7831,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598211593241972e-05,
+      "loss": 0.2504,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595423996525515e-05,
+      "loss": 0.8249,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592633840194446e-05,
+      "loss": 0.8582,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589841125079643e-05,
+      "loss": 0.817,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587045852012745e-05,
+      "loss": 0.8536,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584248021826154e-05,
+      "loss": 0.7948,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858144763535302e-05,
+      "loss": 0.8504,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578644693427272e-05,
+      "loss": 0.8748,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575839196883592e-05,
+      "loss": 0.7541,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573031146557423e-05,
+      "loss": 0.8335,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570220543284966e-05,
+      "loss": 0.7839,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856740738790319e-05,
+      "loss": 0.8022,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564591681249816e-05,
+      "loss": 0.8539,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856177342416333e-05,
+      "loss": 0.8527,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855895261748297e-05,
+      "loss": 0.8342,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556129262048744e-05,
+      "loss": 0.8085,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553303358701413e-05,
+      "loss": 0.8322,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550474908282496e-05,
+      "loss": 0.8478,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854764391163427e-05,
+      "loss": 0.877,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544810369599777e-05,
+      "loss": 0.843,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541974283022807e-05,
+      "loss": 0.8563,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539135652747915e-05,
+      "loss": 0.7965,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536294479620407e-05,
+      "loss": 0.8613,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853345076448636e-05,
+      "loss": 0.7546,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530604508192588e-05,
+      "loss": 0.824,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852775571158668e-05,
+      "loss": 0.8309,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524904375516967e-05,
+      "loss": 0.8485,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852205050083255e-05,
+      "loss": 0.7999,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851919408838327e-05,
+      "loss": 0.9086,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516335139019743e-05,
+      "loss": 0.8919,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851347365359332e-05,
+      "loss": 0.8142,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510609632956124e-05,
+      "loss": 0.8599,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507743077961024e-05,
+      "loss": 0.8848,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504873989461645e-05,
+      "loss": 0.8591,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502002368312367e-05,
+      "loss": 0.9288,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499128215368326e-05,
+      "loss": 0.8765,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849625153148541e-05,
+      "loss": 0.8304,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849337231752026e-05,
+      "loss": 0.805,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490490574330274e-05,
+      "loss": 0.8146,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487606302773596e-05,
+      "loss": 0.8574,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484719503709133e-05,
+      "loss": 0.8671,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481830177996535e-05,
+      "loss": 0.7751,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478938326496213e-05,
+      "loss": 0.8585,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476043950069323e-05,
+      "loss": 0.8798,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473147049577777e-05,
+      "loss": 0.8584,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470247625884236e-05,
+      "loss": 0.8404,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467345679852115e-05,
+      "loss": 0.8825,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464441212345577e-05,
+      "loss": 0.866,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846153422422954e-05,
+      "loss": 0.9068,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458624716369667e-05,
+      "loss": 0.8162,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455712689632375e-05,
+      "loss": 0.8497,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452798144884842e-05,
+      "loss": 0.8458,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844988108299497e-05,
+      "loss": 0.8206,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446961504831428e-05,
+      "loss": 0.7913,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444039411263635e-05,
+      "loss": 0.7978,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441114803161757e-05,
+      "loss": 0.8527,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8438187681396705e-05,
+      "loss": 0.7932,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843525804684014e-05,
+      "loss": 0.8834,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432325900364474e-05,
+      "loss": 0.8737,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429391242842864e-05,
+      "loss": 0.8312,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426454075149215e-05,
+      "loss": 0.6921,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423514398158183e-05,
+      "loss": 0.7831,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420572212745167e-05,
+      "loss": 0.292,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417627519786317e-05,
+      "loss": 0.7814,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414680320158522e-05,
+      "loss": 0.7962,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841173061473943e-05,
+      "loss": 0.7982,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408778404407423e-05,
+      "loss": 0.9168,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405823690041634e-05,
+      "loss": 0.8231,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402866472521944e-05,
+      "loss": 0.837,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399906752728974e-05,
+      "loss": 0.8606,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396944531544098e-05,
+      "loss": 0.7878,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393979809849427e-05,
+      "loss": 0.8673,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391012588527823e-05,
+      "loss": 0.886,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838804286846288e-05,
+      "loss": 0.935,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385070650538955e-05,
+      "loss": 0.8018,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382095935641136e-05,
+      "loss": 0.8439,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379118724655254e-05,
+      "loss": 0.9162,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376139018467895e-05,
+      "loss": 0.8819,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837315681796637e-05,
+      "loss": 0.8955,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837017212403875e-05,
+      "loss": 0.8446,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367184937573833e-05,
+      "loss": 0.7946,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836419525946118e-05,
+      "loss": 0.8904,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836120309059107e-05,
+      "loss": 0.7901,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835820843185454e-05,
+      "loss": 0.8964,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355211284143365e-05,
+      "loss": 0.8595,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835221164835006e-05,
+      "loss": 0.2879,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349209525367878e-05,
+      "loss": 0.7854,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346204916090813e-05,
+      "loss": 0.912,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343197821413605e-05,
+      "loss": 0.8261,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340188242231733e-05,
+      "loss": 0.914,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337176179441408e-05,
+      "loss": 0.8146,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334161633939595e-05,
+      "loss": 0.7964,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331144606623982e-05,
+      "loss": 0.8371,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328125098393007e-05,
+      "loss": 0.8183,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832510311014584e-05,
+      "loss": 0.8356,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322078642782396e-05,
+      "loss": 0.6955,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831905169720333e-05,
+      "loss": 0.85,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831602227431002e-05,
+      "loss": 0.8504,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312990375004598e-05,
+      "loss": 0.8552,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309956000189927e-05,
+      "loss": 0.7764,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830691915076961e-05,
+      "loss": 0.7313,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303879827647977e-05,
+      "loss": 0.739,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300838031730108e-05,
+      "loss": 0.7898,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297793763921814e-05,
+      "loss": 0.8638,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829474702512964e-05,
+      "loss": 0.9214,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291697816260865e-05,
+      "loss": 0.8006,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828864613822351e-05,
+      "loss": 0.8759,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285591991926334e-05,
+      "loss": 0.8157,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282535378278815e-05,
+      "loss": 0.8066,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279476298191184e-05,
+      "loss": 0.7647,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276414752574393e-05,
+      "loss": 0.8383,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273350742340136e-05,
+      "loss": 0.8346,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270284268400837e-05,
+      "loss": 0.8173,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826721533166966e-05,
+      "loss": 0.8241,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264143933060494e-05,
+      "loss": 0.8092,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261070073487967e-05,
+      "loss": 0.8909,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257993753867432e-05,
+      "loss": 0.8353,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254914975114987e-05,
+      "loss": 0.8183,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251833738147455e-05,
+      "loss": 0.8255,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824875004388239e-05,
+      "loss": 0.8107,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245663893238075e-05,
+      "loss": 0.7725,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242575287133536e-05,
+      "loss": 0.8894,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239484226488516e-05,
+      "loss": 0.2655,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236390712223503e-05,
+      "loss": 0.8293,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233294745259703e-05,
+      "loss": 0.8512,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823019632651906e-05,
+      "loss": 0.8535,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822709545692425e-05,
+      "loss": 0.8609,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223992137398666e-05,
+      "loss": 0.822,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822088636886645e-05,
+      "loss": 0.8455,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821777815225245e-05,
+      "loss": 0.7826,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821466748848227e-05,
+      "loss": 0.8296,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211554378482223e-05,
+      "loss": 0.9072,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820843882317935e-05,
+      "loss": 0.8136,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205320823501435e-05,
+      "loss": 0.8153,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202200380376974e-05,
+      "loss": 0.8684,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81990774947352e-05,
+      "loss": 0.821,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195952167506073e-05,
+      "loss": 0.8898,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819282439962028e-05,
+      "loss": 0.8728,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189694192009227e-05,
+      "loss": 0.8129,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186561545605055e-05,
+      "loss": 0.8472,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818342646134063e-05,
+      "loss": 0.8842,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180288940149542e-05,
+      "loss": 0.8157,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177148982966106e-05,
+      "loss": 0.8756,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174006590725364e-05,
+      "loss": 0.8524,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170861764363086e-05,
+      "loss": 0.753,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816771450481576e-05,
+      "loss": 0.8325,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81645648130206e-05,
+      "loss": 0.8606,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161412689915554e-05,
+      "loss": 0.85,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815825813643928e-05,
+      "loss": 0.8644,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155101153531174e-05,
+      "loss": 0.7886,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815194174213134e-05,
+      "loss": 0.2464,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148779903180614e-05,
+      "loss": 0.8703,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814561563762056e-05,
+      "loss": 0.8261,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142448946393455e-05,
+      "loss": 0.8368,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139279830442296e-05,
+      "loss": 0.7845,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136108290710814e-05,
+      "loss": 0.9344,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132934328143457e-05,
+      "loss": 0.8872,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812975794368539e-05,
+      "loss": 0.8378,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126579138282502e-05,
+      "loss": 0.849,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123397912881406e-05,
+      "loss": 0.8081,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120214268429435e-05,
+      "loss": 0.8305,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811702820587463e-05,
+      "loss": 0.8582,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113839726165773e-05,
+      "loss": 0.7501,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110648830252347e-05,
+      "loss": 0.8097,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810745551908457e-05,
+      "loss": 0.8674,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810425979361337e-05,
+      "loss": 0.9013,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101061654790395e-05,
+      "loss": 0.8545,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809786110356801e-05,
+      "loss": 0.8029,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094658140899306e-05,
+      "loss": 0.866,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809145276773808e-05,
+      "loss": 0.8288,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088244985038864e-05,
+      "loss": 0.8179,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085034793756886e-05,
+      "loss": 0.8409,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808182219484811e-05,
+      "loss": 0.2776,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078607189269206e-05,
+      "loss": 0.8331,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807538977797757e-05,
+      "loss": 0.7567,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8072169961931304e-05,
+      "loss": 0.8479,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068947742089225e-05,
+      "loss": 0.2768,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065723119410885e-05,
+      "loss": 0.8457,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062496094856527e-05,
+      "loss": 0.8479,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059266669387128e-05,
+      "loss": 0.7889,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056034843964365e-05,
+      "loss": 0.9217,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052800619550643e-05,
+      "loss": 0.9121,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049563997109076e-05,
+      "loss": 0.8426,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046324977603482e-05,
+      "loss": 0.8169,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043083561998412e-05,
+      "loss": 0.7372,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039839751259118e-05,
+      "loss": 0.8571,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803659354635157e-05,
+      "loss": 0.8037,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033344948242443e-05,
+      "loss": 0.7874,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030093957899138e-05,
+      "loss": 0.8135,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026840576289756e-05,
+      "loss": 0.8043,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023584804383118e-05,
+      "loss": 0.7829,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020326643148754e-05,
+      "loss": 0.8265,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017066093556904e-05,
+      "loss": 0.7964,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801380315657852e-05,
+      "loss": 0.8215,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010537833185267e-05,
+      "loss": 0.8703,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800727012434952e-05,
+      "loss": 0.7525,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004000031044363e-05,
+      "loss": 0.8451,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000727554243587e-05,
+      "loss": 0.8523,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997452694921707e-05,
+      "loss": 0.8697,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994175454053925e-05,
+      "loss": 0.8451,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990895832616173e-05,
+      "loss": 0.804,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987613831585075e-05,
+      "loss": 0.7716,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798432945193798e-05,
+      "loss": 0.7609,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798104269465293e-05,
+      "loss": 0.8023,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977753560708686e-05,
+      "loss": 0.8288,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974462051084714e-05,
+      "loss": 0.9192,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971168166761184e-05,
+      "loss": 0.7088,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967871908718974e-05,
+      "loss": 0.793,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964573277939672e-05,
+      "loss": 0.9313,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796127227540557e-05,
+      "loss": 0.8845,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795796890209967e-05,
+      "loss": 0.8293,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954663159005674e-05,
+      "loss": 0.8227,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795135504710799e-05,
+      "loss": 0.7842,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948044567391747e-05,
+      "loss": 0.8054,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794473172084275e-05,
+      "loss": 0.7911,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941416508447537e-05,
+      "loss": 0.8577,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938098931193333e-05,
+      "loss": 0.8211,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793477899006807e-05,
+      "loss": 0.8687,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931456686060396e-05,
+      "loss": 0.7403,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928132020159648e-05,
+      "loss": 0.8928,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792480499335587e-05,
+      "loss": 0.8253,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921475606639812e-05,
+      "loss": 0.7882,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918143861002928e-05,
+      "loss": 0.9118,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791480975743737e-05,
+      "loss": 0.8808,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911473296935998e-05,
+      "loss": 0.7872,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908134480492364e-05,
+      "loss": 0.7681,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904793309100733e-05,
+      "loss": 0.8068,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790144978375606e-05,
+      "loss": 0.8556,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789810390545401e-05,
+      "loss": 0.8724,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789475567519095e-05,
+      "loss": 0.8466,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.8497,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888052162770738e-05,
+      "loss": 0.8767,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884696882609812e-05,
+      "loss": 0.8359,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881339254480323e-05,
+      "loss": 0.8046,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7877979279382135e-05,
+      "loss": 0.8554,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874616958315804e-05,
+      "loss": 0.8632,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787125229228259e-05,
+      "loss": 0.8299,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867885282284458e-05,
+      "loss": 0.7959,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786451592932405e-05,
+      "loss": 0.8091,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786114423440473e-05,
+      "loss": 0.8025,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857770198530545e-05,
+      "loss": 0.8158,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854393822706243e-05,
+      "loss": 0.862,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851015107937266e-05,
+      "loss": 0.8729,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847634055229758e-05,
+      "loss": 0.8227,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784425066559055e-05,
+      "loss": 0.8484,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840864940027186e-05,
+      "loss": 0.8652,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837476879547883e-05,
+      "loss": 0.7753,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834086485161574e-05,
+      "loss": 0.8243,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830693757877874e-05,
+      "loss": 0.8916,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827298698707093e-05,
+      "loss": 0.2856,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823901308660246e-05,
+      "loss": 0.8189,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820501588749033e-05,
+      "loss": 0.7871,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817099539985848e-05,
+      "loss": 0.8304,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781369516338378e-05,
+      "loss": 0.8401,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781028845995662e-05,
+      "loss": 0.8228,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780687943071883e-05,
+      "loss": 0.8009,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803468076685585e-05,
+      "loss": 0.7816,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800054398872748e-05,
+      "loss": 0.2616,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779663839829687e-05,
+      "loss": 0.8787,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793220075975192e-05,
+      "loss": 0.784,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789799432925655e-05,
+      "loss": 0.9015,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786376470166878e-05,
+      "loss": 0.886,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782951188718187e-05,
+      "loss": 0.8819,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779523589599585e-05,
+      "loss": 0.7886,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777609367383177e-05,
+      "loss": 0.8446,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777266144243613e-05,
+      "loss": 0.8564,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769226896434745e-05,
+      "loss": 0.7726,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776579003685038e-05,
+      "loss": 0.8406,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762350864706487e-05,
+      "loss": 0.8306,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775890938102722e-05,
+      "loss": 0.8063,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755465586837403e-05,
+      "loss": 0.8215,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775201948316256e-05,
+      "loss": 0.2592,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77485710710289e-05,
+      "loss": 0.783,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774512035146332e-05,
+      "loss": 0.7748,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741667325493396e-05,
+      "loss": 0.8417,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738211994147407e-05,
+      "loss": 0.7864,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734754358454303e-05,
+      "loss": 0.8678,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731294419443734e-05,
+      "loss": 0.7736,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727832178146022e-05,
+      "loss": 0.833,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772436763559218e-05,
+      "loss": 0.7968,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720900792813913e-05,
+      "loss": 0.8366,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77174316508436e-05,
+      "loss": 0.2768,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713960210714317e-05,
+      "loss": 0.8299,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771048647345981e-05,
+      "loss": 0.8571,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707010440114517e-05,
+      "loss": 0.847,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703532111713568e-05,
+      "loss": 0.7929,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700051489292756e-05,
+      "loss": 0.7627,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696568573888578e-05,
+      "loss": 0.2429,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76930833665382e-05,
+      "loss": 0.8263,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768959586827947e-05,
+      "loss": 0.8782,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686106080150936e-05,
+      "loss": 0.8666,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682614003191807e-05,
+      "loss": 0.9096,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679119638441984e-05,
+      "loss": 0.2547,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675622986942045e-05,
+      "loss": 0.7747,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672124049733254e-05,
+      "loss": 0.832,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766862282785755e-05,
+      "loss": 0.8176,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665119322357555e-05,
+      "loss": 0.8112,
+      "step": 1457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661613534276575e-05,
+      "loss": 0.7683,
+      "step": 1458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658105464658585e-05,
+      "loss": 0.751,
+      "step": 1459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654595114548255e-05,
+      "loss": 0.8356,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651082484990918e-05,
+      "loss": 0.812,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647567577032596e-05,
+      "loss": 0.8421,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644050391719988e-05,
+      "loss": 0.9043,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640530930100466e-05,
+      "loss": 0.8754,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637009193222088e-05,
+      "loss": 0.8575,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633485182133584e-05,
+      "loss": 0.8656,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629958897884358e-05,
+      "loss": 0.832,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76264303415245e-05,
+      "loss": 0.8298,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762289951410477e-05,
+      "loss": 0.8581,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619366416676608e-05,
+      "loss": 0.8146,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761583105029213e-05,
+      "loss": 0.8711,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761229341600412e-05,
+      "loss": 0.7727,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608753514866048e-05,
+      "loss": 0.8264,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760521134793205e-05,
+      "loss": 0.7936,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601666916256947e-05,
+      "loss": 0.8518,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759812022089622e-05,
+      "loss": 0.8476,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759457126290604e-05,
+      "loss": 0.9086,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591020043343245e-05,
+      "loss": 0.8082,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587466563265337e-05,
+      "loss": 0.7973,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758391082373051e-05,
+      "loss": 0.7842,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580352825797615e-05,
+      "loss": 0.8335,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757679257052618e-05,
+      "loss": 0.8084,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573230058976412e-05,
+      "loss": 0.2419,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569665292209182e-05,
+      "loss": 0.7549,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566098271286034e-05,
+      "loss": 0.8073,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562528997269182e-05,
+      "loss": 0.7954,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755895747122152e-05,
+      "loss": 0.8156,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75553836942066e-05,
+      "loss": 0.8314,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551807667288652e-05,
+      "loss": 0.7557,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548229391532572e-05,
+      "loss": 0.7939,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754464886800393e-05,
+      "loss": 0.8692,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541066097768965e-05,
+      "loss": 0.8263,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537481081894578e-05,
+      "loss": 0.8002,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753389382144835e-05,
+      "loss": 0.8552,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530304317498518e-05,
+      "loss": 0.787,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526712571114e-05,
+      "loss": 0.9631,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752311858336437e-05,
+      "loss": 0.8812,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519522355319877e-05,
+      "loss": 0.8461,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515923888051435e-05,
+      "loss": 0.8496,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751232318263062e-05,
+      "loss": 0.776,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750872024012969e-05,
+      "loss": 0.8309,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505115061621546e-05,
+      "loss": 0.7405,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501507648179773e-05,
+      "loss": 0.8175,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497898000878612e-05,
+      "loss": 0.6991,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494286120792978e-05,
+      "loss": 0.2771,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490672008998443e-05,
+      "loss": 0.8559,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487055666571247e-05,
+      "loss": 0.784,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483437094588295e-05,
+      "loss": 0.829,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747981629412715e-05,
+      "loss": 0.8653,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747619326626605e-05,
+      "loss": 0.8575,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472568012083882e-05,
+      "loss": 0.7278,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468940532660207e-05,
+      "loss": 0.9385,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465310829075247e-05,
+      "loss": 0.7775,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746167890240988e-05,
+      "loss": 0.7256,
+      "step": 1514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458044753745653e-05,
+      "loss": 0.8163,
+      "step": 1515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454408384164775e-05,
+      "loss": 0.8576,
+      "step": 1516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450769794750107e-05,
+      "loss": 0.904,
+      "step": 1517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447128986585184e-05,
+      "loss": 0.8374,
+      "step": 1518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443485960754184e-05,
+      "loss": 0.9048,
+      "step": 1519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439840718341973e-05,
+      "loss": 0.8779,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436193260434047e-05,
+      "loss": 0.879,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743254358811658e-05,
+      "loss": 0.8286,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428891702476404e-05,
+      "loss": 0.833,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425237604601003e-05,
+      "loss": 0.8766,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421581295578524e-05,
+      "loss": 0.7393,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417922776497774e-05,
+      "loss": 0.8142,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414262048448217e-05,
+      "loss": 0.8163,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741059911251997e-05,
+      "loss": 0.7958,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406933969803815e-05,
+      "loss": 0.8115,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403266621391187e-05,
+      "loss": 0.7473,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739959706837418e-05,
+      "loss": 0.8162,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395925311845536e-05,
+      "loss": 0.7944,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392251352898668e-05,
+      "loss": 0.7285,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388575192627635e-05,
+      "loss": 0.7845,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384896832127155e-05,
+      "loss": 0.8312,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381216272492596e-05,
+      "loss": 0.2773,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737753351481999e-05,
+      "loss": 0.7845,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737384856020601e-05,
+      "loss": 0.851,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370161409748002e-05,
+      "loss": 0.8219,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366472064543945e-05,
+      "loss": 0.892,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736278052569249e-05,
+      "loss": 0.8163,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359086794292928e-05,
+      "loss": 0.8569,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355390871445212e-05,
+      "loss": 0.8967,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351692758249942e-05,
+      "loss": 0.7678,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734799245580837e-05,
+      "loss": 0.8248,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344289965222405e-05,
+      "loss": 0.8151,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340585287594605e-05,
+      "loss": 0.7614,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336878424028173e-05,
+      "loss": 0.8058,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333169375626973e-05,
+      "loss": 0.81,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329458143495515e-05,
+      "loss": 0.8395,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732574472873896e-05,
+      "loss": 0.7603,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732202913246312e-05,
+      "loss": 0.7459,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731831135577445e-05,
+      "loss": 0.8778,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314591399780067e-05,
+      "loss": 0.7864,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310869265587722e-05,
+      "loss": 0.7712,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307144954305827e-05,
+      "loss": 0.7669,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303418467043434e-05,
+      "loss": 0.7999,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729968980491025e-05,
+      "loss": 0.867,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295958969016622e-05,
+      "loss": 0.8529,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729222596047355e-05,
+      "loss": 0.7379,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288490780392683e-05,
+      "loss": 0.7529,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728475342988631e-05,
+      "loss": 0.8318,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281013910067365e-05,
+      "loss": 0.8547,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727727222204944e-05,
+      "loss": 0.799,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727352836694676e-05,
+      "loss": 0.8484,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269782345874204e-05,
+      "loss": 0.834,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266034159947288e-05,
+      "loss": 0.791,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262283810282182e-05,
+      "loss": 0.8179,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258531297995694e-05,
+      "loss": 0.2552,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725477662420527e-05,
+      "loss": 0.838,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725101979002902e-05,
+      "loss": 0.8342,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724726079658567e-05,
+      "loss": 0.7543,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243499644994616e-05,
+      "loss": 0.7994,
+      "step": 1573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239736336375878e-05,
+      "loss": 0.7767,
+      "step": 1574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235970871850122e-05,
+      "loss": 0.8145,
+      "step": 1575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232203252538664e-05,
+      "loss": 0.7025,
+      "step": 1576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228433479563454e-05,
+      "loss": 0.8348,
+      "step": 1577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722466155404708e-05,
+      "loss": 0.7912,
+      "step": 1578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220887477112786e-05,
+      "loss": 0.856,
+      "step": 1579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721711124988444e-05,
+      "loss": 0.2708,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213332873486563e-05,
+      "loss": 0.8113,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209552349044304e-05,
+      "loss": 0.8546,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205769677683464e-05,
+      "loss": 0.8751,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201984860530473e-05,
+      "loss": 0.8524,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198197898712402e-05,
+      "loss": 0.8922,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194408793356972e-05,
+      "loss": 0.8698,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190617545592526e-05,
+      "loss": 0.7914,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186824156548052e-05,
+      "loss": 0.808,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183028627353178e-05,
+      "loss": 0.835,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179230959138173e-05,
+      "loss": 0.8106,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175431153033927e-05,
+      "loss": 0.8844,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171629210171984e-05,
+      "loss": 0.8648,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167825131684516e-05,
+      "loss": 0.8327,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164018918704325e-05,
+      "loss": 0.8778,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160210572364868e-05,
+      "loss": 0.8375,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156400093800216e-05,
+      "loss": 0.755,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152587484145086e-05,
+      "loss": 0.8487,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148772744534833e-05,
+      "loss": 0.8437,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714495587610543e-05,
+      "loss": 0.2689,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141136879993506e-05,
+      "loss": 0.8392,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713731575733631e-05,
+      "loss": 0.8266,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133492509271715e-05,
+      "loss": 0.8185,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129667136938254e-05,
+      "loss": 0.8192,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125839641475074e-05,
+      "loss": 0.7623,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712201002402195e-05,
+      "loss": 0.8345,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71181782857193e-05,
+      "loss": 0.8457,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114344427708172e-05,
+      "loss": 0.2552,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110508451130243e-05,
+      "loss": 0.7379,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106670357127818e-05,
+      "loss": 0.8108,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102830146843838e-05,
+      "loss": 0.8157,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709898782142187e-05,
+      "loss": 0.8274,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095143382006116e-05,
+      "loss": 0.8284,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091296829741402e-05,
+      "loss": 0.8192,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087448165773187e-05,
+      "loss": 0.818,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083597391247553e-05,
+      "loss": 0.8323,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079744507311216e-05,
+      "loss": 0.819,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075889515111523e-05,
+      "loss": 0.8347,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707203241579644e-05,
+      "loss": 0.7789,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706817321051457e-05,
+      "loss": 0.8083,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064311900415127e-05,
+      "loss": 0.8476,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060448486647975e-05,
+      "loss": 0.7806,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705658297036359e-05,
+      "loss": 0.861,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052715352713076e-05,
+      "loss": 0.8182,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048845634848156e-05,
+      "loss": 0.7332,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70449738179212e-05,
+      "loss": 0.8317,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041099903085175e-05,
+      "loss": 0.7582,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037223891493696e-05,
+      "loss": 0.7843,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703334578430099e-05,
+      "loss": 0.8739,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702946558266191e-05,
+      "loss": 0.7562,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025583287731934e-05,
+      "loss": 0.7786,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702169890066716e-05,
+      "loss": 0.8587,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701781242262432e-05,
+      "loss": 0.7878,
+      "step": 1632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013923854760754e-05,
+      "loss": 0.864,
+      "step": 1633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010033198234433e-05,
+      "loss": 0.8172,
+      "step": 1634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006140454203948e-05,
+      "loss": 0.2895,
+      "step": 1635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700224562382851e-05,
+      "loss": 0.8025,
+      "step": 1636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699834870826796e-05,
+      "loss": 0.7708,
+      "step": 1637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994449708682745e-05,
+      "loss": 0.8023,
+      "step": 1638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990548626233942e-05,
+      "loss": 0.6643,
+      "step": 1639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698664546208325e-05,
+      "loss": 0.2596,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698274021739298e-05,
+      "loss": 0.8158,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978832893326074e-05,
+      "loss": 0.8119,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697492349104608e-05,
+      "loss": 0.8339,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971012011717173e-05,
+      "loss": 0.7619,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967098456504148e-05,
+      "loss": 0.7637,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696318282657241e-05,
+      "loss": 0.8564,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695926512308799e-05,
+      "loss": 0.7689,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955345347217535e-05,
+      "loss": 0.7916,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69514235001283e-05,
+      "loss": 0.7463,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947499582988177e-05,
+      "loss": 0.8052,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943573596965644e-05,
+      "loss": 0.8788,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693964554322983e-05,
+      "loss": 0.8366,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935715422950455e-05,
+      "loss": 0.8693,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693178323729786e-05,
+      "loss": 0.784,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927848987443006e-05,
+      "loss": 0.8152,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923912674557464e-05,
+      "loss": 0.82,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691997429981342e-05,
+      "loss": 0.8446,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916033864383677e-05,
+      "loss": 0.765,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912091369441652e-05,
+      "loss": 0.7962,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908146816161372e-05,
+      "loss": 0.7983,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690420020571747e-05,
+      "loss": 0.7326,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900251539285207e-05,
+      "loss": 0.8381,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689630081804045e-05,
+      "loss": 0.7981,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892348043159668e-05,
+      "loss": 0.8737,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688839321581996e-05,
+      "loss": 0.7955,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884436337199024e-05,
+      "loss": 0.2692,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880477408475167e-05,
+      "loss": 0.8362,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876516430827314e-05,
+      "loss": 0.8443,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872553405434993e-05,
+      "loss": 0.8606,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868588333478352e-05,
+      "loss": 0.8443,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864621216138134e-05,
+      "loss": 0.8722,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686065205459571e-05,
+      "loss": 0.7604,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685668085003304e-05,
+      "loss": 0.2616,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68527076036327e-05,
+      "loss": 0.8226,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848732316577887e-05,
+      "loss": 0.8285,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844754990052383e-05,
+      "loss": 0.8179,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684077562524059e-05,
+      "loss": 0.7042,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836794223327516e-05,
+      "loss": 0.8743,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683281078549878e-05,
+      "loss": 0.837,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828825312940594e-05,
+      "loss": 0.2705,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682483780683979e-05,
+      "loss": 0.8549,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820848268383797e-05,
+      "loss": 0.8432,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816856698760658e-05,
+      "loss": 0.8679,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812863099159e-05,
+      "loss": 0.8232,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808867470768085e-05,
+      "loss": 0.7979,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804869814777757e-05,
+      "loss": 0.8195,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680087013237847e-05,
+      "loss": 0.7894,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796868424761284e-05,
+      "loss": 0.7307,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792864693117856e-05,
+      "loss": 0.7717,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678885893864045e-05,
+      "loss": 0.793,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784851162521935e-05,
+      "loss": 0.8273,
+      "step": 1691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780841365955776e-05,
+      "loss": 0.7833,
+      "step": 1692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776829550136045e-05,
+      "loss": 0.8281,
+      "step": 1693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.8324,
+      "step": 1694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768799865515147e-05,
+      "loss": 0.8462,
+      "step": 1695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764781999105123e-05,
+      "loss": 0.8307,
+      "step": 1696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760762118223814e-05,
+      "loss": 0.7908,
+      "step": 1697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756740224068293e-05,
+      "loss": 0.8012,
+      "step": 1698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675271631783623e-05,
+      "loss": 0.7448,
+      "step": 1699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748690400725894e-05,
+      "loss": 0.815,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744662473936163e-05,
+      "loss": 0.718,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740632538666494e-05,
+      "loss": 0.8002,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736600596116965e-05,
+      "loss": 0.2931,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732566647488234e-05,
+      "loss": 0.8631,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728530693981567e-05,
+      "loss": 0.8252,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724492736798813e-05,
+      "loss": 0.7923,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720452777142438e-05,
+      "loss": 0.819,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716410816215495e-05,
+      "loss": 0.8314,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671236685522162e-05,
+      "loss": 0.8393,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708320895365066e-05,
+      "loss": 0.8241,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704272937850667e-05,
+      "loss": 0.7552,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670022298388386e-05,
+      "loss": 0.8571,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669617103467067e-05,
+      "loss": 0.7975,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669211709141772e-05,
+      "loss": 0.839,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688061155332226e-05,
+      "loss": 0.8126,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684003227621998e-05,
+      "loss": 0.8228,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679943309495438e-05,
+      "loss": 0.8139,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667588140216154e-05,
+      "loss": 0.8253,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667181750682989e-05,
+      "loss": 0.8104,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667751624710667e-05,
+      "loss": 0.8321,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663683757014654e-05,
+      "loss": 0.7851,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659613904953197e-05,
+      "loss": 0.8467,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665554206973826e-05,
+      "loss": 0.8085,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651468252582383e-05,
+      "loss": 0.7878,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647392454698698e-05,
+      "loss": 0.7995,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643314677300937e-05,
+      "loss": 0.7062,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66392349216034e-05,
+      "loss": 0.8108,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635153188821006e-05,
+      "loss": 0.75,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663106948016923e-05,
+      "loss": 0.8244,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626983796864164e-05,
+      "loss": 0.8196,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622896140122474e-05,
+      "loss": 0.7929,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618806511161406e-05,
+      "loss": 0.8759,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614714911198814e-05,
+      "loss": 0.7967,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610621341453117e-05,
+      "loss": 0.8687,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660652580314334e-05,
+      "loss": 0.852,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602428297489086e-05,
+      "loss": 0.7354,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598328825710536e-05,
+      "loss": 0.8388,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659422738902847e-05,
+      "loss": 0.7637,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590123988664237e-05,
+      "loss": 0.7352,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586018625839796e-05,
+      "loss": 0.8752,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658191130177766e-05,
+      "loss": 0.7864,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577802017700954e-05,
+      "loss": 0.8272,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657369077483336e-05,
+      "loss": 0.7965,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656957757439917e-05,
+      "loss": 0.8713,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656546241762324e-05,
+      "loss": 0.289,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561345305731013e-05,
+      "loss": 0.7777,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655722623994852e-05,
+      "loss": 0.7349,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6553105221502364e-05,
+      "loss": 0.7878,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548982251619744e-05,
+      "loss": 0.9179,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544857331528425e-05,
+      "loss": 0.8167,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540730462456755e-05,
+      "loss": 0.8437,
+      "step": 1751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536601645633676e-05,
+      "loss": 0.8022,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653247088228869e-05,
+      "loss": 0.2642,
+      "step": 1753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528338173651898e-05,
+      "loss": 0.6902,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524203520953962e-05,
+      "loss": 0.8116,
+      "step": 1755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520066925426146e-05,
+      "loss": 0.8219,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515928388300264e-05,
+      "loss": 0.873,
+      "step": 1757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511787910808736e-05,
+      "loss": 0.7592,
+      "step": 1758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650764549418453e-05,
+      "loss": 0.8616,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650350113966123e-05,
+      "loss": 0.8629,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499354848472958e-05,
+      "loss": 0.794,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495206621854433e-05,
+      "loss": 0.7915,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491056461040953e-05,
+      "loss": 0.8733,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648690436726838e-05,
+      "loss": 0.807,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482750341773167e-05,
+      "loss": 0.2614,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478594385792325e-05,
+      "loss": 0.8135,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647443650056345e-05,
+      "loss": 0.7988,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647027668732471e-05,
+      "loss": 0.8236,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466114947314848e-05,
+      "loss": 0.8883,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461951281773183e-05,
+      "loss": 0.7919,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457785691939604e-05,
+      "loss": 0.7961,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453618179054573e-05,
+      "loss": 0.2847,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449448744359124e-05,
+      "loss": 0.7687,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445277389094868e-05,
+      "loss": 0.8176,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644110411450398e-05,
+      "loss": 0.7417,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436928921829218e-05,
+      "loss": 0.7827,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432751812313896e-05,
+      "loss": 0.2731,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642857278720192e-05,
+      "loss": 0.762,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424391847737736e-05,
+      "loss": 0.7844,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420208995166397e-05,
+      "loss": 0.899,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641602423073349e-05,
+      "loss": 0.8551,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411837555685203e-05,
+      "loss": 0.8338,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407648971268266e-05,
+      "loss": 0.8326,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403458478729996e-05,
+      "loss": 0.8152,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399266079318263e-05,
+      "loss": 0.8202,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395071774281527e-05,
+      "loss": 0.7928,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639087556486879e-05,
+      "loss": 0.81,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638667745232964e-05,
+      "loss": 0.7886,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382477437914223e-05,
+      "loss": 0.7527,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637827552287325e-05,
+      "loss": 0.8914,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374071708458005e-05,
+      "loss": 0.7886,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369865995920335e-05,
+      "loss": 0.8906,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365658386512643e-05,
+      "loss": 0.8204,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361448881487913e-05,
+      "loss": 0.7337,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357237482099682e-05,
+      "loss": 0.7496,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353024189602057e-05,
+      "loss": 0.7817,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348809005249703e-05,
+      "loss": 0.8215,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634459193029785e-05,
+      "loss": 0.8442,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6340372966002296e-05,
+      "loss": 0.7583,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63361521136194e-05,
+      "loss": 0.8818,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331929374406077e-05,
+      "loss": 0.8414,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327704749619812e-05,
+      "loss": 0.795,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323478240518645e-05,
+      "loss": 0.8127,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631924984836118e-05,
+      "loss": 0.7391,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315019574406585e-05,
+      "loss": 0.8505,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310787419914582e-05,
+      "loss": 0.8476,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306553386145453e-05,
+      "loss": 0.7795,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630231747436005e-05,
+      "loss": 0.6977,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298079685819773e-05,
+      "loss": 0.7508,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293840021786588e-05,
+      "loss": 0.8119,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6289598483523007e-05,
+      "loss": 0.7019,
+      "step": 1811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628535507229212e-05,
+      "loss": 0.7811,
+      "step": 1812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628110978935756e-05,
+      "loss": 0.8174,
+      "step": 1813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627686263598352e-05,
+      "loss": 0.8013,
+      "step": 1814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272613613434755e-05,
+      "loss": 0.7188,
+      "step": 1815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626836272297657e-05,
+      "loss": 0.8148,
+      "step": 1816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626410996587483e-05,
+      "loss": 0.8622,
+      "step": 1817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625985534339596e-05,
+      "loss": 0.8104,
+      "step": 1818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625559885680693e-05,
+      "loss": 0.2783,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625134050737527e-05,
+      "loss": 0.8701,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624708029636907e-05,
+      "loss": 0.8423,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242818225056968e-05,
+      "loss": 0.8025,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623855429470816e-05,
+      "loss": 0.8298,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623428850659239e-05,
+      "loss": 0.7561,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230020861979962e-05,
+      "loss": 0.8168,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225751362141726e-05,
+      "loss": 0.8087,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221480008349088e-05,
+      "loss": 0.7203,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217206801874013e-05,
+      "loss": 0.8332,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212931743989004e-05,
+      "loss": 0.7886,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208654835967123e-05,
+      "loss": 0.8212,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204376079081986e-05,
+      "loss": 0.7425,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200095474607753e-05,
+      "loss": 0.8119,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195813023819136e-05,
+      "loss": 0.7652,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61915287279914e-05,
+      "loss": 0.7448,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187242588400352e-05,
+      "loss": 0.8035,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182954606322364e-05,
+      "loss": 0.8867,
+      "step": 1836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178664783034332e-05,
+      "loss": 0.8491,
+      "step": 1837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174373119813723e-05,
+      "loss": 0.8248,
+      "step": 1838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617007961793854e-05,
+      "loss": 0.8432,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616578427868734e-05,
+      "loss": 0.8112,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616148710333922e-05,
+      "loss": 0.769,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157188093173833e-05,
+      "loss": 0.823,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152887249471363e-05,
+      "loss": 0.8938,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148584573512557e-05,
+      "loss": 0.9052,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61442800665787e-05,
+      "loss": 0.7884,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139973729951622e-05,
+      "loss": 0.7315,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61356655649137e-05,
+      "loss": 0.8,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131355572747853e-05,
+      "loss": 0.8444,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612704375473754e-05,
+      "loss": 0.8216,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612273011216678e-05,
+      "loss": 0.723,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118414646320115e-05,
+      "loss": 0.3008,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611409735848264e-05,
+      "loss": 0.7919,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109778249939994e-05,
+      "loss": 0.7656,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610545732197836e-05,
+      "loss": 0.8594,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610113457588445e-05,
+      "loss": 0.7456,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096810012945532e-05,
+      "loss": 0.2754,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092483634449408e-05,
+      "loss": 0.8619,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088155441684422e-05,
+      "loss": 0.8771,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608382543593946e-05,
+      "loss": 0.8441,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079493618503944e-05,
+      "loss": 0.8022,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075159990667834e-05,
+      "loss": 0.8007,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070824553721637e-05,
+      "loss": 0.7941,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066487308956396e-05,
+      "loss": 0.8325,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062148257663687e-05,
+      "loss": 0.7903,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057807401135623e-05,
+      "loss": 0.8735,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053464740664874e-05,
+      "loss": 0.8636,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049120277544616e-05,
+      "loss": 0.8184,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044774013068588e-05,
+      "loss": 0.8942,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040425948531054e-05,
+      "loss": 0.8101,
+      "step": 1869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6036076085226813e-05,
+      "loss": 0.7258,
+      "step": 1870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603172442445121e-05,
+      "loss": 0.8347,
+      "step": 1871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027370967500104e-05,
+      "loss": 0.7758,
+      "step": 1872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023015715669914e-05,
+      "loss": 0.7448,
+      "step": 1873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601865867025758e-05,
+      "loss": 0.7807,
+      "step": 1874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014299832560577e-05,
+      "loss": 0.7673,
+      "step": 1875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600993920387691e-05,
+      "loss": 0.8249,
+      "step": 1876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005576785505127e-05,
+      "loss": 0.8238,
+      "step": 1877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001212578744306e-05,
+      "loss": 0.8672,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599684658489405e-05,
+      "loss": 0.8018,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992478805254507e-05,
+      "loss": 0.7957,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598810924112634e-05,
+      "loss": 0.8645,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983737893810764e-05,
+      "loss": 0.802,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979364764609506e-05,
+      "loss": 0.7482,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974989854824832e-05,
+      "loss": 0.8468,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970613165759545e-05,
+      "loss": 0.7953,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966234698716963e-05,
+      "loss": 0.8124,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961854455000944e-05,
+      "loss": 0.7976,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595747243591587e-05,
+      "loss": 0.8384,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595308864276666e-05,
+      "loss": 0.8628,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594870307685875e-05,
+      "loss": 0.7868,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944315739498113e-05,
+      "loss": 0.8628,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593992663199124e-05,
+      "loss": 0.8233,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935535755645162e-05,
+      "loss": 0.2712,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931143111767422e-05,
+      "loss": 0.7439,
+      "step": 1894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59267487016661e-05,
+      "loss": 0.2658,
+      "step": 1895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922352526649803e-05,
+      "loss": 0.8102,
+      "step": 1896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591795458802765e-05,
+      "loss": 0.8397,
+      "step": 1897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5913554887109305e-05,
+      "loss": 0.7797,
+      "step": 1898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590915342520494e-05,
+      "loss": 0.258,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590475020362526e-05,
+      "loss": 0.8152,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900345223681493e-05,
+      "loss": 0.7785,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895938486685384e-05,
+      "loss": 0.8549,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5891529993949214e-05,
+      "loss": 0.8399,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887119746785776e-05,
+      "loss": 0.7862,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882707746508387e-05,
+      "loss": 0.7526,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878293994430887e-05,
+      "loss": 0.7641,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587387849186765e-05,
+      "loss": 0.7149,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586946124013354e-05,
+      "loss": 0.8426,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865042240543984e-05,
+      "loss": 0.8485,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860621494414892e-05,
+      "loss": 0.8109,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585619900306271e-05,
+      "loss": 0.7557,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851774767804415e-05,
+      "loss": 0.8788,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847348789957476e-05,
+      "loss": 0.8903,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842921070839908e-05,
+      "loss": 0.7727,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838491611770227e-05,
+      "loss": 0.7824,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834060414067474e-05,
+      "loss": 0.7429,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582962747905121e-05,
+      "loss": 0.7742,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825192808041504e-05,
+      "loss": 0.8594,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582075640235895e-05,
+      "loss": 0.8159,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816318263324663e-05,
+      "loss": 0.7337,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811878392260263e-05,
+      "loss": 0.7193,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580743679048789e-05,
+      "loss": 0.8533,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802993459330197e-05,
+      "loss": 0.7337,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579854840011036e-05,
+      "loss": 0.8151,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794101614152068e-05,
+      "loss": 0.2839,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789653102779514e-05,
+      "loss": 0.8561,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578520286731741e-05,
+      "loss": 0.8378,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578075090909099e-05,
+      "loss": 0.8528,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776297229425984e-05,
+      "loss": 0.776,
+      "step": 1929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771841829648652e-05,
+      "loss": 0.8202,
+      "step": 1930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576738471108576e-05,
+      "loss": 0.8555,
+      "step": 1931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762925875064584e-05,
+      "loss": 0.8689,
+      "step": 1932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575846532291291e-05,
+      "loss": 0.8063,
+      "step": 1933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754003055959035e-05,
+      "loss": 0.8435,
+      "step": 1934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749539075531773e-05,
+      "loss": 0.791,
+      "step": 1935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745073382960443e-05,
+      "loss": 0.7871,
+      "step": 1936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740605979574877e-05,
+      "loss": 0.8169,
+      "step": 1937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736136866705408e-05,
+      "loss": 0.7544,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573166604568289e-05,
+      "loss": 0.8046,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572719351783868e-05,
+      "loss": 0.8006,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722719284504638e-05,
+      "loss": 0.7947,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571824334701314e-05,
+      "loss": 0.7299,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571376570669707e-05,
+      "loss": 0.6858,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709286364889814e-05,
+      "loss": 0.8801,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570480532292527e-05,
+      "loss": 0.7987,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570032258213783e-05,
+      "loss": 0.8313,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695838143862404e-05,
+      "loss": 0.7568,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691352009434408e-05,
+      "loss": 0.7827,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686864180189758e-05,
+      "loss": 0.79,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682374657464873e-05,
+      "loss": 0.8181,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567788344259668e-05,
+      "loss": 0.7744,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673390536922612e-05,
+      "loss": 0.8488,
+      "step": 1952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56688959417806e-05,
+      "loss": 0.8406,
+      "step": 1953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664399658509084e-05,
+      "loss": 0.7342,
+      "step": 1954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659901688447e-05,
+      "loss": 0.7921,
+      "step": 1955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565540203293379e-05,
+      "loss": 0.8067,
+      "step": 1956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56509006933094e-05,
+      "loss": 0.8341,
+      "step": 1957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564639767091427e-05,
+      "loss": 0.7807,
+      "step": 1958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641892967089352e-05,
+      "loss": 0.8404,
+      "step": 1959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637386583176094e-05,
+      "loss": 0.8112,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632878520516438e-05,
+      "loss": 0.8251,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562836878045283e-05,
+      "loss": 0.8147,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623857364328227e-05,
+      "loss": 0.8151,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619344273486063e-05,
+      "loss": 0.7665,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561482950927029e-05,
+      "loss": 0.2823,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610313073025347e-05,
+      "loss": 0.8562,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605794966096177e-05,
+      "loss": 0.8102,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560127518982822e-05,
+      "loss": 0.8654,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596753745567407e-05,
+      "loss": 0.782,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559223063466017e-05,
+      "loss": 0.8083,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587705858453442e-05,
+      "loss": 0.7973,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583179418294644e-05,
+      "loss": 0.7675,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578651315531695e-05,
+      "loss": 0.8711,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574121551513013e-05,
+      "loss": 0.7528,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569590127587506e-05,
+      "loss": 0.8707,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556505704510458e-05,
+      "loss": 0.7373,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560522305414126e-05,
+      "loss": 0.8162,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555985909866543e-05,
+      "loss": 0.8098,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551447859812716e-05,
+      "loss": 0.805,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546908156604015e-05,
+      "loss": 0.8296,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542366801592312e-05,
+      "loss": 0.793,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537823796129972e-05,
+      "loss": 0.8035,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533279141569843e-05,
+      "loss": 0.261,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528732839265272e-05,
+      "loss": 0.6645,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524184890570094e-05,
+      "loss": 0.8541,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519635296838633e-05,
+      "loss": 0.873,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55150840594257e-05,
+      "loss": 0.8525,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510531179686603e-05,
+      "loss": 0.7493,
+      "step": 1988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505976658977136e-05,
+      "loss": 0.2604,
+      "step": 1989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550142049865358e-05,
+      "loss": 0.8645,
+      "step": 1990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54968627000727e-05,
+      "loss": 0.8521,
+      "step": 1991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549230326459176e-05,
+      "loss": 0.8158,
+      "step": 1992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487742193568505e-05,
+      "loss": 0.8392,
+      "step": 1993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548317948836117e-05,
+      "loss": 0.2919,
+      "step": 1994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478615150328465e-05,
+      "loss": 0.7258,
+      "step": 1995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474049180829603e-05,
+      "loss": 0.2784,
+      "step": 1996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.8503,
+      "step": 1997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464912352872646e-05,
+      "loss": 0.7442,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5460341497135397e-05,
+      "loss": 0.8374,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545576901537366e-05,
+      "loss": 0.7697,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451194908949073e-05,
+      "loss": 0.2849,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446619179223743e-05,
+      "loss": 0.8419,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442041827560274e-05,
+      "loss": 0.8282,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437462855321746e-05,
+      "loss": 0.8342,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543288226387172e-05,
+      "loss": 0.8075,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542830005457424e-05,
+      "loss": 0.7755,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423716228793834e-05,
+      "loss": 0.7904,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419130787895517e-05,
+      "loss": 0.2522,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541454373324477e-05,
+      "loss": 0.7973,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409955066207568e-05,
+      "loss": 0.8202,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405364788150362e-05,
+      "loss": 0.8238,
+      "step": 2011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540077290044008e-05,
+      "loss": 0.7859,
+      "step": 2012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396179404444123e-05,
+      "loss": 0.7833,
+      "step": 2013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391584301530397e-05,
+      "loss": 0.7599,
+      "step": 2014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386987593067253e-05,
+      "loss": 0.8366,
+      "step": 2015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382389280423542e-05,
+      "loss": 0.8065,
+      "step": 2016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377789364968588e-05,
+      "loss": 0.8078,
+      "step": 2017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373187848072194e-05,
+      "loss": 0.8541,
+      "step": 2018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368584731104627e-05,
+      "loss": 0.71,
+      "step": 2019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363980015436646e-05,
+      "loss": 0.7184,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359373702439477e-05,
+      "loss": 0.8468,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354765793484834e-05,
+      "loss": 0.8051,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350156289944887e-05,
+      "loss": 0.7439,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345545193192298e-05,
+      "loss": 0.8449,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534093250460019e-05,
+      "loss": 0.7634,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533631822554217e-05,
+      "loss": 0.8453,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533170235739232e-05,
+      "loss": 0.8447,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327084901525183e-05,
+      "loss": 0.7703,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322465859315788e-05,
+      "loss": 0.7976,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317845232139622e-05,
+      "loss": 0.8825,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313223021372665e-05,
+      "loss": 0.2767,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308599228391343e-05,
+      "loss": 0.8301,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303973854572575e-05,
+      "loss": 0.72,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299346901293743e-05,
+      "loss": 0.7124,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294718369932692e-05,
+      "loss": 0.8097,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529008826186775e-05,
+      "loss": 0.7406,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52854565784777e-05,
+      "loss": 0.8288,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528082332114181e-05,
+      "loss": 0.8158,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276188491239802e-05,
+      "loss": 0.7866,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527155209015188e-05,
+      "loss": 0.7326,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52669141192587e-05,
+      "loss": 0.8742,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262274579941406e-05,
+      "loss": 0.7832,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257633473581584e-05,
+      "loss": 0.8232,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252990801561312e-05,
+      "loss": 0.7741,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248346565263114e-05,
+      "loss": 0.8434,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243700766069994e-05,
+      "loss": 0.8411,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239053405365409e-05,
+      "loss": 0.7774,
+      "step": 2047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5234404484533294e-05,
+      "loss": 0.8126,
+      "step": 2048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522975400495804e-05,
+      "loss": 0.7691,
+      "step": 2049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225101968024504e-05,
+      "loss": 0.792,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220448375118003e-05,
+      "loss": 0.8499,
+      "step": 2051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521579322762433e-05,
+      "loss": 0.8316,
+      "step": 2052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211136526929719e-05,
+      "loss": 0.8201,
+      "step": 2053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206478274420895e-05,
+      "loss": 0.8498,
+      "step": 2054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201818471485022e-05,
+      "loss": 0.8095,
+      "step": 2055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197157119509735e-05,
+      "loss": 0.9041,
+      "step": 2056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192494219883132e-05,
+      "loss": 0.8085,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518782977399376e-05,
+      "loss": 0.8464,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183163783230646e-05,
+      "loss": 0.8079,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178496248983254e-05,
+      "loss": 0.873,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173827172641536e-05,
+      "loss": 0.793,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516915655559587e-05,
+      "loss": 0.7984,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164484399237123e-05,
+      "loss": 0.7423,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159810704956602e-05,
+      "loss": 0.8858,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155135474146078e-05,
+      "loss": 0.8741,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515045870819778e-05,
+      "loss": 0.8389,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145780408504389e-05,
+      "loss": 0.2863,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141100576459053e-05,
+      "loss": 0.7687,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136419213455367e-05,
+      "loss": 0.8333,
+      "step": 2069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513173632088738e-05,
+      "loss": 0.85,
+      "step": 2070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127051900149611e-05,
+      "loss": 0.8569,
+      "step": 2071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122365952637017e-05,
+      "loss": 0.2591,
+      "step": 2072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117678479745022e-05,
+      "loss": 0.7723,
+      "step": 2073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112989482869498e-05,
+      "loss": 0.8087,
+      "step": 2074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108298963406772e-05,
+      "loss": 0.8522,
+      "step": 2075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103606922753622e-05,
+      "loss": 0.7617,
+      "step": 2076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098913362307284e-05,
+      "loss": 0.8095,
+      "step": 2077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094218283465445e-05,
+      "loss": 0.8485,
+      "step": 2078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089521687626243e-05,
+      "loss": 0.8191,
+      "step": 2079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508482357618827e-05,
+      "loss": 0.7907,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080123950550563e-05,
+      "loss": 0.8625,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507542281211262e-05,
+      "loss": 0.7421,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070720162274375e-05,
+      "loss": 0.8156,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506601600243623e-05,
+      "loss": 0.853,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061310333999025e-05,
+      "loss": 0.8093,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056603158364049e-05,
+      "loss": 0.7579,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051894476933044e-05,
+      "loss": 0.7741,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047184291108204e-05,
+      "loss": 0.8394,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504247260229216e-05,
+      "loss": 0.8408,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037759411887995e-05,
+      "loss": 0.775,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503304472129925e-05,
+      "loss": 0.7816,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028328531929894e-05,
+      "loss": 0.8243,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023610845184359e-05,
+      "loss": 0.8146,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501889166246751e-05,
+      "loss": 0.8088,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014170985184669e-05,
+      "loss": 0.8742,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009448814741592e-05,
+      "loss": 0.7898,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004725152544483e-05,
+      "loss": 0.8366,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8701,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995273358515236e-05,
+      "loss": 0.8372,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990545229497722e-05,
+      "loss": 0.8534,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985815614355442e-05,
+      "loss": 0.7862,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981084514496822e-05,
+      "loss": 0.8309,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976351931330722e-05,
+      "loss": 0.7415,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497161786626645e-05,
+      "loss": 0.8338,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966882320713758e-05,
+      "loss": 0.7728,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962145296082828e-05,
+      "loss": 0.8263,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957406793784295e-05,
+      "loss": 0.8117,
+      "step": 2107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952666815229224e-05,
+      "loss": 0.8359,
+      "step": 2108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947925361829127e-05,
+      "loss": 0.7948,
+      "step": 2109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943182434995956e-05,
+      "loss": 0.8055,
+      "step": 2110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493843803614209e-05,
+      "loss": 0.922,
+      "step": 2111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933692166680355e-05,
+      "loss": 0.8099,
+      "step": 2112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928944828024024e-05,
+      "loss": 0.7797,
+      "step": 2113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924196021586785e-05,
+      "loss": 0.8328,
+      "step": 2114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919445748782783e-05,
+      "loss": 0.8083,
+      "step": 2115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4914694011026594e-05,
+      "loss": 0.801,
+      "step": 2116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909940809733223e-05,
+      "loss": 0.8542,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490518614631812e-05,
+      "loss": 0.797,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490043002219717e-05,
+      "loss": 0.8292,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489567243878668e-05,
+      "loss": 0.7966,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890913397503415e-05,
+      "loss": 0.778,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886152899764548e-05,
+      "loss": 0.7504,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881390946987708e-05,
+      "loss": 0.7843,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876627540590941e-05,
+      "loss": 0.8172,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487186268199274e-05,
+      "loss": 0.786,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486709637261201e-05,
+      "loss": 0.7609,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862328613868116e-05,
+      "loss": 0.7932,
+      "step": 2127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857559407180831e-05,
+      "loss": 0.8677,
+      "step": 2128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852788753970369e-05,
+      "loss": 0.7753,
+      "step": 2129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848016655657375e-05,
+      "loss": 0.8723,
+      "step": 2130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843243113662919e-05,
+      "loss": 0.8916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838468129408512e-05,
+      "loss": 0.8098,
+      "step": 2132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483369170431608e-05,
+      "loss": 0.8356,
+      "step": 2133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828913839807992e-05,
+      "loss": 0.8051,
+      "step": 2134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824134537307033e-05,
+      "loss": 0.7097,
+      "step": 2135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819353798236427e-05,
+      "loss": 0.8702,
+      "step": 2136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814571624019815e-05,
+      "loss": 0.8103,
+      "step": 2137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809788016081277e-05,
+      "loss": 0.8681,
+      "step": 2138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805002975845313e-05,
+      "loss": 0.8154,
+      "step": 2139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800216504736848e-05,
+      "loss": 0.7359,
+      "step": 2140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795428604181241e-05,
+      "loss": 0.7863,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790639275604269e-05,
+      "loss": 0.8305,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785848520432133e-05,
+      "loss": 0.788,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781056340091468e-05,
+      "loss": 0.7207,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776262736009326e-05,
+      "loss": 0.8535,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771467709613185e-05,
+      "loss": 0.7761,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476667126233095e-05,
+      "loss": 0.8409,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761873395590937e-05,
+      "loss": 0.7751,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757074110821905e-05,
+      "loss": 0.8201,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752273409453015e-05,
+      "loss": 0.7997,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747471292913864e-05,
+      "loss": 0.7574,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742667762634457e-05,
+      "loss": 0.8406,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737862820045239e-05,
+      "loss": 0.8105,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733056466577051e-05,
+      "loss": 0.717,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728248703661183e-05,
+      "loss": 0.7868,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723439532729318e-05,
+      "loss": 0.8395,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718628955213574e-05,
+      "loss": 0.7833,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713816972546484e-05,
+      "loss": 0.7951,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709003586160995e-05,
+      "loss": 0.2882,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704188797490484e-05,
+      "loss": 0.8651,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699372607968727e-05,
+      "loss": 0.7702,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694555019029939e-05,
+      "loss": 0.7476,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689736032108728e-05,
+      "loss": 0.7807,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684915648640143e-05,
+      "loss": 0.698,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680093870059625e-05,
+      "loss": 0.7911,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675270697803055e-05,
+      "loss": 0.7911,
+      "step": 2166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670446133306706e-05,
+      "loss": 0.2855,
+      "step": 2167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665620178007278e-05,
+      "loss": 0.7924,
+      "step": 2168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660792833341883e-05,
+      "loss": 0.8302,
+      "step": 2169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465596410074805e-05,
+      "loss": 0.7989,
+      "step": 2170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651133981663714e-05,
+      "loss": 0.7654,
+      "step": 2171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646302477527227e-05,
+      "loss": 0.7695,
+      "step": 2172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641469589777354e-05,
+      "loss": 0.7789,
+      "step": 2173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636635319853274e-05,
+      "loss": 0.7955,
+      "step": 2174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463179966919457e-05,
+      "loss": 0.7978,
+      "step": 2175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626962639241247e-05,
+      "loss": 0.7875,
+      "step": 2176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622124231433707e-05,
+      "loss": 0.7512,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617284447212775e-05,
+      "loss": 0.7756,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461244328801968e-05,
+      "loss": 0.8493,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607600755296063e-05,
+      "loss": 0.7925,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602756850483968e-05,
+      "loss": 0.7889,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597911575025856e-05,
+      "loss": 0.8236,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593064930364593e-05,
+      "loss": 0.8138,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588216917943448e-05,
+      "loss": 0.7757,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583367539206102e-05,
+      "loss": 0.7309,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578516795596644e-05,
+      "loss": 0.7675,
+      "step": 2186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573664688559568e-05,
+      "loss": 0.8176,
+      "step": 2187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568811219539771e-05,
+      "loss": 0.841,
+      "step": 2188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456395638998256e-05,
+      "loss": 0.7972,
+      "step": 2189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559100201333647e-05,
+      "loss": 0.7838,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554242655039139e-05,
+      "loss": 0.8535,
+      "step": 2191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549383752545567e-05,
+      "loss": 0.8507,
+      "step": 2192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544523495299843e-05,
+      "loss": 0.8053,
+      "step": 2193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45396618847493e-05,
+      "loss": 0.7312,
+      "step": 2194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534798922341667e-05,
+      "loss": 0.2575,
+      "step": 2195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529934609525076e-05,
+      "loss": 0.8884,
+      "step": 2196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525068947748058e-05,
+      "loss": 0.8705,
+      "step": 2197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520201938459552e-05,
+      "loss": 0.8277,
+      "step": 2198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515333583108896e-05,
+      "loss": 0.8147,
+      "step": 2199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510463883145826e-05,
+      "loss": 0.8416,
+      "step": 2200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505592840020479e-05,
+      "loss": 0.7821,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45007204551834e-05,
+      "loss": 0.83,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495846730085519e-05,
+      "loss": 0.7736,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490971666178177e-05,
+      "loss": 0.7868,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486095264913113e-05,
+      "loss": 0.7873,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481217527742458e-05,
+      "loss": 0.81,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476338456118739e-05,
+      "loss": 0.8254,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471458051494894e-05,
+      "loss": 0.8358,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466576315324241e-05,
+      "loss": 0.7677,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461693249060509e-05,
+      "loss": 0.7788,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456808854157816e-05,
+      "loss": 0.7574,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445192313207067e-05,
+      "loss": 0.8196,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447036084253992e-05,
+      "loss": 0.7462,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442147712163076e-05,
+      "loss": 0.8636,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437258017253627e-05,
+      "loss": 0.8793,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432367000981736e-05,
+      "loss": 0.8139,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442747466480389e-05,
+      "loss": 0.8548,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422581010176963e-05,
+      "loss": 0.2918,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417686038558233e-05,
+      "loss": 0.7946,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412789751405365e-05,
+      "loss": 0.7562,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440789215017641e-05,
+      "loss": 0.7775,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440299323632982e-05,
+      "loss": 0.7108,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398093011324425e-05,
+      "loss": 0.8047,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393191476619466e-05,
+      "loss": 0.7711,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388288633674545e-05,
+      "loss": 0.7095,
+      "step": 2225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4383384483949687e-05,
+      "loss": 0.7382,
+      "step": 2226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437847902890528e-05,
+      "loss": 0.8324,
+      "step": 2227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437357227000211e-05,
+      "loss": 0.3083,
+      "step": 2228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368664208701353e-05,
+      "loss": 0.7751,
+      "step": 2229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436375484646457e-05,
+      "loss": 0.7922,
+      "step": 2230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358844184753713e-05,
+      "loss": 0.825,
+      "step": 2231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353932225031113e-05,
+      "loss": 0.8079,
+      "step": 2232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349018968759503e-05,
+      "loss": 0.7874,
+      "step": 2233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344104417401978e-05,
+      "loss": 0.843,
+      "step": 2234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339188572422047e-05,
+      "loss": 0.7863,
+      "step": 2235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433427143528358e-05,
+      "loss": 0.6592,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329353007450846e-05,
+      "loss": 0.8397,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324433290388492e-05,
+      "loss": 0.8418,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319512285561553e-05,
+      "loss": 0.8139,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314589994435442e-05,
+      "loss": 0.7347,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309666418475963e-05,
+      "loss": 0.8031,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304741559149294e-05,
+      "loss": 0.8897,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299815417922003e-05,
+      "loss": 0.8177,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294887996261034e-05,
+      "loss": 0.7753,
+      "step": 2244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289959295633713e-05,
+      "loss": 0.7845,
+      "step": 2245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285029317507753e-05,
+      "loss": 0.7843,
+      "step": 2246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280098063351238e-05,
+      "loss": 0.8453,
+      "step": 2247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275165534632639e-05,
+      "loss": 0.7941,
+      "step": 2248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270231732820801e-05,
+      "loss": 0.7343,
+      "step": 2249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265296659384956e-05,
+      "loss": 0.7555,
+      "step": 2250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260360315794705e-05,
+      "loss": 0.7743,
+      "step": 2251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255422703520037e-05,
+      "loss": 0.8057,
+      "step": 2252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250483824031309e-05,
+      "loss": 0.7586,
+      "step": 2253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245543678799265e-05,
+      "loss": 0.2467,
+      "step": 2254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240602269295016e-05,
+      "loss": 0.738,
+      "step": 2255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235659596990064e-05,
+      "loss": 0.8521,
+      "step": 2256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230715663356265e-05,
+      "loss": 0.8982,
+      "step": 2257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225770469865874e-05,
+      "loss": 0.7787,
+      "step": 2258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220824017991505e-05,
+      "loss": 0.733,
+      "step": 2259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215876309206152e-05,
+      "loss": 0.7704,
+      "step": 2260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210927344983189e-05,
+      "loss": 0.8446,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205977126796351e-05,
+      "loss": 0.7527,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201025656119763e-05,
+      "loss": 0.6987,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196072934427902e-05,
+      "loss": 0.8272,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191118963195643e-05,
+      "loss": 0.8119,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186163743898207e-05,
+      "loss": 0.8364,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181207278011206e-05,
+      "loss": 0.8286,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176249567010617e-05,
+      "loss": 0.8175,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171290612372781e-05,
+      "loss": 0.7382,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166330415574422e-05,
+      "loss": 0.8179,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161368978092624e-05,
+      "loss": 0.737,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156406301404847e-05,
+      "loss": 0.2667,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151442386988911e-05,
+      "loss": 0.7603,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146477236323015e-05,
+      "loss": 0.7703,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141510850885723e-05,
+      "loss": 0.8102,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413654323215596e-05,
+      "loss": 0.8091,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131574381613024e-05,
+      "loss": 0.8489,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126604300736585e-05,
+      "loss": 0.8649,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412163299100667e-05,
+      "loss": 0.821,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116660453903676e-05,
+      "loss": 0.7766,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111686690908365e-05,
+      "loss": 0.7745,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106711703501866e-05,
+      "loss": 0.7332,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101735493165667e-05,
+      "loss": 0.785,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096758061381629e-05,
+      "loss": 0.7829,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091779409631964e-05,
+      "loss": 0.803,
+      "step": 2285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086799539399266e-05,
+      "loss": 0.8189,
+      "step": 2286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081818452166469e-05,
+      "loss": 0.7943,
+      "step": 2287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076836149416889e-05,
+      "loss": 0.8314,
+      "step": 2288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071852632634196e-05,
+      "loss": 0.864,
+      "step": 2289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066867903302419e-05,
+      "loss": 0.8061,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061881962905955e-05,
+      "loss": 0.7537,
+      "step": 2291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056894812929549e-05,
+      "loss": 0.8363,
+      "step": 2292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051906454858326e-05,
+      "loss": 0.7926,
+      "step": 2293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404691689017775e-05,
+      "loss": 0.8439,
+      "step": 2294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404192612037366e-05,
+      "loss": 0.892,
+      "step": 2295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403693414693224e-05,
+      "loss": 0.7664,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403194097134005e-05,
+      "loss": 0.7734,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026946595083995e-05,
+      "loss": 0.7892,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021951019651334e-05,
+      "loss": 0.7914,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.8004,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401195627720706e-05,
+      "loss": 0.8273,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400695711317176e-05,
+      "loss": 0.7182,
+      "step": 2302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001956755912487e-05,
+      "loss": 0.7499,
+      "step": 2303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399695520691829e-05,
+      "loss": 0.8259,
+      "step": 2304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991952467678567e-05,
+      "loss": 0.7721,
+      "step": 2305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986948539683078e-05,
+      "loss": 0.7538,
+      "step": 2306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981943424421932e-05,
+      "loss": 0.7665,
+      "step": 2307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976937123385588e-05,
+      "loss": 0.7607,
+      "step": 2308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971929638064868e-05,
+      "loss": 0.7867,
+      "step": 2309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966920969950941e-05,
+      "loss": 0.2687,
+      "step": 2310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961911120535325e-05,
+      "loss": 0.2809,
+      "step": 2311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956900091309894e-05,
+      "loss": 0.7938,
+      "step": 2312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395188788376688e-05,
+      "loss": 0.7422,
+      "step": 2313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946874499398851e-05,
+      "loss": 0.7165,
+      "step": 2314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394185993969873e-05,
+      "loss": 0.8553,
+      "step": 2315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936844206159798e-05,
+      "loss": 0.8607,
+      "step": 2316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931827300275683e-05,
+      "loss": 0.2698,
+      "step": 2317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926809223540357e-05,
+      "loss": 0.7864,
+      "step": 2318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921789977448143e-05,
+      "loss": 0.7905,
+      "step": 2319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391676956349371e-05,
+      "loss": 0.8546,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911747983172083e-05,
+      "loss": 0.6836,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906725237978625e-05,
+      "loss": 0.8266,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901701329409052e-05,
+      "loss": 0.8331,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3896676258959417e-05,
+      "loss": 0.8565,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891650028126135e-05,
+      "loss": 0.8435,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886622638405953e-05,
+      "loss": 0.8861,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881594091295969e-05,
+      "loss": 0.8695,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387656438829362e-05,
+      "loss": 0.868,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38715335308967e-05,
+      "loss": 0.7474,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386650152060333e-05,
+      "loss": 0.8481,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861468358911987e-05,
+      "loss": 0.8298,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856434047321482e-05,
+      "loss": 0.8015,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851398587330982e-05,
+      "loss": 0.8093,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846361980439978e-05,
+      "loss": 0.8234,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841324228148309e-05,
+      "loss": 0.7583,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836285331956172e-05,
+      "loss": 0.8008,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831245293364075e-05,
+      "loss": 0.8635,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382620411387289e-05,
+      "loss": 0.2554,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821161794983815e-05,
+      "loss": 0.7412,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816118338198396e-05,
+      "loss": 0.7975,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811073745018513e-05,
+      "loss": 0.758,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806028016946384e-05,
+      "loss": 0.7938,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380098115548457e-05,
+      "loss": 0.2815,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795933162135962e-05,
+      "loss": 0.7919,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3790884038403796e-05,
+      "loss": 0.8049,
+      "step": 2345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785833785791636e-05,
+      "loss": 0.852,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3780782405803393e-05,
+      "loss": 0.8344,
+      "step": 2347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775729899943303e-05,
+      "loss": 0.8235,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377067626971594e-05,
+      "loss": 0.8241,
+      "step": 2349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765621516626225e-05,
+      "loss": 0.7415,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376056564217939e-05,
+      "loss": 0.8322,
+      "step": 2351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755508647881025e-05,
+      "loss": 0.8288,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750450535237036e-05,
+      "loss": 0.8546,
+      "step": 2353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745391305753669e-05,
+      "loss": 0.8328,
+      "step": 2354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740330960937505e-05,
+      "loss": 0.7701,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373526950229545e-05,
+      "loss": 0.7727,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730206931334751e-05,
+      "loss": 0.7351,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725143249562975e-05,
+      "loss": 0.8242,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720078458488036e-05,
+      "loss": 0.8739,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715012559618156e-05,
+      "loss": 0.2843,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709945554461908e-05,
+      "loss": 0.7861,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704877444528182e-05,
+      "loss": 0.8743,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36998082313262e-05,
+      "loss": 0.744,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694737916365517e-05,
+      "loss": 0.7866,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689666501156006e-05,
+      "loss": 0.7794,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684593987207881e-05,
+      "loss": 0.8291,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679520376031673e-05,
+      "loss": 0.7674,
+      "step": 2367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674445669138241e-05,
+      "loss": 0.7881,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669369868038773e-05,
+      "loss": 0.7528,
+      "step": 2369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664292974244784e-05,
+      "loss": 0.8149,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365921498926811e-05,
+      "loss": 0.8262,
+      "step": 2371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654135914620916e-05,
+      "loss": 0.7634,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364905575181569e-05,
+      "loss": 0.8068,
+      "step": 2373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364397450236524e-05,
+      "loss": 0.7213,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638892167782704e-05,
+      "loss": 0.8109,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633808749581542e-05,
+      "loss": 0.7722,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628724249275536e-05,
+      "loss": 0.8671,
+      "step": 2377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623638668378783e-05,
+      "loss": 0.7089,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618552008405715e-05,
+      "loss": 0.799,
+      "step": 2379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613464270871078e-05,
+      "loss": 0.7721,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608375457289934e-05,
+      "loss": 0.8298,
+      "step": 2381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603285569177675e-05,
+      "loss": 0.816,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598194608050011e-05,
+      "loss": 0.8353,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593102575422965e-05,
+      "loss": 0.2466,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588009472812889e-05,
+      "loss": 0.7244,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582915301736439e-05,
+      "loss": 0.8044,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577820063710607e-05,
+      "loss": 0.8627,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572723760252688e-05,
+      "loss": 0.7978,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567626392880308e-05,
+      "loss": 0.7877,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562527963111394e-05,
+      "loss": 0.7487,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35574284724642e-05,
+      "loss": 0.8229,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552327922457297e-05,
+      "loss": 0.7886,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547226314609562e-05,
+      "loss": 0.7786,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35421236504402e-05,
+      "loss": 0.756,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537019931468716e-05,
+      "loss": 0.7509,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531915159214941e-05,
+      "loss": 0.7417,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526809335199013e-05,
+      "loss": 0.7781,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521702460941391e-05,
+      "loss": 0.784,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516594537962831e-05,
+      "loss": 0.7668,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511485567784421e-05,
+      "loss": 0.8321,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506375551927546e-05,
+      "loss": 0.8055,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501264491913909e-05,
+      "loss": 0.7876,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496152389265524e-05,
+      "loss": 0.7651,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349103924550471e-05,
+      "loss": 0.8569,
+      "step": 2404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485925062154106e-05,
+      "loss": 0.8147,
+      "step": 2405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480809840736651e-05,
+      "loss": 0.7364,
+      "step": 2406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475693582775598e-05,
+      "loss": 0.8187,
+      "step": 2407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470576289794505e-05,
+      "loss": 0.7622,
+      "step": 2408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465457963317246e-05,
+      "loss": 0.8041,
+      "step": 2409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460338604867992e-05,
+      "loss": 0.778,
+      "step": 2410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345521821597123e-05,
+      "loss": 0.8142,
+      "step": 2411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345009679815175e-05,
+      "loss": 0.8296,
+      "step": 2412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344497435293465e-05,
+      "loss": 0.26,
+      "step": 2413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343985088184533e-05,
+      "loss": 0.7757,
+      "step": 2414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434726386409505e-05,
+      "loss": 0.8101,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342960086815318e-05,
+      "loss": 0.7881,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424474328602677e-05,
+      "loss": 0.7433,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419346769284616e-05,
+      "loss": 0.7992,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341421819172593e-05,
+      "loss": 0.8107,
+      "step": 2419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409088597453841e-05,
+      "loss": 0.8326,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403957987995884e-05,
+      "loss": 0.8177,
+      "step": 2421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398826364879892e-05,
+      "loss": 0.7816,
+      "step": 2422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393693729634e-05,
+      "loss": 0.7403,
+      "step": 2423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388560083786652e-05,
+      "loss": 0.7433,
+      "step": 2424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383425428866577e-05,
+      "loss": 0.834,
+      "step": 2425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337828976640282e-05,
+      "loss": 0.7621,
+      "step": 2426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373153097924716e-05,
+      "loss": 0.8082,
+      "step": 2427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336801542496191e-05,
+      "loss": 0.7433,
+      "step": 2428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362876749044334e-05,
+      "loss": 0.7749,
+      "step": 2429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357737071702225e-05,
+      "loss": 0.8079,
+      "step": 2430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352596394466119e-05,
+      "loss": 0.7808,
+      "step": 2431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347454718866848e-05,
+      "loss": 0.8283,
+      "step": 2432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342312046435541e-05,
+      "loss": 0.8636,
+      "step": 2433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3337168378703624e-05,
+      "loss": 0.7638,
+      "step": 2434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333202371720282e-05,
+      "loss": 0.774,
+      "step": 2435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326878063465149e-05,
+      "loss": 0.7446,
+      "step": 2436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321731419022925e-05,
+      "loss": 0.2649,
+      "step": 2437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316583785408755e-05,
+      "loss": 0.264,
+      "step": 2438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311435164155546e-05,
+      "loss": 0.3009,
+      "step": 2439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306285556796494e-05,
+      "loss": 0.8464,
+      "step": 2440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301134964865091e-05,
+      "loss": 0.8037,
+      "step": 2441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295983389895123e-05,
+      "loss": 0.8332,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290830833420666e-05,
+      "loss": 0.8802,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285677296976086e-05,
+      "loss": 0.8399,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280522782096055e-05,
+      "loss": 0.7117,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275367290315514e-05,
+      "loss": 0.7334,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270210823169718e-05,
+      "loss": 0.7719,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265053382194192e-05,
+      "loss": 0.8482,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259894968924765e-05,
+      "loss": 0.6507,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325473558489755e-05,
+      "loss": 0.7543,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324957523164895e-05,
+      "loss": 0.7855,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244413910715656e-05,
+      "loss": 0.7736,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323925162363465e-05,
+      "loss": 0.8685,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234088371943199e-05,
+      "loss": 0.7745,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228924157178851e-05,
+      "loss": 0.8027,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223758980879458e-05,
+      "loss": 0.826,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218592844583137e-05,
+      "loss": 0.8426,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213425749828315e-05,
+      "loss": 0.7455,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208257698153677e-05,
+      "loss": 0.8622,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203088691098218e-05,
+      "loss": 0.8113,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197918730201199e-05,
+      "loss": 0.679,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192747817002179e-05,
+      "loss": 0.813,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187575953040988e-05,
+      "loss": 0.7819,
+      "step": 2463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318240313985775e-05,
+      "loss": 0.7742,
+      "step": 2464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177229378992862e-05,
+      "loss": 0.7201,
+      "step": 2465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172054671987014e-05,
+      "loss": 0.817,
+      "step": 2466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316687902038117e-05,
+      "loss": 0.7242,
+      "step": 2467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316170242571657e-05,
+      "loss": 0.7794,
+      "step": 2468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156524889534759e-05,
+      "loss": 0.7851,
+      "step": 2469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151346413377528e-05,
+      "loss": 0.7847,
+      "step": 2470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146166998786975e-05,
+      "loss": 0.773,
+      "step": 2471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140986647305467e-05,
+      "loss": 0.8291,
+      "step": 2472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135805360475646e-05,
+      "loss": 0.7559,
+      "step": 2473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313062313984044e-05,
+      "loss": 0.7667,
+      "step": 2474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125439986943056e-05,
+      "loss": 0.7924,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120255903326968e-05,
+      "loss": 0.8464,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115070890535942e-05,
+      "loss": 0.6997,
+      "step": 2477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109884950114007e-05,
+      "loss": 0.7944,
+      "step": 2478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104698083605474e-05,
+      "loss": 0.7618,
+      "step": 2479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099510292554936e-05,
+      "loss": 0.7591,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094321578507247e-05,
+      "loss": 0.7498,
+      "step": 2481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089131943007551e-05,
+      "loss": 0.772,
+      "step": 2482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083941387601256e-05,
+      "loss": 0.7652,
+      "step": 2483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078749913834048e-05,
+      "loss": 0.7678,
+      "step": 2484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073557523251885e-05,
+      "loss": 0.2867,
+      "step": 2485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068364217401004e-05,
+      "loss": 0.8538,
+      "step": 2486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063169997827898e-05,
+      "loss": 0.8499,
+      "step": 2487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057974866079359e-05,
+      "loss": 0.7053,
+      "step": 2488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305277882370242e-05,
+      "loss": 0.7799,
+      "step": 2489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047581872244412e-05,
+      "loss": 0.7715,
+      "step": 2490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042384013252918e-05,
+      "loss": 0.7637,
+      "step": 2491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037185248275803e-05,
+      "loss": 0.822,
+      "step": 2492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031985578861195e-05,
+      "loss": 0.681,
+      "step": 2493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026785006557491e-05,
+      "loss": 0.7895,
+      "step": 2494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021583532913362e-05,
+      "loss": 0.7675,
+      "step": 2495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016381159477745e-05,
+      "loss": 0.8361,
+      "step": 2496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011177887799846e-05,
+      "loss": 0.7734,
+      "step": 2497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005973719429133e-05,
+      "loss": 0.7525,
+      "step": 2498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300076865591535e-05,
+      "loss": 0.793,
+      "step": 2499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995562698808494e-05,
+      "loss": 0.7311,
+      "step": 2500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990355849658851e-05,
+      "loss": 0.8079,
+      "step": 2501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985148110016947e-05,
+      "loss": 0.786,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297993948143359e-05,
+      "loss": 0.8107,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974729965459843e-05,
+      "loss": 0.7702,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969519563647042e-05,
+      "loss": 0.7745,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964308277546778e-05,
+      "loss": 0.8005,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959096108710916e-05,
+      "loss": 0.7933,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953883058691572e-05,
+      "loss": 0.7857,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948669129041131e-05,
+      "loss": 0.2833,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943454321312243e-05,
+      "loss": 0.7662,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938238637057811e-05,
+      "loss": 0.8195,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933022077831006e-05,
+      "loss": 0.8281,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927804645185256e-05,
+      "loss": 0.7077,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922586340674248e-05,
+      "loss": 0.8637,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917367165851935e-05,
+      "loss": 0.8452,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912147122272523e-05,
+      "loss": 0.7496,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906926211490476e-05,
+      "loss": 0.8095,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901704435060526e-05,
+      "loss": 0.7658,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896481794537646e-05,
+      "loss": 0.7262,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891258291477086e-05,
+      "loss": 0.7945,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886033927434331e-05,
+      "loss": 0.8389,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880808703965148e-05,
+      "loss": 0.7498,
+      "step": 2522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2875582622625535e-05,
+      "loss": 0.7176,
+      "step": 2523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870355684971763e-05,
+      "loss": 0.7234,
+      "step": 2524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286512789256035e-05,
+      "loss": 0.8628,
+      "step": 2525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859899246948067e-05,
+      "loss": 0.8226,
+      "step": 2526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854669749691947e-05,
+      "loss": 0.752,
+      "step": 2527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849439402349267e-05,
+      "loss": 0.8582,
+      "step": 2528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844208206477569e-05,
+      "loss": 0.8381,
+      "step": 2529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283897616363463e-05,
+      "loss": 0.8084,
+      "step": 2530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28337432753785e-05,
+      "loss": 0.8068,
+      "step": 2531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828509543267468e-05,
+      "loss": 0.8183,
+      "step": 2532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823274968860073e-05,
+      "loss": 0.7959,
+      "step": 2533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818039553715114e-05,
+      "loss": 0.7422,
+      "step": 2534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812803299391629e-05,
+      "loss": 0.828,
+      "step": 2535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807566207448919e-05,
+      "loss": 0.2583,
+      "step": 2536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802328279446521e-05,
+      "loss": 0.7607,
+      "step": 2537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797089516944233e-05,
+      "loss": 0.8225,
+      "step": 2538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791849921502091e-05,
+      "loss": 0.8107,
+      "step": 2539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786609494680387e-05,
+      "loss": 0.7384,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781368238039655e-05,
+      "loss": 0.8722,
+      "step": 2541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776126153140683e-05,
+      "loss": 0.7831,
+      "step": 2542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770883241544496e-05,
+      "loss": 0.3169,
+      "step": 2543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765639504812375e-05,
+      "loss": 0.8071,
+      "step": 2544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276039494450584e-05,
+      "loss": 0.6907,
+      "step": 2545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755149562186656e-05,
+      "loss": 0.7334,
+      "step": 2546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749903359416842e-05,
+      "loss": 0.7949,
+      "step": 2547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274465633775865e-05,
+      "loss": 0.2743,
+      "step": 2548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739408498774577e-05,
+      "loss": 0.7907,
+      "step": 2549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734159844027377e-05,
+      "loss": 0.8106,
+      "step": 2550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728910375080021e-05,
+      "loss": 0.8775,
+      "step": 2551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723660093495753e-05,
+      "loss": 0.2646,
+      "step": 2552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718409000838039e-05,
+      "loss": 0.8213,
+      "step": 2553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271315709867059e-05,
+      "loss": 0.7314,
+      "step": 2554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707904388557362e-05,
+      "loss": 0.8561,
+      "step": 2555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702650872062545e-05,
+      "loss": 0.7474,
+      "step": 2556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697396550750582e-05,
+      "loss": 0.7727,
+      "step": 2557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692141426186137e-05,
+      "loss": 0.2694,
+      "step": 2558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686885499934134e-05,
+      "loss": 0.8094,
+      "step": 2559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681628773559713e-05,
+      "loss": 0.2905,
+      "step": 2560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676371248628275e-05,
+      "loss": 0.7692,
+      "step": 2561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267111292670544e-05,
+      "loss": 0.7211,
+      "step": 2562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665853809357081e-05,
+      "loss": 0.6896,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660593898149296e-05,
+      "loss": 0.7842,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655333194648423e-05,
+      "loss": 0.8249,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650071700421036e-05,
+      "loss": 0.792,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264480941703395e-05,
+      "loss": 0.8006,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639546346054205e-05,
+      "loss": 0.7899,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634282489049086e-05,
+      "loss": 0.7806,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26290178475861e-05,
+      "loss": 0.8083,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623752423233002e-05,
+      "loss": 0.8367,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618486217557766e-05,
+      "loss": 0.7534,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613219232128608e-05,
+      "loss": 0.8161,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607951468513972e-05,
+      "loss": 0.7833,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260268292828254e-05,
+      "loss": 0.7212,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597413613003214e-05,
+      "loss": 0.7328,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592143524245132e-05,
+      "loss": 0.7281,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586872663577672e-05,
+      "loss": 0.8458,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581601032570427e-05,
+      "loss": 0.7785,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257632863279323e-05,
+      "loss": 0.8275,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571055465816131e-05,
+      "loss": 0.7233,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565781533209423e-05,
+      "loss": 0.7982,
+      "step": 2582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560506836543617e-05,
+      "loss": 0.806,
+      "step": 2583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255523137738946e-05,
+      "loss": 0.8195,
+      "step": 2584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549955157317912e-05,
+      "loss": 0.7876,
+      "step": 2585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544678177900175e-05,
+      "loss": 0.7878,
+      "step": 2586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539400440707672e-05,
+      "loss": 0.8239,
+      "step": 2587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253412194731204e-05,
+      "loss": 0.7855,
+      "step": 2588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528842699285162e-05,
+      "loss": 0.7285,
+      "step": 2589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523562698199127e-05,
+      "loss": 0.8598,
+      "step": 2590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518281945626266e-05,
+      "loss": 0.2896,
+      "step": 2591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513000443139112e-05,
+      "loss": 0.7918,
+      "step": 2592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507718192310443e-05,
+      "loss": 0.8461,
+      "step": 2593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502435194713245e-05,
+      "loss": 0.2561,
+      "step": 2594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497151451920735e-05,
+      "loss": 0.7273,
+      "step": 2595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491866965506345e-05,
+      "loss": 0.8804,
+      "step": 2596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486581737043735e-05,
+      "loss": 0.7774,
+      "step": 2597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248129576810678e-05,
+      "loss": 0.7675,
+      "step": 2598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476009060269577e-05,
+      "loss": 0.716,
+      "step": 2599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470721615106454e-05,
+      "loss": 0.7927,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465433434191938e-05,
+      "loss": 0.7832,
+      "step": 2601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460144519100794e-05,
+      "loss": 0.8017,
+      "step": 2602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.7692,
+      "step": 2603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244956449268873e-05,
+      "loss": 0.8069,
+      "step": 2604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444273384518418e-05,
+      "loss": 0.806,
+      "step": 2605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438981548472682e-05,
+      "loss": 0.7397,
+      "step": 2606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433688986127373e-05,
+      "loss": 0.8725,
+      "step": 2607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428395699058549e-05,
+      "loss": 0.7945,
+      "step": 2608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423101688842488e-05,
+      "loss": 0.8729,
+      "step": 2609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417806957055684e-05,
+      "loss": 0.8105,
+      "step": 2610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412511505274845e-05,
+      "loss": 0.8441,
+      "step": 2611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407215335076894e-05,
+      "loss": 0.7563,
+      "step": 2612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401918448038965e-05,
+      "loss": 0.7572,
+      "step": 2613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239662084573841e-05,
+      "loss": 0.8206,
+      "step": 2614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391322529752789e-05,
+      "loss": 0.7721,
+      "step": 2615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238602350165988e-05,
+      "loss": 0.6918,
+      "step": 2616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380723763037668e-05,
+      "loss": 0.8062,
+      "step": 2617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375423315464355e-05,
+      "loss": 0.7691,
+      "step": 2618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237012216051835e-05,
+      "loss": 0.7575,
+      "step": 2619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364820299778266e-05,
+      "loss": 0.7805,
+      "step": 2620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359517734822948e-05,
+      "loss": 0.8797,
+      "step": 2621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354214467231422e-05,
+      "loss": 0.6804,
+      "step": 2622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348910498582951e-05,
+      "loss": 0.7797,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343605830456982e-05,
+      "loss": 0.7654,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338300464433193e-05,
+      "loss": 0.7551,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233299440209145e-05,
+      "loss": 0.7437,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327687645011832e-05,
+      "loss": 0.7994,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322380194774641e-05,
+      "loss": 0.8018,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231707205296036e-05,
+      "loss": 0.7934,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23117632211497e-05,
+      "loss": 0.7802,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306453700923555e-05,
+      "loss": 0.7729,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301143493863052e-05,
+      "loss": 0.8402,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295832601549498e-05,
+      "loss": 0.7531,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290521025564413e-05,
+      "loss": 0.7039,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285208767489522e-05,
+      "loss": 0.7403,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279895828906756e-05,
+      "loss": 0.8297,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274582211398241e-05,
+      "loss": 0.8079,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269267916546308e-05,
+      "loss": 0.7275,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226395294593349e-05,
+      "loss": 0.7753,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258637301142526e-05,
+      "loss": 0.7849,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253320983756352e-05,
+      "loss": 0.8041,
+      "step": 2641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248003995358098e-05,
+      "loss": 0.8353,
+      "step": 2642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242686337531102e-05,
+      "loss": 0.7495,
+      "step": 2643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237368011858902e-05,
+      "loss": 0.8266,
+      "step": 2644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232049019925227e-05,
+      "loss": 0.7887,
+      "step": 2645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226729363314012e-05,
+      "loss": 0.8614,
+      "step": 2646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221409043609389e-05,
+      "loss": 0.7841,
+      "step": 2647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216088062395681e-05,
+      "loss": 0.7897,
+      "step": 2648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221076642125742e-05,
+      "loss": 0.7668,
+      "step": 2649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205444121779318e-05,
+      "loss": 0.2766,
+      "step": 2650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200121165546299e-05,
+      "loss": 0.7357,
+      "step": 2651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194797554143471e-05,
+      "loss": 0.7993,
+      "step": 2652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189473289156149e-05,
+      "loss": 0.7217,
+      "step": 2653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184148372169826e-05,
+      "loss": 0.7894,
+      "step": 2654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178822804770206e-05,
+      "loss": 0.7894,
+      "step": 2655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173496588543175e-05,
+      "loss": 0.7832,
+      "step": 2656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168169725074821e-05,
+      "loss": 0.2832,
+      "step": 2657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162842215951419e-05,
+      "loss": 0.7374,
+      "step": 2658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157514062759432e-05,
+      "loss": 0.7641,
+      "step": 2659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215218526708553e-05,
+      "loss": 0.2666,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146855830516554e-05,
+      "loss": 0.812,
+      "step": 2661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141525754639558e-05,
+      "loss": 0.7385,
+      "step": 2662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136195041041766e-05,
+      "loss": 0.7632,
+      "step": 2663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130863691310603e-05,
+      "loss": 0.7563,
+      "step": 2664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125531707033684e-05,
+      "loss": 0.7086,
+      "step": 2665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120199089798808e-05,
+      "loss": 0.7662,
+      "step": 2666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114865841193965e-05,
+      "loss": 0.7255,
+      "step": 2667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109531962807333e-05,
+      "loss": 0.7792,
+      "step": 2668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104197456227275e-05,
+      "loss": 0.7654,
+      "step": 2669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098862323042348e-05,
+      "loss": 0.8456,
+      "step": 2670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209352656484129e-05,
+      "loss": 0.8106,
+      "step": 2671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088190183213021e-05,
+      "loss": 0.8107,
+      "step": 2672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082853179746657e-05,
+      "loss": 0.8189,
+      "step": 2673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077515556031492e-05,
+      "loss": 0.7883,
+      "step": 2674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072177313657005e-05,
+      "loss": 0.7959,
+      "step": 2675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066838454212866e-05,
+      "loss": 0.8536,
+      "step": 2676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061498979288916e-05,
+      "loss": 0.7491,
+      "step": 2677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056158890475194e-05,
+      "loss": 0.7996,
+      "step": 2678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050818189361912e-05,
+      "loss": 0.7792,
+      "step": 2679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045476877539465e-05,
+      "loss": 0.865,
+      "step": 2680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040134956598434e-05,
+      "loss": 0.802,
+      "step": 2681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034792428129578e-05,
+      "loss": 0.6874,
+      "step": 2682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029449293723841e-05,
+      "loss": 0.7794,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202410555497234e-05,
+      "loss": 0.7669,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018761213466381e-05,
+      "loss": 0.7865,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013416270797443e-05,
+      "loss": 0.6993,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008070728557186e-05,
+      "loss": 0.7685,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200272458833745e-05,
+      "loss": 0.7898,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997377851730248e-05,
+      "loss": 0.8009,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199203052032778e-05,
+      "loss": 0.8463,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986682595722413e-05,
+      "loss": 0.8061,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981334079506696e-05,
+      "loss": 0.8286,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975984973273353e-05,
+      "loss": 0.8192,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970635278615289e-05,
+      "loss": 0.7127,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965284997125568e-05,
+      "loss": 0.7893,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959934130397453e-05,
+      "loss": 0.7894,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954582680024359e-05,
+      "loss": 0.8171,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949230647599888e-05,
+      "loss": 0.2781,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943878034717808e-05,
+      "loss": 0.7762,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938524842972068e-05,
+      "loss": 0.7305,
+      "step": 2700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1933171073956781e-05,
+      "loss": 0.8287,
+      "step": 2701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927816729266235e-05,
+      "loss": 0.7099,
+      "step": 2702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922461810494893e-05,
+      "loss": 0.7253,
+      "step": 2703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917106319237386e-05,
+      "loss": 0.7834,
+      "step": 2704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911750257088513e-05,
+      "loss": 0.769,
+      "step": 2705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906393625643244e-05,
+      "loss": 0.7976,
+      "step": 2706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901036426496724e-05,
+      "loss": 0.7649,
+      "step": 2707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895678661244258e-05,
+      "loss": 0.8416,
+      "step": 2708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189032033148133e-05,
+      "loss": 0.7043,
+      "step": 2709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884961438803582e-05,
+      "loss": 0.2661,
+      "step": 2710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187960198480683e-05,
+      "loss": 0.8224,
+      "step": 2711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874241971087058e-05,
+      "loss": 0.711,
+      "step": 2712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868881399240408e-05,
+      "loss": 0.8006,
+      "step": 2713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18635202708632e-05,
+      "loss": 0.2967,
+      "step": 2714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858158587551913e-05,
+      "loss": 0.8193,
+      "step": 2715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852796350903189e-05,
+      "loss": 0.836,
+      "step": 2716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847433562513841e-05,
+      "loss": 0.8125,
+      "step": 2717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842070223980842e-05,
+      "loss": 0.778,
+      "step": 2718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836706336901332e-05,
+      "loss": 0.7932,
+      "step": 2719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183134190287261e-05,
+      "loss": 0.7399,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825976923492142e-05,
+      "loss": 0.7977,
+      "step": 2721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1820611400357556e-05,
+      "loss": 0.7803,
+      "step": 2722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815245335066633e-05,
+      "loss": 0.7977,
+      "step": 2723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809878729217337e-05,
+      "loss": 0.7854,
+      "step": 2724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804511584407763e-05,
+      "loss": 0.2773,
+      "step": 2725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799143902236198e-05,
+      "loss": 0.7043,
+      "step": 2726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793775684301063e-05,
+      "loss": 0.8176,
+      "step": 2727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788406932200955e-05,
+      "loss": 0.7895,
+      "step": 2728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783037647534621e-05,
+      "loss": 0.8031,
+      "step": 2729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777667831900967e-05,
+      "loss": 0.7757,
+      "step": 2730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772297486899069e-05,
+      "loss": 0.7512,
+      "step": 2731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766926614128142e-05,
+      "loss": 0.7738,
+      "step": 2732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176155521518758e-05,
+      "loss": 0.7633,
+      "step": 2733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756183291676903e-05,
+      "loss": 0.8016,
+      "step": 2734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750810845195826e-05,
+      "loss": 0.7706,
+      "step": 2735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745437877344186e-05,
+      "loss": 0.8163,
+      "step": 2736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174006438972199e-05,
+      "loss": 0.8232,
+      "step": 2737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734690383929404e-05,
+      "loss": 0.8627,
+      "step": 2738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729315861566737e-05,
+      "loss": 0.715,
+      "step": 2739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723940824234462e-05,
+      "loss": 0.7777,
+      "step": 2740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718565273533198e-05,
+      "loss": 0.8296,
+      "step": 2741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171318921106372e-05,
+      "loss": 0.7781,
+      "step": 2742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707812638426957e-05,
+      "loss": 0.773,
+      "step": 2743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702435557223988e-05,
+      "loss": 0.8661,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169705796905604e-05,
+      "loss": 0.8165,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691679875524499e-05,
+      "loss": 0.8066,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686301278230895e-05,
+      "loss": 0.2396,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680922178776912e-05,
+      "loss": 0.8022,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675542578764379e-05,
+      "loss": 0.27,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670162479795277e-05,
+      "loss": 0.783,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664781883471734e-05,
+      "loss": 0.7809,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659400791396034e-05,
+      "loss": 0.7715,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165401920517059e-05,
+      "loss": 0.7857,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648637126397989e-05,
+      "loss": 0.7559,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643254556680934e-05,
+      "loss": 0.778,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637871497622304e-05,
+      "loss": 0.7783,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632487950825097e-05,
+      "loss": 0.829,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162710391789248e-05,
+      "loss": 0.7089,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621719400427751e-05,
+      "loss": 0.9216,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616334400034348e-05,
+      "loss": 0.8096,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161094891831587e-05,
+      "loss": 0.7751,
+      "step": 2761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160556295687604e-05,
+      "loss": 0.2904,
+      "step": 2762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600176517318742e-05,
+      "loss": 0.8127,
+      "step": 2763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594789601247982e-05,
+      "loss": 0.8683,
+      "step": 2764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158940221026793e-05,
+      "loss": 0.7457,
+      "step": 2765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584014345982882e-05,
+      "loss": 0.6804,
+      "step": 2766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157862600999728e-05,
+      "loss": 0.7672,
+      "step": 2767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573237203915705e-05,
+      "loss": 0.2635,
+      "step": 2768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156784792934288e-05,
+      "loss": 0.7819,
+      "step": 2769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156245818788367e-05,
+      "loss": 0.2649,
+      "step": 2770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155706798114307e-05,
+      "loss": 0.7144,
+      "step": 2771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551677310726222e-05,
+      "loss": 0.8071,
+      "step": 2772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546286178238404e-05,
+      "loss": 0.7844,
+      "step": 2773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540894585285027e-05,
+      "loss": 0.7806,
+      "step": 2774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535502533471647e-05,
+      "loss": 0.7577,
+      "step": 2775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530110024403953e-05,
+      "loss": 0.7237,
+      "step": 2776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524717059687767e-05,
+      "loss": 0.3114,
+      "step": 2777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519323640929051e-05,
+      "loss": 0.7935,
+      "step": 2778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15139297697339e-05,
+      "loss": 0.8446,
+      "step": 2779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508535447708547e-05,
+      "loss": 0.7913,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150314067645935e-05,
+      "loss": 0.2378,
+      "step": 2781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497745457592817e-05,
+      "loss": 0.8148,
+      "step": 2782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492349792715571e-05,
+      "loss": 0.8542,
+      "step": 2783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486953683434383e-05,
+      "loss": 0.8127,
+      "step": 2784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481557131356144e-05,
+      "loss": 0.7938,
+      "step": 2785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476160138087888e-05,
+      "loss": 0.7828,
+      "step": 2786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470762705236774e-05,
+      "loss": 0.7642,
+      "step": 2787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465364834410088e-05,
+      "loss": 0.7626,
+      "step": 2788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459966527215262e-05,
+      "loss": 0.6966,
+      "step": 2789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454567785259836e-05,
+      "loss": 0.785,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449168610151498e-05,
+      "loss": 0.7286,
+      "step": 2791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443769003498054e-05,
+      "loss": 0.7818,
+      "step": 2792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438368966907445e-05,
+      "loss": 0.8503,
+      "step": 2793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432968501987735e-05,
+      "loss": 0.8269,
+      "step": 2794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427567610347124e-05,
+      "loss": 0.7257,
+      "step": 2795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422166293593921e-05,
+      "loss": 0.8014,
+      "step": 2796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416764553336587e-05,
+      "loss": 0.7892,
+      "step": 2797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141136239118369e-05,
+      "loss": 0.7697,
+      "step": 2798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405959808743927e-05,
+      "loss": 0.7666,
+      "step": 2799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400556807626124e-05,
+      "loss": 0.7608,
+      "step": 2800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395153389439232e-05,
+      "loss": 0.8069,
+      "step": 2801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389749555792325e-05,
+      "loss": 0.8411,
+      "step": 2802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384345308294596e-05,
+      "loss": 0.7881,
+      "step": 2803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378940648555367e-05,
+      "loss": 0.7956,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373535578184083e-05,
+      "loss": 0.7613,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368130098790309e-05,
+      "loss": 0.6989,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362724211983732e-05,
+      "loss": 0.7811,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357317919374157e-05,
+      "loss": 0.7926,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135191122257152e-05,
+      "loss": 0.7808,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346504123185866e-05,
+      "loss": 0.8178,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341096622827366e-05,
+      "loss": 0.2652,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335688723106306e-05,
+      "loss": 0.7117,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13302804256331e-05,
+      "loss": 0.8859,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132487173201827e-05,
+      "loss": 0.7454,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131946264387246e-05,
+      "loss": 0.8426,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314053162806435e-05,
+      "loss": 0.7475,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308643290431073e-05,
+      "loss": 0.7659,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303233028357368e-05,
+      "loss": 0.7986,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297822378196432e-05,
+      "loss": 0.7786,
+      "step": 2819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129241134155949e-05,
+      "loss": 0.7544,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128699992005789e-05,
+      "loss": 0.7404,
+      "step": 2821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281588115303081e-05,
+      "loss": 0.7727,
+      "step": 2822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127617592890664e-05,
+      "loss": 0.8426,
+      "step": 2823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1270763362480248e-05,
+      "loss": 0.7828,
+      "step": 2824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265350417635706e-05,
+      "loss": 0.8467,
+      "step": 2825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259937095984918e-05,
+      "loss": 0.8335,
+      "step": 2826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254523399139914e-05,
+      "loss": 0.2445,
+      "step": 2827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249109328712824e-05,
+      "loss": 0.7499,
+      "step": 2828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243694886315895e-05,
+      "loss": 0.813,
+      "step": 2829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123828007356148e-05,
+      "loss": 0.7849,
+      "step": 2830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123286489206205e-05,
+      "loss": 0.2987,
+      "step": 2831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227449343430179e-05,
+      "loss": 0.733,
+      "step": 2832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222033429278551e-05,
+      "loss": 0.7625,
+      "step": 2833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1216617151219969e-05,
+      "loss": 0.7874,
+      "step": 2834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211200510867321e-05,
+      "loss": 0.691,
+      "step": 2835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205783509833634e-05,
+      "loss": 0.8211,
+      "step": 2836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200366149732015e-05,
+      "loss": 0.825,
+      "step": 2837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194948432175696e-05,
+      "loss": 0.7521,
+      "step": 2838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189530358778005e-05,
+      "loss": 0.7966,
+      "step": 2839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184111931152377e-05,
+      "loss": 0.8253,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178693150912361e-05,
+      "loss": 0.7416,
+      "step": 2841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11732740196716e-05,
+      "loss": 0.2304,
+      "step": 2842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167854539043849e-05,
+      "loss": 0.8463,
+      "step": 2843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162434710642965e-05,
+      "loss": 0.2871,
+      "step": 2844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157014536082907e-05,
+      "loss": 0.8389,
+      "step": 2845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151594016977737e-05,
+      "loss": 0.7556,
+      "step": 2846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146173154941626e-05,
+      "loss": 0.8236,
+      "step": 2847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140751951588837e-05,
+      "loss": 0.2668,
+      "step": 2848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113533040853374e-05,
+      "loss": 0.8544,
+      "step": 2849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112990852739081e-05,
+      "loss": 0.8208,
+      "step": 2850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124486309774611e-05,
+      "loss": 0.8073,
+      "step": 2851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119063757299822e-05,
+      "loss": 0.7597,
+      "step": 2852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113640871581213e-05,
+      "loss": 0.7903,
+      "step": 2853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108217654233649e-05,
+      "loss": 0.8064,
+      "step": 2854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102794106872105e-05,
+      "loss": 0.7902,
+      "step": 2855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097370231111647e-05,
+      "loss": 0.8057,
+      "step": 2856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091946028567442e-05,
+      "loss": 0.2636,
+      "step": 2857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086521500854746e-05,
+      "loss": 0.7737,
+      "step": 2858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081096649588929e-05,
+      "loss": 0.2696,
+      "step": 2859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107567147638544e-05,
+      "loss": 0.7989,
+      "step": 2860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070245982859828e-05,
+      "loss": 0.7007,
+      "step": 2861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064820170627751e-05,
+      "loss": 0.7422,
+      "step": 2862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105939404130494e-05,
+      "loss": 0.781,
+      "step": 2863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105396759650724e-05,
+      "loss": 0.7294,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048540837850572e-05,
+      "loss": 0.7821,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043113766950973e-05,
+      "loss": 0.758,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103768638542455e-05,
+      "loss": 0.8612,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103225869488751e-05,
+      "loss": 0.8284,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026830696956163e-05,
+      "loss": 0.7925,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021402393246899e-05,
+      "loss": 0.7764,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10159737853762e-05,
+      "loss": 0.7589,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010544874960641e-05,
+      "loss": 0.7941,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005115663616887e-05,
+      "loss": 0.2516,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999686152961693e-05,
+      "loss": 0.7672,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994256344611902e-05,
+      "loss": 0.7678,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988826240184447e-05,
+      "loss": 0.8249,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983395841296349e-05,
+      "loss": 0.7657,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977965149564714e-05,
+      "loss": 0.698,
+      "step": 2878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0972534166606734e-05,
+      "loss": 0.7114,
+      "step": 2879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967102894039699e-05,
+      "loss": 0.7591,
+      "step": 2880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096167133348097e-05,
+      "loss": 0.7754,
+      "step": 2881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956239486548004e-05,
+      "loss": 0.8317,
+      "step": 2882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950807354858339e-05,
+      "loss": 0.7986,
+      "step": 2883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945374940029598e-05,
+      "loss": 0.7582,
+      "step": 2884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939942243679495e-05,
+      "loss": 0.7779,
+      "step": 2885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934509267425814e-05,
+      "loss": 0.7448,
+      "step": 2886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929076012886437e-05,
+      "loss": 0.7739,
+      "step": 2887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923642481679316e-05,
+      "loss": 0.7042,
+      "step": 2888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918208675422497e-05,
+      "loss": 0.7118,
+      "step": 2889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912774595734102e-05,
+      "loss": 0.8369,
+      "step": 2890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907340244232332e-05,
+      "loss": 0.7806,
+      "step": 2891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901905622535477e-05,
+      "loss": 0.7836,
+      "step": 2892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896470732261897e-05,
+      "loss": 0.8277,
+      "step": 2893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089103557503004e-05,
+      "loss": 0.7946,
+      "step": 2894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885600152458429e-05,
+      "loss": 0.7755,
+      "step": 2895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880164466165675e-05,
+      "loss": 0.2933,
+      "step": 2896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874728517770447e-05,
+      "loss": 0.8342,
+      "step": 2897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086929230889152e-05,
+      "loss": 0.256,
+      "step": 2898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863855841147723e-05,
+      "loss": 0.7463,
+      "step": 2899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858419116157975e-05,
+      "loss": 0.7263,
+      "step": 2900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852982135541265e-05,
+      "loss": 0.866,
+      "step": 2901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847544900916665e-05,
+      "loss": 0.6963,
+      "step": 2902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842107413903314e-05,
+      "loss": 0.6755,
+      "step": 2903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083666967612043e-05,
+      "loss": 0.7957,
+      "step": 2904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831231689187311e-05,
+      "loss": 0.7935,
+      "step": 2905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.831,
+      "step": 2906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820354974347908e-05,
+      "loss": 0.8014,
+      "step": 2907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814916249680579e-05,
+      "loss": 0.874,
+      "step": 2908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809477282340924e-05,
+      "loss": 0.7827,
+      "step": 2909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08040380739486e-05,
+      "loss": 0.8243,
+      "step": 2910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798598626123343e-05,
+      "loss": 0.7324,
+      "step": 2911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793158940484951e-05,
+      "loss": 0.2955,
+      "step": 2912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07877190186533e-05,
+      "loss": 0.7757,
+      "step": 2913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782278862248332e-05,
+      "loss": 0.7637,
+      "step": 2914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776838472890065e-05,
+      "loss": 0.8287,
+      "step": 2915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771397852198574e-05,
+      "loss": 0.8457,
+      "step": 2916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765957001794018e-05,
+      "loss": 0.8584,
+      "step": 2917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760515923296617e-05,
+      "loss": 0.862,
+      "step": 2918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755074618326655e-05,
+      "loss": 0.836,
+      "step": 2919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074963308850449e-05,
+      "loss": 0.7696,
+      "step": 2920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744191335450544e-05,
+      "loss": 0.7697,
+      "step": 2921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0738749360785308e-05,
+      "loss": 0.7617,
+      "step": 2922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733307166129333e-05,
+      "loss": 0.8019,
+      "step": 2923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727864753103245e-05,
+      "loss": 0.8252,
+      "step": 2924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722422123327722e-05,
+      "loss": 0.7859,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716979278423523e-05,
+      "loss": 0.7832,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711536220011456e-05,
+      "loss": 0.803,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706092949712403e-05,
+      "loss": 0.6882,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07006494691473e-05,
+      "loss": 0.7805,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695205779937154e-05,
+      "loss": 0.7533,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689761883703028e-05,
+      "loss": 0.8473,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684317782066054e-05,
+      "loss": 0.7633,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678873476647416e-05,
+      "loss": 0.2376,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673428969068365e-05,
+      "loss": 0.8493,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066798426095021e-05,
+      "loss": 0.7942,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662539353914316e-05,
+      "loss": 0.7747,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657094249582122e-05,
+      "loss": 0.8185,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651648949575106e-05,
+      "loss": 0.764,
+      "step": 2938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646203455514815e-05,
+      "loss": 0.767,
+      "step": 2939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640757769022853e-05,
+      "loss": 0.8146,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635311891720879e-05,
+      "loss": 0.8033,
+      "step": 2941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629865825230611e-05,
+      "loss": 0.7918,
+      "step": 2942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624419571173825e-05,
+      "loss": 0.2817,
+      "step": 2943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618973131172348e-05,
+      "loss": 0.7697,
+      "step": 2944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613526506848065e-05,
+      "loss": 0.7278,
+      "step": 2945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608079699822916e-05,
+      "loss": 0.8573,
+      "step": 2946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602632711718892e-05,
+      "loss": 0.7705,
+      "step": 2947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597185544158046e-05,
+      "loss": 0.8335,
+      "step": 2948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591738198762477e-05,
+      "loss": 0.7537,
+      "step": 2949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058629067715434e-05,
+      "loss": 0.7905,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580842980955841e-05,
+      "loss": 0.7426,
+      "step": 2951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575395111789242e-05,
+      "loss": 0.8317,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569947071276847e-05,
+      "loss": 0.7629,
+      "step": 2953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564498861041025e-05,
+      "loss": 0.826,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055905048270418e-05,
+      "loss": 0.8011,
+      "step": 2955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553601937888782e-05,
+      "loss": 0.7762,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548153228217335e-05,
+      "loss": 0.8554,
+      "step": 2957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542704355312407e-05,
+      "loss": 0.8197,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05372553207966e-05,
+      "loss": 0.8235,
+      "step": 2959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531806126292581e-05,
+      "loss": 0.7427,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526356773423042e-05,
+      "loss": 0.765,
+      "step": 2961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520907263810746e-05,
+      "loss": 0.7963,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515457599078489e-05,
+      "loss": 0.7411,
+      "step": 2963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510007780849112e-05,
+      "loss": 0.7767,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504557810745512e-05,
+      "loss": 0.7369,
+      "step": 2965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499107690390617e-05,
+      "loss": 0.7662,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493657421407419e-05,
+      "loss": 0.7072,
+      "step": 2967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488207005418935e-05,
+      "loss": 0.7601,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482756444048236e-05,
+      "loss": 0.791,
+      "step": 2969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477305738918433e-05,
+      "loss": 0.7736,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471854891652679e-05,
+      "loss": 0.8482,
+      "step": 2971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466403903874176e-05,
+      "loss": 0.8403,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046095277720616e-05,
+      "loss": 0.8047,
+      "step": 2973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455501513271914e-05,
+      "loss": 0.7461,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450050113694756e-05,
+      "loss": 0.7604,
+      "step": 2975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044459858009805e-05,
+      "loss": 0.7942,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439146914105196e-05,
+      "loss": 0.7943,
+      "step": 2977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433695117339638e-05,
+      "loss": 0.8128,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428243191424852e-05,
+      "loss": 0.8168,
+      "step": 2979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422791137984359e-05,
+      "loss": 0.7803,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417338958641712e-05,
+      "loss": 0.7415,
+      "step": 2981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411886655020508e-05,
+      "loss": 0.8224,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040643422874438e-05,
+      "loss": 0.7529,
+      "step": 2983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400981681436995e-05,
+      "loss": 0.7647,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395529014722048e-05,
+      "loss": 0.7596,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390076230223289e-05,
+      "loss": 0.7747,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384623329564482e-05,
+      "loss": 0.8304,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037917031436945e-05,
+      "loss": 0.7692,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373717186262018e-05,
+      "loss": 0.7483,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368263946866077e-05,
+      "loss": 0.7896,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362810597805526e-05,
+      "loss": 0.8097,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357357140704316e-05,
+      "loss": 0.7864,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0351903577186417e-05,
+      "loss": 0.7784,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346449908875833e-05,
+      "loss": 0.7673,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340996137396608e-05,
+      "loss": 0.7696,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335542264372802e-05,
+      "loss": 0.8098,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330088291428521e-05,
+      "loss": 0.7509,
+      "step": 2997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324634220187886e-05,
+      "loss": 0.7496,
+      "step": 2998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031918005227506e-05,
+      "loss": 0.7696,
+      "step": 2999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313725789314226e-05,
+      "loss": 0.8077,
+      "step": 3000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308271432929598e-05,
+      "loss": 0.7965,
+      "step": 3001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302816984745418e-05,
+      "loss": 0.7602,
+      "step": 3002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297362446385957e-05,
+      "loss": 0.8377,
+      "step": 3003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291907819475513e-05,
+      "loss": 0.7165,
+      "step": 3004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286453105638404e-05,
+      "loss": 0.7438,
+      "step": 3005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280998306498984e-05,
+      "loss": 0.7897,
+      "step": 3006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275543423681622e-05,
+      "loss": 0.2704,
+      "step": 3007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027008845881072e-05,
+      "loss": 0.7599,
+      "step": 3008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02646334135107e-05,
+      "loss": 0.7353,
+      "step": 3009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259178289406011e-05,
+      "loss": 0.7783,
+      "step": 3010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253723088121123e-05,
+      "loss": 0.801,
+      "step": 3011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024826781128053e-05,
+      "loss": 0.875,
+      "step": 3012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242812460508746e-05,
+      "loss": 0.7915,
+      "step": 3013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237357037430313e-05,
+      "loss": 0.7059,
+      "step": 3014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231901543669788e-05,
+      "loss": 0.7507,
+      "step": 3015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0226445980851756e-05,
+      "loss": 0.7572,
+      "step": 3016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022099035060081e-05,
+      "loss": 0.7721,
+      "step": 3017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215534654541583e-05,
+      "loss": 0.7412,
+      "step": 3018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210078894298704e-05,
+      "loss": 0.7853,
+      "step": 3019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204623071496842e-05,
+      "loss": 0.741,
+      "step": 3020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199167187760674e-05,
+      "loss": 0.7236,
+      "step": 3021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193711244714897e-05,
+      "loss": 0.8024,
+      "step": 3022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188255243984224e-05,
+      "loss": 0.769,
+      "step": 3023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182799187193386e-05,
+      "loss": 0.7345,
+      "step": 3024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177343075967137e-05,
+      "loss": 0.8123,
+      "step": 3025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171886911930233e-05,
+      "loss": 0.7193,
+      "step": 3026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166430696707464e-05,
+      "loss": 0.7021,
+      "step": 3027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160974431923616e-05,
+      "loss": 0.8083,
+      "step": 3028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155518119203511e-05,
+      "loss": 0.7863,
+      "step": 3029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150061760171963e-05,
+      "loss": 0.7175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144605356453819e-05,
+      "loss": 0.7938,
+      "step": 3031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139148909673924e-05,
+      "loss": 0.2667,
+      "step": 3032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133692421457146e-05,
+      "loss": 0.8496,
+      "step": 3033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128235893428362e-05,
+      "loss": 0.7774,
+      "step": 3034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012277932721246e-05,
+      "loss": 0.7812,
+      "step": 3035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011732272443434e-05,
+      "loss": 0.7427,
+      "step": 3036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111866086718916e-05,
+      "loss": 0.8255,
+      "step": 3037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106409415691105e-05,
+      "loss": 0.7683,
+      "step": 3038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100952712975842e-05,
+      "loss": 0.69,
+      "step": 3039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095495980198066e-05,
+      "loss": 0.7791,
+      "step": 3040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009003921898273e-05,
+      "loss": 0.6995,
+      "step": 3041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084582430954787e-05,
+      "loss": 0.8154,
+      "step": 3042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079125617739209e-05,
+      "loss": 0.7727,
+      "step": 3043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073668780960966e-05,
+      "loss": 0.7936,
+      "step": 3044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006821192224504e-05,
+      "loss": 0.7118,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062755043216419e-05,
+      "loss": 0.7507,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057298145500091e-05,
+      "loss": 0.7457,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051841230721065e-05,
+      "loss": 0.7544,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046384300504337e-05,
+      "loss": 0.7873,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040927356474918e-05,
+      "loss": 0.8095,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035470400257823e-05,
+      "loss": 0.7152,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030013433478065e-05,
+      "loss": 0.7841,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024556457760666e-05,
+      "loss": 0.2769,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019099474730646e-05,
+      "loss": 0.7861,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013642486013033e-05,
+      "loss": 0.8083,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008185493232848e-05,
+      "loss": 0.8013,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002728498015122e-05,
+      "loss": 0.737,
+      "step": 3057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99727150198488e-06,
+      "loss": 0.7338,
+      "step": 3058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991814506767154e-06,
+      "loss": 0.8137,
+      "step": 3059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986357513986972e-06,
+      "loss": 0.7687,
+      "step": 3060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980900525269357e-06,
+      "loss": 0.7749,
+      "step": 3061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975443542239334e-06,
+      "loss": 0.8107,
+      "step": 3062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969986566521938e-06,
+      "loss": 0.7165,
+      "step": 3063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96452959974218e-06,
+      "loss": 0.798,
+      "step": 3064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959072643525083e-06,
+      "loss": 0.6934,
+      "step": 3065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953615699495664e-06,
+      "loss": 0.7213,
+      "step": 3066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948158769278939e-06,
+      "loss": 0.772,
+      "step": 3067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94270185449991e-06,
+      "loss": 0.771,
+      "step": 3068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937244956783585e-06,
+      "loss": 0.8115,
+      "step": 3069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931788077754967e-06,
+      "loss": 0.6833,
+      "step": 3070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926331219039039e-06,
+      "loss": 0.7664,
+      "step": 3071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920874382260794e-06,
+      "loss": 0.7604,
+      "step": 3072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915417569045213e-06,
+      "loss": 0.7664,
+      "step": 3073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909960781017275e-06,
+      "loss": 0.7697,
+      "step": 3074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904504019801935e-06,
+      "loss": 0.7867,
+      "step": 3075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899047287024161e-06,
+      "loss": 0.7163,
+      "step": 3076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8935905843089e-06,
+      "loss": 0.7735,
+      "step": 3077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888133913281088e-06,
+      "loss": 0.7752,
+      "step": 3078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882677275565662e-06,
+      "loss": 0.7447,
+      "step": 3079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877220672787544e-06,
+      "loss": 0.7729,
+      "step": 3080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871764106571645e-06,
+      "loss": 0.7261,
+      "step": 3081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866307578542858e-06,
+      "loss": 0.7585,
+      "step": 3082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860851090326079e-06,
+      "loss": 0.787,
+      "step": 3083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855394643546184e-06,
+      "loss": 0.732,
+      "step": 3084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84993823982804e-06,
+      "loss": 0.7826,
+      "step": 3085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844481880796492e-06,
+      "loss": 0.7092,
+      "step": 3086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839025568076383e-06,
+      "loss": 0.7926,
+      "step": 3087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833569303292541e-06,
+      "loss": 0.7972,
+      "step": 3088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82811308806977e-06,
+      "loss": 0.793,
+      "step": 3089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822656924032867e-06,
+      "loss": 0.8069,
+      "step": 3090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817200812806616e-06,
+      "loss": 0.772,
+      "step": 3091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81174475601578e-06,
+      "loss": 0.8151,
+      "step": 3092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806288755285107e-06,
+      "loss": 0.7662,
+      "step": 3093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800832812239326e-06,
+      "loss": 0.7762,
+      "step": 3094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79537692850316e-06,
+      "loss": 0.7171,
+      "step": 3095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789921105701297e-06,
+      "loss": 0.7512,
+      "step": 3096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784465345458419e-06,
+      "loss": 0.8282,
+      "step": 3097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779009649399188e-06,
+      "loss": 0.7106,
+      "step": 3098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773554019148249e-06,
+      "loss": 0.8164,
+      "step": 3099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768098456330213e-06,
+      "loss": 0.7767,
+      "step": 3100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762642962569687e-06,
+      "loss": 0.2898,
+      "step": 3101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757187539491259e-06,
+      "loss": 0.2594,
+      "step": 3102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751732188719474e-06,
+      "loss": 0.8166,
+      "step": 3103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74627691187888e-06,
+      "loss": 0.7874,
+      "step": 3104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740821710593989e-06,
+      "loss": 0.8606,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735366586489303e-06,
+      "loss": 0.2549,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729911541189282e-06,
+      "loss": 0.8141,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724456576318383e-06,
+      "loss": 0.8356,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719001693501023e-06,
+      "loss": 0.8087,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713546894361598e-06,
+      "loss": 0.713,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70809218052449e-06,
+      "loss": 0.7578,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702637553614045e-06,
+      "loss": 0.7844,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697183015254587e-06,
+      "loss": 0.7602,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691728567070405e-06,
+      "loss": 0.6738,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686274210685778e-06,
+      "loss": 0.7321,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680819947724943e-06,
+      "loss": 0.7989,
+      "step": 3116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675365779812118e-06,
+      "loss": 0.2653,
+      "step": 3117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66991170857148e-06,
+      "loss": 0.8042,
+      "step": 3118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6644577356272e-06,
+      "loss": 0.7842,
+      "step": 3119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659003862603395e-06,
+      "loss": 0.7629,
+      "step": 3120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653550091124169e-06,
+      "loss": 0.7788,
+      "step": 3121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648096422813584e-06,
+      "loss": 0.8476,
+      "step": 3122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642642859295685e-06,
+      "loss": 0.8006,
+      "step": 3123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637189402194477e-06,
+      "loss": 0.8145,
+      "step": 3124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631736053133926e-06,
+      "loss": 0.7503,
+      "step": 3125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626282813737983e-06,
+      "loss": 0.7053,
+      "step": 3126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620829685630557e-06,
+      "loss": 0.8039,
+      "step": 3127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61537667043552e-06,
+      "loss": 0.8221,
+      "step": 3128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609923769776714e-06,
+      "loss": 0.7556,
+      "step": 3129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604470985277952e-06,
+      "loss": 0.2892,
+      "step": 3130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599018318563011e-06,
+      "loss": 0.7378,
+      "step": 3131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593565771255623e-06,
+      "loss": 0.7854,
+      "step": 3132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588113344979492e-06,
+      "loss": 0.8037,
+      "step": 3133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582661041358292e-06,
+      "loss": 0.8116,
+      "step": 3134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577208862015646e-06,
+      "loss": 0.7934,
+      "step": 3135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571756808575153e-06,
+      "loss": 0.2843,
+      "step": 3136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566304882660364e-06,
+      "loss": 0.7096,
+      "step": 3137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560853085894806e-06,
+      "loss": 0.7399,
+      "step": 3138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555401419901953e-06,
+      "loss": 0.7124,
+      "step": 3139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549949886305246e-06,
+      "loss": 0.7776,
+      "step": 3140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54449848672809e-06,
+      "loss": 0.774,
+      "step": 3141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539047222793841e-06,
+      "loss": 0.7932,
+      "step": 3142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533596096125826e-06,
+      "loss": 0.7824,
+      "step": 3143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52814510834732e-06,
+      "loss": 0.7783,
+      "step": 3144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522694261081573e-06,
+      "loss": 0.8182,
+      "step": 3145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517243555951768e-06,
+      "loss": 0.7945,
+      "step": 3146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511792994581068e-06,
+      "loss": 0.8287,
+      "step": 3147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506342578592584e-06,
+      "loss": 0.7225,
+      "step": 3148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500892309609384e-06,
+      "loss": 0.8073,
+      "step": 3149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495442189254491e-06,
+      "loss": 0.796,
+      "step": 3150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48999221915089e-06,
+      "loss": 0.796,
+      "step": 3151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484542400921518e-06,
+      "loss": 0.7531,
+      "step": 3152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479092736189255e-06,
+      "loss": 0.7644,
+      "step": 3153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47364322657696e-06,
+      "loss": 0.8665,
+      "step": 3154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468193873707425e-06,
+      "loss": 0.7662,
+      "step": 3155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462744679203401e-06,
+      "loss": 0.7118,
+      "step": 3156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457295644687596e-06,
+      "loss": 0.8578,
+      "step": 3157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451846771782666e-06,
+      "loss": 0.7691,
+      "step": 3158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446398062111221e-06,
+      "loss": 0.7562,
+      "step": 3159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440949517295822e-06,
+      "loss": 0.7569,
+      "step": 3160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435501138958976e-06,
+      "loss": 0.2699,
+      "step": 3161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430052928723153e-06,
+      "loss": 0.7758,
+      "step": 3162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424604888210762e-06,
+      "loss": 0.7468,
+      "step": 3163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41915701904416e-06,
+      "loss": 0.7913,
+      "step": 3164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41370932284566e-06,
+      "loss": 0.8044,
+      "step": 3165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408261801237524e-06,
+      "loss": 0.7513,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402814455841957e-06,
+      "loss": 0.7403,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39736728828111e-06,
+      "loss": 0.2459,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391920300177086e-06,
+      "loss": 0.7615,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386473493151938e-06,
+      "loss": 0.7421,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381026868827655e-06,
+      "loss": 0.7992,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375580428826175e-06,
+      "loss": 0.7925,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370134174769392e-06,
+      "loss": 0.751,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364688108279123e-06,
+      "loss": 0.7546,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35924223097715e-06,
+      "loss": 0.8674,
+      "step": 3175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.353796544485185e-06,
+      "loss": 0.8254,
+      "step": 3176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348351050424899e-06,
+      "loss": 0.8441,
+      "step": 3177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34290575041788e-06,
+      "loss": 0.7615,
+      "step": 3178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337460646085684e-06,
+      "loss": 0.8089,
+      "step": 3179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332015739049796e-06,
+      "loss": 0.8275,
+      "step": 3180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326571030931636e-06,
+      "loss": 0.7873,
+      "step": 3181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321126523352585e-06,
+      "loss": 0.7727,
+      "step": 3182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31568221793395e-06,
+      "loss": 0.7824,
+      "step": 3183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310238116296977e-06,
+      "loss": 0.7341,
+      "step": 3184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304794220062849e-06,
+      "loss": 0.7553,
+      "step": 3185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299350530852703e-06,
+      "loss": 0.7774,
+      "step": 3186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293907050287602e-06,
+      "loss": 0.7582,
+      "step": 3187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288463779988548e-06,
+      "loss": 0.7973,
+      "step": 3188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283020721576479e-06,
+      "loss": 0.8034,
+      "step": 3189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27757787667228e-06,
+      "loss": 0.8193,
+      "step": 3190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272135246896761e-06,
+      "loss": 0.7952,
+      "step": 3191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26669283387067e-06,
+      "loss": 0.8186,
+      "step": 3192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261250639214693e-06,
+      "loss": 0.7498,
+      "step": 3193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255808664549459e-06,
+      "loss": 0.823,
+      "step": 3194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250366911495513e-06,
+      "loss": 0.7535,
+      "step": 3195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244925381673348e-06,
+      "loss": 0.7695,
+      "step": 3196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239484076703385e-06,
+      "loss": 0.7696,
+      "step": 3197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234042998205984e-06,
+      "loss": 0.7824,
+      "step": 3198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228602147801427e-06,
+      "loss": 0.7671,
+      "step": 3199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223161527109938e-06,
+      "loss": 0.7669,
+      "step": 3200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217721137751668e-06,
+      "loss": 0.7972,
+      "step": 3201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212280981346704e-06,
+      "loss": 0.7551,
+      "step": 3202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206841059515052e-06,
+      "loss": 0.8028,
+      "step": 3203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201401373876659e-06,
+      "loss": 0.8048,
+      "step": 3204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195961926051404e-06,
+      "loss": 0.7867,
+      "step": 3205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19052271765908e-06,
+      "loss": 0.8042,
+      "step": 3206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185083750319423e-06,
+      "loss": 0.7364,
+      "step": 3207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179645025652092e-06,
+      "loss": 0.7605,
+      "step": 3208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.8176,
+      "step": 3209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16876831081269e-06,
+      "loss": 0.7913,
+      "step": 3210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163330323879571e-06,
+      "loss": 0.6424,
+      "step": 3211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157892586096693e-06,
+      "loss": 0.7674,
+      "step": 3212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152455099083339e-06,
+      "loss": 0.2775,
+      "step": 3213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147017864458736e-06,
+      "loss": 0.7731,
+      "step": 3214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141580883842027e-06,
+      "loss": 0.7872,
+      "step": 3215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136144158852282e-06,
+      "loss": 0.7392,
+      "step": 3216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130707691108483e-06,
+      "loss": 0.798,
+      "step": 3217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125271482229553e-06,
+      "loss": 0.8128,
+      "step": 3218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119835533834332e-06,
+      "loss": 0.7652,
+      "step": 3219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114399847541573e-06,
+      "loss": 0.7129,
+      "step": 3220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108964424969962e-06,
+      "loss": 0.7502,
+      "step": 3221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103529267738105e-06,
+      "loss": 0.8589,
+      "step": 3222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098094377464528e-06,
+      "loss": 0.814,
+      "step": 3223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09265975576767e-06,
+      "loss": 0.73,
+      "step": 3224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0872254042659e-06,
+      "loss": 0.7379,
+      "step": 3225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081791324577505e-06,
+      "loss": 0.761,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076357518320686e-06,
+      "loss": 0.8147,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070923987113567e-06,
+      "loss": 0.8198,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065490732574187e-06,
+      "loss": 0.7421,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06005775632051e-06,
+      "loss": 0.765,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054625059970403e-06,
+      "loss": 0.7947,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049192645141663e-06,
+      "loss": 0.7934,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043760513452003e-06,
+      "loss": 0.8324,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038328666519035e-06,
+      "loss": 0.7369,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032897105960306e-06,
+      "loss": 0.8276,
+      "step": 3235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027465833393266e-06,
+      "loss": 0.27,
+      "step": 3236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022034850435291e-06,
+      "loss": 0.7806,
+      "step": 3237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016604158703654e-06,
+      "loss": 0.7976,
+      "step": 3238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011173759815554e-06,
+      "loss": 0.7344,
+      "step": 3239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005743655388097e-06,
+      "loss": 0.7759,
+      "step": 3240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000313847038309e-06,
+      "loss": 0.8033,
+      "step": 3241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994884336383116e-06,
+      "loss": 0.8079,
+      "step": 3242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989455125039362e-06,
+      "loss": 0.686,
+      "step": 3243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984026214623807e-06,
+      "loss": 0.7307,
+      "step": 3244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978597606753106e-06,
+      "loss": 0.687,
+      "step": 3245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973169303043839e-06,
+      "loss": 0.7288,
+      "step": 3246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96774130511249e-06,
+      "loss": 0.6905,
+      "step": 3247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962313614575457e-06,
+      "loss": 0.7921,
+      "step": 3248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95688623304903e-06,
+      "loss": 0.7972,
+      "step": 3249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951459162149428e-06,
+      "loss": 0.7899,
+      "step": 3250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946032403492765e-06,
+      "loss": 0.6844,
+      "step": 3251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940605958695064e-06,
+      "loss": 0.7692,
+      "step": 3252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93517982937225e-06,
+      "loss": 0.7451,
+      "step": 3253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929754017140172e-06,
+      "loss": 0.8212,
+      "step": 3254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924328523614565e-06,
+      "loss": 0.7479,
+      "step": 3255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918903350411074e-06,
+      "loss": 0.7306,
+      "step": 3256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913478499145255e-06,
+      "loss": 0.7399,
+      "step": 3257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908053971432565e-06,
+      "loss": 0.7686,
+      "step": 3258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902629768888359e-06,
+      "loss": 0.2626,
+      "step": 3259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897205893127896e-06,
+      "loss": 0.2779,
+      "step": 3260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891782345766351e-06,
+      "loss": 0.7794,
+      "step": 3261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886359128418792e-06,
+      "loss": 0.2584,
+      "step": 3262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880936242700181e-06,
+      "loss": 0.7671,
+      "step": 3263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875513690225387e-06,
+      "loss": 0.7484,
+      "step": 3264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870091472609194e-06,
+      "loss": 0.7323,
+      "step": 3265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864669591466261e-06,
+      "loss": 0.7334,
+      "step": 3266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859248048411165e-06,
+      "loss": 0.7844,
+      "step": 3267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853826845058374e-06,
+      "loss": 0.7683,
+      "step": 3268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848405983022264e-06,
+      "loss": 0.7442,
+      "step": 3269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842985463917095e-06,
+      "loss": 0.7457,
+      "step": 3270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837565289357037e-06,
+      "loss": 0.7766,
+      "step": 3271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832145460956156e-06,
+      "loss": 0.7785,
+      "step": 3272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826725980328403e-06,
+      "loss": 0.8196,
+      "step": 3273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821306849087644e-06,
+      "loss": 0.7244,
+      "step": 3274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815888068847623e-06,
+      "loss": 0.737,
+      "step": 3275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810469641222001e-06,
+      "loss": 0.777,
+      "step": 3276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805051567824307e-06,
+      "loss": 0.7303,
+      "step": 3277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799633850267986e-06,
+      "loss": 0.7666,
+      "step": 3278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794216490166366e-06,
+      "loss": 0.7565,
+      "step": 3279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78879948913268e-06,
+      "loss": 0.8024,
+      "step": 3280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783382848780036e-06,
+      "loss": 0.7809,
+      "step": 3281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77796657072145e-06,
+      "loss": 0.8359,
+      "step": 3282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772550656569826e-06,
+      "loss": 0.7754,
+      "step": 3283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767135107937952e-06,
+      "loss": 0.7812,
+      "step": 3284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761719926438523e-06,
+      "loss": 0.7323,
+      "step": 3285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75630511368411e-06,
+      "loss": 0.6961,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75089067128718e-06,
+      "loss": 0.7376,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745476600860088e-06,
+      "loss": 0.789,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740062904015084e-06,
+      "loss": 0.7544,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734649582364298e-06,
+      "loss": 0.7889,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729236637519754e-06,
+      "loss": 0.7408,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72382407109336e-06,
+      "loss": 0.7459,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718411884696919e-06,
+      "loss": 0.7686,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713000079942113e-06,
+      "loss": 0.7649,
+      "step": 3294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707588658440511e-06,
+      "loss": 0.767,
+      "step": 3295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70217762180357e-06,
+      "loss": 0.6751,
+      "step": 3296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696766971642635e-06,
+      "loss": 0.7585,
+      "step": 3297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69135670956893e-06,
+      "loss": 0.7811,
+      "step": 3298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685946837193567e-06,
+      "loss": 0.7864,
+      "step": 3299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68053735612754e-06,
+      "loss": 0.7944,
+      "step": 3300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675128267981734e-06,
+      "loss": 0.7314,
+      "step": 3301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669719574366902e-06,
+      "loss": 0.7186,
+      "step": 3302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664311276893693e-06,
+      "loss": 0.7816,
+      "step": 3303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65890337717264e-06,
+      "loss": 0.7925,
+      "step": 3304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653495876814137e-06,
+      "loss": 0.786,
+      "step": 3305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648088777428482e-06,
+      "loss": 0.7348,
+      "step": 3306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.642682080625843e-06,
+      "loss": 0.782,
+      "step": 3307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637275788016273e-06,
+      "loss": 0.72,
+      "step": 3308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631869901209694e-06,
+      "loss": 0.7201,
+      "step": 3309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626464421815919e-06,
+      "loss": 0.856,
+      "step": 3310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621059351444638e-06,
+      "loss": 0.7788,
+      "step": 3311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615654691705408e-06,
+      "loss": 0.7654,
+      "step": 3312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610250444207679e-06,
+      "loss": 0.8038,
+      "step": 3313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604846610560771e-06,
+      "loss": 0.8211,
+      "step": 3314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599443192373881e-06,
+      "loss": 0.8283,
+      "step": 3315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594040191256076e-06,
+      "loss": 0.7033,
+      "step": 3316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588637608816314e-06,
+      "loss": 0.7629,
+      "step": 3317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583235446663414e-06,
+      "loss": 0.7919,
+      "step": 3318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57783370640608e-06,
+      "loss": 0.7265,
+      "step": 3319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572432389652881e-06,
+      "loss": 0.7891,
+      "step": 3320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567031498012267e-06,
+      "loss": 0.8343,
+      "step": 3321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561631033092558e-06,
+      "loss": 0.819,
+      "step": 3322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55623099650195e-06,
+      "loss": 0.7187,
+      "step": 3323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550831389848504e-06,
+      "loss": 0.7501,
+      "step": 3324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545432214740166e-06,
+      "loss": 0.7249,
+      "step": 3325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540033472784743e-06,
+      "loss": 0.2443,
+      "step": 3326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534635165589914e-06,
+      "loss": 0.7626,
+      "step": 3327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529237294763228e-06,
+      "loss": 0.6637,
+      "step": 3328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523839861912114e-06,
+      "loss": 0.7688,
+      "step": 3329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518442868643857e-06,
+      "loss": 0.2477,
+      "step": 3330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513046316565618e-06,
+      "loss": 0.7176,
+      "step": 3331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507650207284429e-06,
+      "loss": 0.7649,
+      "step": 3332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502254542407186e-06,
+      "loss": 0.2615,
+      "step": 3333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49685932354065e-06,
+      "loss": 0.7427,
+      "step": 3334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491464552291455e-06,
+      "loss": 0.7853,
+      "step": 3335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486070230266104e-06,
+      "loss": 0.8023,
+      "step": 3336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480676359070952e-06,
+      "loss": 0.8121,
+      "step": 3337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475282940312234e-06,
+      "loss": 0.8115,
+      "step": 3338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469889975596047e-06,
+      "loss": 0.7598,
+      "step": 3339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464497466528355e-06,
+      "loss": 0.6931,
+      "step": 3340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459105414714975e-06,
+      "loss": 0.7323,
+      "step": 3341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4537138217616e-06,
+      "loss": 0.8232,
+      "step": 3342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448322689273783e-06,
+      "loss": 0.8636,
+      "step": 3343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442932018856933e-06,
+      "loss": 0.8194,
+      "step": 3344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437541812116334e-06,
+      "loss": 0.8161,
+      "step": 3345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432152070657121e-06,
+      "loss": 0.8522,
+      "step": 3346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4267627960843e-06,
+      "loss": 0.7303,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421373990002722e-06,
+      "loss": 0.8024,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41598565401712e-06,
+      "loss": 0.6997,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410597789732074e-06,
+      "loss": 0.7592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40521039875202e-06,
+      "loss": 0.8167,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399823482681263e-06,
+      "loss": 0.8031,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39443704312396e-06,
+      "loss": 0.8119,
+      "step": 3353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.389051081684134e-06,
+      "loss": 0.8067,
+      "step": 3354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383665599965654e-06,
+      "loss": 0.7058,
+      "step": 3355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378280599572252e-06,
+      "loss": 0.7665,
+      "step": 3356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37289608210752e-06,
+      "loss": 0.7773,
+      "step": 3357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367512049174905e-06,
+      "loss": 0.7404,
+      "step": 3358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3621285023777e-06,
+      "loss": 0.7886,
+      "step": 3359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356745443319068e-06,
+      "loss": 0.8408,
+      "step": 3360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351362873602018e-06,
+      "loss": 0.7722,
+      "step": 3361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345980794829412e-06,
+      "loss": 0.7495,
+      "step": 3362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34059920860397e-06,
+      "loss": 0.7771,
+      "step": 3363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335218116528266e-06,
+      "loss": 0.7377,
+      "step": 3364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329837520204726e-06,
+      "loss": 0.7985,
+      "step": 3365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324457421235624e-06,
+      "loss": 0.8418,
+      "step": 3366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31907782122309e-06,
+      "loss": 0.7986,
+      "step": 3367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313698721769106e-06,
+      "loss": 0.7626,
+      "step": 3368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308320124475503e-06,
+      "loss": 0.7795,
+      "step": 3369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302942030943961e-06,
+      "loss": 0.7112,
+      "step": 3370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297564442776014e-06,
+      "loss": 0.7547,
+      "step": 3371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292187361573045e-06,
+      "loss": 0.7764,
+      "step": 3372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286810788936283e-06,
+      "loss": 0.7561,
+      "step": 3373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281434726466806e-06,
+      "loss": 0.7481,
+      "step": 3374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276059175765543e-06,
+      "loss": 0.8361,
+      "step": 3375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270684138433266e-06,
+      "loss": 0.7613,
+      "step": 3376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2653096160706e-06,
+      "loss": 0.7755,
+      "step": 3377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25993561027801e-06,
+      "loss": 0.7497,
+      "step": 3378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25456212265582e-06,
+      "loss": 0.8111,
+      "step": 3379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249189154804177e-06,
+      "loss": 0.7705,
+      "step": 3380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243816708323095e-06,
+      "loss": 0.7154,
+      "step": 3381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238444784812427e-06,
+      "loss": 0.7099,
+      "step": 3382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23307338587186e-06,
+      "loss": 0.7588,
+      "step": 3383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227702513100933e-06,
+      "loss": 0.8187,
+      "step": 3384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222332168099033e-06,
+      "loss": 0.8104,
+      "step": 3385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216962352465385e-06,
+      "loss": 0.766,
+      "step": 3386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211593067799048e-06,
+      "loss": 0.8151,
+      "step": 3387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206224315698939e-06,
+      "loss": 0.7166,
+      "step": 3388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200856097763807e-06,
+      "loss": 0.809,
+      "step": 3389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195488415592238e-06,
+      "loss": 0.252,
+      "step": 3390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190121270782668e-06,
+      "loss": 0.7865,
+      "step": 3391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184754664933368e-06,
+      "loss": 0.7713,
+      "step": 3392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17938859964245e-06,
+      "loss": 0.797,
+      "step": 3393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174023076507863e-06,
+      "loss": 0.6767,
+      "step": 3394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168658097127391e-06,
+      "loss": 0.827,
+      "step": 3395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16329366309867e-06,
+      "loss": 0.7764,
+      "step": 3396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15792977601916e-06,
+      "loss": 0.7528,
+      "step": 3397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152566437486162e-06,
+      "loss": 0.7943,
+      "step": 3398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147203649096811e-06,
+      "loss": 0.8325,
+      "step": 3399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14184141244809e-06,
+      "loss": 0.7577,
+      "step": 3400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136479729136801e-06,
+      "loss": 0.7607,
+      "step": 3401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131118600759593e-06,
+      "loss": 0.7676,
+      "step": 3402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125758028912944e-06,
+      "loss": 0.7757,
+      "step": 3403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120398015193171e-06,
+      "loss": 0.8033,
+      "step": 3404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11503856119642e-06,
+      "loss": 0.7794,
+      "step": 3405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10967966851867e-06,
+      "loss": 0.2775,
+      "step": 3406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104321338755745e-06,
+      "loss": 0.8041,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09896357350328e-06,
+      "loss": 0.8008,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09360637435676e-06,
+      "loss": 0.7832,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088249742911489e-06,
+      "loss": 0.8003,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082893680762619e-06,
+      "loss": 0.7734,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077538189505108e-06,
+      "loss": 0.7682,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072183270733767e-06,
+      "loss": 0.7771,
+      "step": 3413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066828926043224e-06,
+      "loss": 0.7472,
+      "step": 3414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061475157027935e-06,
+      "loss": 0.7246,
+      "step": 3415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056121965282194e-06,
+      "loss": 0.6657,
+      "step": 3416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050769352400117e-06,
+      "loss": 0.8081,
+      "step": 3417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045417319975646e-06,
+      "loss": 0.7227,
+      "step": 3418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04006586960255e-06,
+      "loss": 0.8212,
+      "step": 3419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034715002874432e-06,
+      "loss": 0.7622,
+      "step": 3420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029364721384718e-06,
+      "loss": 0.7953,
+      "step": 3421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02401502672665e-06,
+      "loss": 0.72,
+      "step": 3422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018665920493305e-06,
+      "loss": 0.7845,
+      "step": 3423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013317404277589e-06,
+      "loss": 0.8443,
+      "step": 3424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007969479672224e-06,
+      "loss": 0.7827,
+      "step": 3425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002622148269755e-06,
+      "loss": 0.7194,
+      "step": 3426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997275411662553e-06,
+      "loss": 0.7907,
+      "step": 3427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991929271442817e-06,
+      "loss": 0.7528,
+      "step": 3428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98658372920256e-06,
+      "loss": 0.805,
+      "step": 3429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981238786533622e-06,
+      "loss": 0.7648,
+      "step": 3430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97589444502766e-06,
+      "loss": 0.7972,
+      "step": 3431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970550706276164e-06,
+      "loss": 0.8427,
+      "step": 3432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965207571870425e-06,
+      "loss": 0.7819,
+      "step": 3433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959865043401567e-06,
+      "loss": 0.7113,
+      "step": 3434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954523122460537e-06,
+      "loss": 0.7648,
+      "step": 3435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949181810638091e-06,
+      "loss": 0.7605,
+      "step": 3436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943841109524808e-06,
+      "loss": 0.8357,
+      "step": 3437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938501020711084e-06,
+      "loss": 0.7522,
+      "step": 3438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93316154578714e-06,
+      "loss": 0.7415,
+      "step": 3439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927822686342997e-06,
+      "loss": 0.7469,
+      "step": 3440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922484443968511e-06,
+      "loss": 0.764,
+      "step": 3441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917146820253343e-06,
+      "loss": 0.7893,
+      "step": 3442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911809816786982e-06,
+      "loss": 0.7311,
+      "step": 3443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906473435158714e-06,
+      "loss": 0.7995,
+      "step": 3444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901137676957653e-06,
+      "loss": 0.7886,
+      "step": 3445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895802543772728e-06,
+      "loss": 0.7539,
+      "step": 3446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89046803719267e-06,
+      "loss": 0.7961,
+      "step": 3447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885134158806038e-06,
+      "loss": 0.7566,
+      "step": 3448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879800910201194e-06,
+      "loss": 0.7956,
+      "step": 3449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874468292966321e-06,
+      "loss": 0.7433,
+      "step": 3450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869136308689398e-06,
+      "loss": 0.7888,
+      "step": 3451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863804958958236e-06,
+      "loss": 0.7448,
+      "step": 3452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858474245360447e-06,
+      "loss": 0.7951,
+      "step": 3453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853144169483447e-06,
+      "loss": 0.2701,
+      "step": 3454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847814732914473e-06,
+      "loss": 0.7735,
+      "step": 3455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842485937240568e-06,
+      "loss": 0.7884,
+      "step": 3456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837157784048586e-06,
+      "loss": 0.7471,
+      "step": 3457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831830274925182e-06,
+      "loss": 0.7915,
+      "step": 3458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826503411456825e-06,
+      "loss": 0.7898,
+      "step": 3459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821177195229798e-06,
+      "loss": 0.7746,
+      "step": 3460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815851627830177e-06,
+      "loss": 0.81,
+      "step": 3461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810526710843855e-06,
+      "loss": 0.6415,
+      "step": 3462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80520244585653e-06,
+      "loss": 0.7482,
+      "step": 3463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799878834453704e-06,
+      "loss": 0.7602,
+      "step": 3464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794555878220685e-06,
+      "loss": 0.7167,
+      "step": 3465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789233578742583e-06,
+      "loss": 0.8281,
+      "step": 3466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783911937604322e-06,
+      "loss": 0.6709,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778590956390614e-06,
+      "loss": 0.676,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77327063668599e-06,
+      "loss": 0.8026,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767950980074775e-06,
+      "loss": 0.7819,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762631988141103e-06,
+      "loss": 0.7405,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757313662468901e-06,
+      "loss": 0.7303,
+      "step": 3472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751996004641907e-06,
+      "loss": 0.7931,
+      "step": 3473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74667901624365e-06,
+      "loss": 0.7761,
+      "step": 3474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741362698857475e-06,
+      "loss": 0.7443,
+      "step": 3475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736047054066512e-06,
+      "loss": 0.7705,
+      "step": 3476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730732083453695e-06,
+      "loss": 0.7797,
+      "step": 3477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725417788601766e-06,
+      "loss": 0.8328,
+      "step": 3478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720104171093247e-06,
+      "loss": 0.8034,
+      "step": 3479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71479123251048e-06,
+      "loss": 0.8302,
+      "step": 3480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709478974435587e-06,
+      "loss": 0.715,
+      "step": 3481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704167398450508e-06,
+      "loss": 0.7752,
+      "step": 3482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69885650613695e-06,
+      "loss": 0.8088,
+      "step": 3483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693546299076443e-06,
+      "loss": 0.6692,
+      "step": 3484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688236778850307e-06,
+      "loss": 0.8021,
+      "step": 3485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682927947039643e-06,
+      "loss": 0.8256,
+      "step": 3486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677619805225362e-06,
+      "loss": 0.8116,
+      "step": 3487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672312354988168e-06,
+      "loss": 0.7508,
+      "step": 3488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667005597908557e-06,
+      "loss": 0.7318,
+      "step": 3489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661699535566812e-06,
+      "loss": 0.7486,
+      "step": 3490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656394169543018e-06,
+      "loss": 0.7404,
+      "step": 3491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651089501417052e-06,
+      "loss": 0.8286,
+      "step": 3492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64578553276858e-06,
+      "loss": 0.8214,
+      "step": 3493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640482265177056e-06,
+      "loss": 0.8025,
+      "step": 3494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635179700221733e-06,
+      "loss": 0.7644,
+      "step": 3495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629877839481657e-06,
+      "loss": 0.7082,
+      "step": 3496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624576684535649e-06,
+      "loss": 0.8197,
+      "step": 3497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619276236962332e-06,
+      "loss": 0.7499,
+      "step": 3498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613976498340123e-06,
+      "loss": 0.726,
+      "step": 3499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6086774702472145e-06,
+      "loss": 0.7517,
+      "step": 3500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603379154261595e-06,
+      "loss": 0.7486,
+      "step": 3501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598081551961037e-06,
+      "loss": 0.7961,
+      "step": 3502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59278466492311e-06,
+      "loss": 0.757,
+      "step": 3503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587488494725157e-06,
+      "loss": 0.6924,
+      "step": 3504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582193042944319e-06,
+      "loss": 0.8356,
+      "step": 3505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5768983111575165e-06,
+      "loss": 0.7662,
+      "step": 3506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571604300941454e-06,
+      "loss": 0.7836,
+      "step": 3507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56631101387263e-06,
+      "loss": 0.8199,
+      "step": 3508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561018451527318e-06,
+      "loss": 0.2517,
+      "step": 3509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555726615481588e-06,
+      "loss": 0.7482,
+      "step": 3510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550435507311273e-06,
+      "loss": 0.763,
+      "step": 3511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.8133,
+      "step": 3512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539855480899207e-06,
+      "loss": 0.8195,
+      "step": 3513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534566565808065e-06,
+      "loss": 0.7622,
+      "step": 3514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529278384893548e-06,
+      "loss": 0.7442,
+      "step": 3515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5239909397304225e-06,
+      "loss": 0.6598,
+      "step": 3516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518704231893225e-06,
+      "loss": 0.6952,
+      "step": 3517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513418262956269e-06,
+      "loss": 0.7591,
+      "step": 3518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508133034493659e-06,
+      "loss": 0.7331,
+      "step": 3519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502848548079269e-06,
+      "loss": 0.2514,
+      "step": 3520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497564805286759e-06,
+      "loss": 0.7574,
+      "step": 3521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49228180768956e-06,
+      "loss": 0.7044,
+      "step": 3522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48699955686089e-06,
+      "loss": 0.758,
+      "step": 3523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48171805437374e-06,
+      "loss": 0.2473,
+      "step": 3524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476437301800876e-06,
+      "loss": 0.7801,
+      "step": 3525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471157300714841e-06,
+      "loss": 0.7699,
+      "step": 3526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4658780526879605e-06,
+      "loss": 0.294,
+      "step": 3527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460599559292334e-06,
+      "loss": 0.7704,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455321822099827e-06,
+      "loss": 0.7486,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450044842682087e-06,
+      "loss": 0.7452,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444768622610543e-06,
+      "loss": 0.7601,
+      "step": 3531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4394931634563834e-06,
+      "loss": 0.7538,
+      "step": 3532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43421846679058e-06,
+      "loss": 0.7557,
+      "step": 3533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42894453418387e-06,
+      "loss": 0.7421,
+      "step": 3534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423671367206775e-06,
+      "loss": 0.7789,
+      "step": 3535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4183989674295755e-06,
+      "loss": 0.7669,
+      "step": 3536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4131273364223275e-06,
+      "loss": 0.7334,
+      "step": 3537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407856475754871e-06,
+      "loss": 0.7015,
+      "step": 3538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4025863869967905e-06,
+      "loss": 0.2578,
+      "step": 3539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397317071717465e-06,
+      "loss": 0.7908,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392048531486027e-06,
+      "loss": 0.8272,
+      "step": 3541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3867807678713965e-06,
+      "loss": 0.7554,
+      "step": 3542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3815137824422365e-06,
+      "loss": 0.78,
+      "step": 3543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3762475767670015e-06,
+      "loss": 0.8006,
+      "step": 3544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370982152413903e-06,
+      "loss": 0.789,
+      "step": 3545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365717510950917e-06,
+      "loss": 0.7409,
+      "step": 3546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360453653945796e-06,
+      "loss": 0.8149,
+      "step": 3547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355190582966053e-06,
+      "loss": 0.7387,
+      "step": 3548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349928299578968e-06,
+      "loss": 0.741,
+      "step": 3549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344666805351581e-06,
+      "loss": 0.7852,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339406101850708e-06,
+      "loss": 0.6757,
+      "step": 3551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3341461906429216e-06,
+      "loss": 0.7281,
+      "step": 3552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328887073294564e-06,
+      "loss": 0.785,
+      "step": 3553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3236287513717275e-06,
+      "loss": 0.779,
+      "step": 3554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318371226440288e-06,
+      "loss": 0.749,
+      "step": 3555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313114500065872e-06,
+      "loss": 0.7972,
+      "step": 3556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307858573813865e-06,
+      "loss": 0.7621,
+      "step": 3557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302603449249419e-06,
+      "loss": 0.7157,
+      "step": 3558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297349127937454e-06,
+      "loss": 0.8137,
+      "step": 3559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292095611442642e-06,
+      "loss": 0.7892,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286842901329413e-06,
+      "loss": 0.7814,
+      "step": 3561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281590999161962e-06,
+      "loss": 0.7396,
+      "step": 3562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276339906504249e-06,
+      "loss": 0.2658,
+      "step": 3563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2710896249199805e-06,
+      "loss": 0.7819,
+      "step": 3564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265840155972628e-06,
+      "loss": 0.7843,
+      "step": 3565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2605915012254225e-06,
+      "loss": 0.7638,
+      "step": 3566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255343662241354e-06,
+      "loss": 0.7654,
+      "step": 3567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250096640583161e-06,
+      "loss": 0.7979,
+      "step": 3568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244850437813343e-06,
+      "loss": 0.7802,
+      "step": 3569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239605055494165e-06,
+      "loss": 0.7472,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234360495187628e-06,
+      "loss": 0.8673,
+      "step": 3571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2291167584555055e-06,
+      "loss": 0.7335,
+      "step": 3572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223873846859318e-06,
+      "loss": 0.7479,
+      "step": 3573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2186317619603475e-06,
+      "loss": 0.7548,
+      "step": 3574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213390505319618e-06,
+      "loss": 0.7771,
+      "step": 3575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208150078497913e-06,
+      "loss": 0.7742,
+      "step": 3576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202910483055774e-06,
+      "loss": 0.825,
+      "step": 3577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1976717205534814e-06,
+      "loss": 0.7412,
+      "step": 3578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192433792551085e-06,
+      "loss": 0.6938,
+      "step": 3579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187196700608373e-06,
+      "loss": 0.7278,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181960446284893e-06,
+      "loss": 0.7833,
+      "step": 3581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176725031139928e-06,
+      "loss": 0.7829,
+      "step": 3582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171490456732535e-06,
+      "loss": 0.8124,
+      "step": 3583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166256724621503e-06,
+      "loss": 0.8236,
+      "step": 3584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161023836365373e-06,
+      "loss": 0.7582,
+      "step": 3585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155791793522435e-06,
+      "loss": 0.8009,
+      "step": 3586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150560597650734e-06,
+      "loss": 0.8453,
+      "step": 3587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145330250308058e-06,
+      "loss": 0.7609,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140100753051938e-06,
+      "loss": 0.3183,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134872107439652e-06,
+      "loss": 0.7365,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1296443150282395e-06,
+      "loss": 0.7906,
+      "step": 3591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124417377374469e-06,
+      "loss": 0.7529,
+      "step": 3592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119191296034854e-06,
+      "loss": 0.781,
+      "step": 3593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113966072565668e-06,
+      "loss": 0.7469,
+      "step": 3594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1087417085229195e-06,
+      "loss": 0.7957,
+      "step": 3595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103518205462356e-06,
+      "loss": 0.7386,
+      "step": 3596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098295564939477e-06,
+      "loss": 0.77,
+      "step": 3597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0930737885095235e-06,
+      "loss": 0.7514,
+      "step": 3598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0878528777274814e-06,
+      "loss": 0.6911,
+      "step": 3599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082632834148069e-06,
+      "loss": 0.7177,
+      "step": 3600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077413659325753e-06,
+      "loss": 0.7714,
+      "step": 3601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072195354814748e-06,
+      "loss": 0.738,
+      "step": 3602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066977922168998e-06,
+      "loss": 0.8031,
+      "step": 3603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061761362942192e-06,
+      "loss": 0.7485,
+      "step": 3604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0565456786877586e-06,
+      "loss": 0.7851,
+      "step": 3605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05133087095887e-06,
+      "loss": 0.7845,
+      "step": 3606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0461169413084315e-06,
+      "loss": 0.7598,
+      "step": 3607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040903891289088e-06,
+      "loss": 0.8704,
+      "step": 3608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035691722453226e-06,
+      "loss": 0.2558,
+      "step": 3609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0304804363529625e-06,
+      "loss": 0.6901,
+      "step": 3610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02527003454016e-06,
+      "loss": 0.7904,
+      "step": 3611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0200605185664135e-06,
+      "loss": 0.7659,
+      "step": 3612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014851889983058e-06,
+      "loss": 0.7193,
+      "step": 3613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009644150341152e-06,
+      "loss": 0.7323,
+      "step": 3614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004437301191505e-06,
+      "loss": 0.7124,
+      "step": 3615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999231344084656e-06,
+      "loss": 0.7955,
+      "step": 3616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994026280570871e-06,
+      "loss": 0.7217,
+      "step": 3617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988822112200157e-06,
+      "loss": 0.8229,
+      "step": 3618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983618840522256e-06,
+      "loss": 0.7274,
+      "step": 3619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978416467086642e-06,
+      "loss": 0.2659,
+      "step": 3620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.973214993442512e-06,
+      "loss": 0.7134,
+      "step": 3621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968014421138809e-06,
+      "loss": 0.7197,
+      "step": 3622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962814751724201e-06,
+      "loss": 0.7714,
+      "step": 3623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957615986747085e-06,
+      "loss": 0.7327,
+      "step": 3624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952418127755589e-06,
+      "loss": 0.8412,
+      "step": 3625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94722117629758e-06,
+      "loss": 0.7778,
+      "step": 3626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942025133920647e-06,
+      "loss": 0.7834,
+      "step": 3627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936830002172104e-06,
+      "loss": 0.7536,
+      "step": 3628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931635782599e-06,
+      "loss": 0.7776,
+      "step": 3629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926442476748115e-06,
+      "loss": 0.7443,
+      "step": 3630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921250086165955e-06,
+      "loss": 0.8452,
+      "step": 3631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9160586123987475e-06,
+      "loss": 0.8721,
+      "step": 3632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91086805699245e-06,
+      "loss": 0.7574,
+      "step": 3633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905678421492754e-06,
+      "loss": 0.7999,
+      "step": 3634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900489707445069e-06,
+      "loss": 0.8412,
+      "step": 3635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895301916394528e-06,
+      "loss": 0.7462,
+      "step": 3636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890115049885995e-06,
+      "loss": 0.7596,
+      "step": 3637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884929109464061e-06,
+      "loss": 0.8338,
+      "step": 3638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879744096673033e-06,
+      "loss": 0.7515,
+      "step": 3639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874560013056948e-06,
+      "loss": 0.7648,
+      "step": 3640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869376860159563e-06,
+      "loss": 0.7695,
+      "step": 3641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8641946395243574e-06,
+      "loss": 0.8051,
+      "step": 3642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8590133526945366e-06,
+      "loss": 0.7531,
+      "step": 3643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853833001213024e-06,
+      "loss": 0.7755,
+      "step": 3644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848653586622475e-06,
+      "loss": 0.7341,
+      "step": 3645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843475110465245e-06,
+      "loss": 0.2724,
+      "step": 3646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8382975742834275e-06,
+      "loss": 0.2508,
+      "step": 3647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833120979618837e-06,
+      "loss": 0.7795,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827945328012988e-06,
+      "loss": 0.7099,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82277062100714e-06,
+      "loss": 0.7732,
+      "step": 3650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817596860142255e-06,
+      "loss": 0.2514,
+      "step": 3651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812424046959018e-06,
+      "loss": 0.7791,
+      "step": 3652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8072521829978256e-06,
+      "loss": 0.7291,
+      "step": 3653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802081269798802e-06,
+      "loss": 0.7723,
+      "step": 3654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796911308901787e-06,
+      "loss": 0.7296,
+      "step": 3655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791742301846325e-06,
+      "loss": 0.7059,
+      "step": 3656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7865742501716884e-06,
+      "loss": 0.722,
+      "step": 3657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781407155416861e-06,
+      "loss": 0.7872,
+      "step": 3658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776241019120547e-06,
+      "loss": 0.7708,
+      "step": 3659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771075842821151e-06,
+      "loss": 0.7541,
+      "step": 3660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765911628056805e-06,
+      "loss": 0.7592,
+      "step": 3661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760748376365354e-06,
+      "loss": 0.6944,
+      "step": 3662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755586089284346e-06,
+      "loss": 0.7845,
+      "step": 3663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750424768351053e-06,
+      "loss": 0.8116,
+      "step": 3664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74526441510245e-06,
+      "loss": 0.8174,
+      "step": 3665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740105031075237e-06,
+      "loss": 0.8031,
+      "step": 3666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734946617805811e-06,
+      "loss": 0.7667,
+      "step": 3667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7297891768302855e-06,
+      "loss": 0.6948,
+      "step": 3668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724632709684485e-06,
+      "loss": 0.7505,
+      "step": 3669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719477217903948e-06,
+      "loss": 0.7601,
+      "step": 3670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714322703023914e-06,
+      "loss": 0.6889,
+      "step": 3671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709169166579334e-06,
+      "loss": 0.3017,
+      "step": 3672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704016610104882e-06,
+      "loss": 0.7246,
+      "step": 3673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69886503513491e-06,
+      "loss": 0.7443,
+      "step": 3674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693714443203507e-06,
+      "loss": 0.6399,
+      "step": 3675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688564835844453e-06,
+      "loss": 0.8137,
+      "step": 3676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683416214591249e-06,
+      "loss": 0.7757,
+      "step": 3677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678268580977079e-06,
+      "loss": 0.7587,
+      "step": 3678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6731219365348535e-06,
+      "loss": 0.8055,
+      "step": 3679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667976282797184e-06,
+      "loss": 0.7149,
+      "step": 3680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66283162129638e-06,
+      "loss": 0.7688,
+      "step": 3681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657687953564462e-06,
+      "loss": 0.7354,
+      "step": 3682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652545281133156e-06,
+      "loss": 0.7996,
+      "step": 3683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647403605533886e-06,
+      "loss": 0.7613,
+      "step": 3684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642262928297778e-06,
+      "loss": 0.7315,
+      "step": 3685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637123250955668e-06,
+      "loss": 0.8287,
+      "step": 3686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631984575038094e-06,
+      "loss": 0.7352,
+      "step": 3687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626846902075286e-06,
+      "loss": 0.2542,
+      "step": 3688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621710233597183e-06,
+      "loss": 0.6787,
+      "step": 3689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616574571133425e-06,
+      "loss": 0.7424,
+      "step": 3690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6114399162133535e-06,
+      "loss": 0.7769,
+      "step": 3691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606306270366002e-06,
+      "loss": 0.7837,
+      "step": 3692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601173635120111e-06,
+      "loss": 0.6466,
+      "step": 3693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59604201200412e-06,
+      "loss": 0.814,
+      "step": 3694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590911402546163e-06,
+      "loss": 0.8041,
+      "step": 3695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585781808274074e-06,
+      "loss": 0.2655,
+      "step": 3696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580653230715384e-06,
+      "loss": 0.7676,
+      "step": 3697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575525671397327e-06,
+      "loss": 0.2337,
+      "step": 3698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5703991318468256e-06,
+      "loss": 0.6391,
+      "step": 3699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565273613590498e-06,
+      "loss": 0.7396,
+      "step": 3700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560149118154673e-06,
+      "loss": 0.8106,
+      "step": 3701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5550256470653514e-06,
+      "loss": 0.7608,
+      "step": 3702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549903201848251e-06,
+      "loss": 0.7952,
+      "step": 3703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544781784028769e-06,
+      "loss": 0.82,
+      "step": 3704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.539661395132011e-06,
+      "loss": 0.7884,
+      "step": 3705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534542036682756e-06,
+      "loss": 0.741,
+      "step": 3706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529423710205496e-06,
+      "loss": 0.6941,
+      "step": 3707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524306417224406e-06,
+      "loss": 0.7823,
+      "step": 3708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519190159263351e-06,
+      "loss": 0.801,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5140749378458954e-06,
+      "loss": 0.8106,
+      "step": 3710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508960754495292e-06,
+      "loss": 0.2792,
+      "step": 3711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503847610734482e-06,
+      "loss": 0.7746,
+      "step": 3712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498735508086094e-06,
+      "loss": 0.7815,
+      "step": 3713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4936244480724575e-06,
+      "loss": 0.7333,
+      "step": 3714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488514432215582e-06,
+      "loss": 0.6778,
+      "step": 3715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483405462037173e-06,
+      "loss": 0.7685,
+      "step": 3716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478297539058614e-06,
+      "loss": 0.7324,
+      "step": 3717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4731906648009875e-06,
+      "loss": 0.6963,
+      "step": 3718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468084840785062e-06,
+      "loss": 0.7475,
+      "step": 3719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462980068531289e-06,
+      "loss": 0.6981,
+      "step": 3720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4578763495598025e-06,
+      "loss": 0.7675,
+      "step": 3721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452773685390439e-06,
+      "loss": 0.8151,
+      "step": 3722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447672077542707e-06,
+      "loss": 0.7187,
+      "step": 3723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442571527535802e-06,
+      "loss": 0.8051,
+      "step": 3724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437472036888608e-06,
+      "loss": 0.822,
+      "step": 3725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432373607119698e-06,
+      "loss": 0.754,
+      "step": 3726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4272762397473145e-06,
+      "loss": 0.7706,
+      "step": 3727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422179936289394e-06,
+      "loss": 0.7332,
+      "step": 3728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.417084698263563e-06,
+      "loss": 0.7698,
+      "step": 3729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411990527187118e-06,
+      "loss": 0.795,
+      "step": 3730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406897424577037e-06,
+      "loss": 0.7961,
+      "step": 3731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40180539194999e-06,
+      "loss": 0.7602,
+      "step": 3732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3967144308223285e-06,
+      "loss": 0.8233,
+      "step": 3733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391624542710069e-06,
+      "loss": 0.7356,
+      "step": 3734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386535729128926e-06,
+      "loss": 0.6952,
+      "step": 3735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381447991594285e-06,
+      "loss": 0.7259,
+      "step": 3736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376361331621219e-06,
+      "loss": 0.7793,
+      "step": 3737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371275750724468e-06,
+      "loss": 0.6889,
+      "step": 3738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366191250418461e-06,
+      "loss": 0.756,
+      "step": 3739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3611078322173e-06,
+      "loss": 0.7156,
+      "step": 3740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356025497634763e-06,
+      "loss": 0.76,
+      "step": 3741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350944248184314e-06,
+      "loss": 0.7877,
+      "step": 3742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345864085379087e-06,
+      "loss": 0.755,
+      "step": 3743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340785010731893e-06,
+      "loss": 0.7843,
+      "step": 3744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335707025755218e-06,
+      "loss": 0.7758,
+      "step": 3745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330630131961228e-06,
+      "loss": 0.7437,
+      "step": 3746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325554330861763e-06,
+      "loss": 0.7751,
+      "step": 3747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320479623968332e-06,
+      "loss": 0.8152,
+      "step": 3748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3154060127921206e-06,
+      "loss": 0.7208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310333498843994e-06,
+      "loss": 0.7281,
+      "step": 3750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305262083634488e-06,
+      "loss": 0.7553,
+      "step": 3751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300191768673805e-06,
+      "loss": 0.7505,
+      "step": 3752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295122555471821e-06,
+      "loss": 0.8009,
+      "step": 3753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290054445538095e-06,
+      "loss": 0.8204,
+      "step": 3754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2849874403818485e-06,
+      "loss": 0.8105,
+      "step": 3755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279921541511969e-06,
+      "loss": 0.8327,
+      "step": 3756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2748567504370265e-06,
+      "loss": 0.7359,
+      "step": 3757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269793068665254e-06,
+      "loss": 0.2546,
+      "step": 3758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264730497704554e-06,
+      "loss": 0.7305,
+      "step": 3759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2596690390625e-06,
+      "loss": 0.7555,
+      "step": 3760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254608694246334e-06,
+      "loss": 0.73,
+      "step": 3761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24954946476297e-06,
+      "loss": 0.7331,
+      "step": 3762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24449135211898e-06,
+      "loss": 0.8218,
+      "step": 3763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23943435782061e-06,
+      "loss": 0.287,
+      "step": 3764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2343784833737795e-06,
+      "loss": 0.7635,
+      "step": 3765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22932373028406e-06,
+      "loss": 0.7935,
+      "step": 3766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224270100056701e-06,
+      "loss": 0.7351,
+      "step": 3767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21921759419661e-06,
+      "loss": 0.769,
+      "step": 3768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214166214208366e-06,
+      "loss": 0.2276,
+      "step": 3769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209115961596208e-06,
+      "loss": 0.7156,
+      "step": 3770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20406683786404e-06,
+      "loss": 0.8084,
+      "step": 3771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199018844515435e-06,
+      "loss": 0.7861,
+      "step": 3772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193971983053618e-06,
+      "loss": 0.7991,
+      "step": 3773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18892625498149e-06,
+      "loss": 0.721,
+      "step": 3774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183881661801605e-06,
+      "loss": 0.7324,
+      "step": 3775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178838205016189e-06,
+      "loss": 0.7813,
+      "step": 3776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173795886127113e-06,
+      "loss": 0.7704,
+      "step": 3777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168754706635927e-06,
+      "loss": 0.7554,
+      "step": 3778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163714668043834e-06,
+      "loss": 0.87,
+      "step": 3779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158675771851692e-06,
+      "loss": 0.7959,
+      "step": 3780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153638019560027e-06,
+      "loss": 0.7092,
+      "step": 3781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148601412669022e-06,
+      "loss": 0.701,
+      "step": 3782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143565952678521e-06,
+      "loss": 0.7621,
+      "step": 3783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1385316410880165e-06,
+      "loss": 0.279,
+      "step": 3784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133498479396672e-06,
+      "loss": 0.6974,
+      "step": 3785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128466469103305e-06,
+      "loss": 0.7911,
+      "step": 3786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123435611706383e-06,
+      "loss": 0.7735,
+      "step": 3787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118405908704034e-06,
+      "loss": 0.7316,
+      "step": 3788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113377361594048e-06,
+      "loss": 0.7666,
+      "step": 3789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108349971873868e-06,
+      "loss": 0.8213,
+      "step": 3790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103323741040586e-06,
+      "loss": 0.7652,
+      "step": 3791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0982986705909505e-06,
+      "loss": 0.7387,
+      "step": 3792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093274762021375e-06,
+      "loss": 0.7552,
+      "step": 3793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08825201682792e-06,
+      "loss": 0.7666,
+      "step": 3794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083230436506291e-06,
+      "loss": 0.7499,
+      "step": 3795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078210022551858e-06,
+      "loss": 0.8243,
+      "step": 3796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073190776459645e-06,
+      "loss": 0.7079,
+      "step": 3797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06817269972432e-06,
+      "loss": 0.8287,
+      "step": 3798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063155793840203e-06,
+      "loss": 0.7405,
+      "step": 3799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0581400603012715e-06,
+      "loss": 0.7033,
+      "step": 3800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053125500601156e-06,
+      "loss": 0.7533,
+      "step": 3801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048112116233125e-06,
+      "loss": 0.3022,
+      "step": 3802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043099908690105e-06,
+      "loss": 0.7242,
+      "step": 3803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0380888794646796e-06,
+      "loss": 0.7495,
+      "step": 3804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033079030049063e-06,
+      "loss": 0.7054,
+      "step": 3805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028070361935134e-06,
+      "loss": 0.2446,
+      "step": 3806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023062876614412e-06,
+      "loss": 0.8047,
+      "step": 3807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018056575578075e-06,
+      "loss": 0.7759,
+      "step": 3808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013051460316926e-06,
+      "loss": 0.7825,
+      "step": 3809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0080475323214346e-06,
+      "loss": 0.8031,
+      "step": 3810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003044793081716e-06,
+      "loss": 0.8199,
+      "step": 3811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998043244087514e-06,
+      "loss": 0.773,
+      "step": 3812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993042886828243e-06,
+      "loss": 0.7781,
+      "step": 3813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988043722792942e-06,
+      "loss": 0.8205,
+      "step": 3814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.7338,
+      "step": 3815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978048980348669e-06,
+      "loss": 0.7134,
+      "step": 3816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97305340491601e-06,
+      "loss": 0.7214,
+      "step": 3817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9680590286599536e-06,
+      "loss": 0.7381,
+      "step": 3818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963065853067762e-06,
+      "loss": 0.2665,
+      "step": 3819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958073879626344e-06,
+      "loss": 0.2761,
+      "step": 3820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953083109822253e-06,
+      "loss": 0.7866,
+      "step": 3821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94809354514168e-06,
+      "loss": 0.7403,
+      "step": 3822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943105187070454e-06,
+      "loss": 0.7279,
+      "step": 3823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938118037094049e-06,
+      "loss": 0.7057,
+      "step": 3824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9331320966975845e-06,
+      "loss": 0.8319,
+      "step": 3825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928147367365808e-06,
+      "loss": 0.7474,
+      "step": 3826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923163850583114e-06,
+      "loss": 0.8442,
+      "step": 3827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9181815478335325e-06,
+      "loss": 0.783,
+      "step": 3828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9132004606007394e-06,
+      "loss": 0.7302,
+      "step": 3829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908220590368039e-06,
+      "loss": 0.7133,
+      "step": 3830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903241938618374e-06,
+      "loss": 0.7712,
+      "step": 3831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898264506834335e-06,
+      "loss": 0.7319,
+      "step": 3832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893288296498138e-06,
+      "loss": 0.6677,
+      "step": 3833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888313309091638e-06,
+      "loss": 0.7694,
+      "step": 3834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883339546096325e-06,
+      "loss": 0.7972,
+      "step": 3835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878367008993334e-06,
+      "loss": 0.7431,
+      "step": 3836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873395699263416e-06,
+      "loss": 0.8047,
+      "step": 3837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868425618386977e-06,
+      "loss": 0.7647,
+      "step": 3838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863456767844042e-06,
+      "loss": 0.7823,
+      "step": 3839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858489149114281e-06,
+      "loss": 0.7054,
+      "step": 3840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853522763676987e-06,
+      "loss": 0.7771,
+      "step": 3841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848557613011092e-06,
+      "loss": 0.7684,
+      "step": 3842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843593698595159e-06,
+      "loss": 0.7928,
+      "step": 3843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838631021907377e-06,
+      "loss": 0.8039,
+      "step": 3844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833669584425581e-06,
+      "loss": 0.7178,
+      "step": 3845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828709387627219e-06,
+      "loss": 0.7785,
+      "step": 3846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823750432989387e-06,
+      "loss": 0.7549,
+      "step": 3847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818792721988798e-06,
+      "loss": 0.668,
+      "step": 3848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813836256101795e-06,
+      "loss": 0.8339,
+      "step": 3849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8088810368043635e-06,
+      "loss": 0.7527,
+      "step": 3850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803927065572098e-06,
+      "loss": 0.7636,
+      "step": 3851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798974343880242e-06,
+      "loss": 0.6697,
+      "step": 3852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79402287320365e-06,
+      "loss": 0.7271,
+      "step": 3853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789072655016817e-06,
+      "loss": 0.7399,
+      "step": 3854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784123690793849e-06,
+      "loss": 0.7522,
+      "step": 3855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7791759820085e-06,
+      "loss": 0.7806,
+      "step": 3856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774229530134133e-06,
+      "loss": 0.7975,
+      "step": 3857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769284336643739e-06,
+      "loss": 0.7575,
+      "step": 3858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76434040300994e-06,
+      "loss": 0.6933,
+      "step": 3859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759397730704985e-06,
+      "loss": 0.2773,
+      "step": 3860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754456321200741e-06,
+      "loss": 0.7769,
+      "step": 3861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7495161759686945e-06,
+      "loss": 0.7836,
+      "step": 3862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744577296479965e-06,
+      "loss": 0.7175,
+      "step": 3863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739639684205298e-06,
+      "loss": 0.7068,
+      "step": 3864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7347033406150494e-06,
+      "loss": 0.7391,
+      "step": 3865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729768267179201e-06,
+      "loss": 0.7453,
+      "step": 3866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724834465367363e-06,
+      "loss": 0.7981,
+      "step": 3867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719901936648765e-06,
+      "loss": 0.7081,
+      "step": 3868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714970682492251e-06,
+      "loss": 0.2697,
+      "step": 3869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710040704366289e-06,
+      "loss": 0.7662,
+      "step": 3870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705112003738967e-06,
+      "loss": 0.797,
+      "step": 3871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700184582078e-06,
+      "loss": 0.718,
+      "step": 3872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695258440850706e-06,
+      "loss": 0.7286,
+      "step": 3873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690333581524042e-06,
+      "loss": 0.7584,
+      "step": 3874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685410005564563e-06,
+      "loss": 0.7027,
+      "step": 3875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6804877144384515e-06,
+      "loss": 0.8011,
+      "step": 3876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675566709611509e-06,
+      "loss": 0.7389,
+      "step": 3877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670646992549158e-06,
+      "loss": 0.8865,
+      "step": 3878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665728564716425e-06,
+      "loss": 0.8072,
+      "step": 3879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660811427577957e-06,
+      "loss": 0.2716,
+      "step": 3880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65589558259802e-06,
+      "loss": 0.7405,
+      "step": 3881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650981031240501e-06,
+      "loss": 0.6986,
+      "step": 3882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646067774968888e-06,
+      "loss": 0.7546,
+      "step": 3883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64115581524629e-06,
+      "loss": 0.7367,
+      "step": 3884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636245153535429e-06,
+      "loss": 0.7541,
+      "step": 3885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631335791298649e-06,
+      "loss": 0.7926,
+      "step": 3886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626427729997894e-06,
+      "loss": 0.7191,
+      "step": 3887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621520971094723e-06,
+      "loss": 0.7322,
+      "step": 3888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616615516050317e-06,
+      "loss": 0.7695,
+      "step": 3889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611711366325454e-06,
+      "loss": 0.7395,
+      "step": 3890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6068085233805405e-06,
+      "loss": 0.7391,
+      "step": 3891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601906988675575e-06,
+      "loss": 0.7707,
+      "step": 3892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597006763670185e-06,
+      "loss": 0.7209,
+      "step": 3893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592107849823591e-06,
+      "loss": 0.7931,
+      "step": 3894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587210248594637e-06,
+      "loss": 0.6792,
+      "step": 3895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58231396144177e-06,
+      "loss": 0.7474,
+      "step": 3896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577418989823038e-06,
+      "loss": 0.718,
+      "step": 3897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572525335196113e-06,
+      "loss": 0.7175,
+      "step": 3898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567632999018266e-06,
+      "loss": 0.781,
+      "step": 3899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562741982746377e-06,
+      "loss": 0.7774,
+      "step": 3900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557852287836925e-06,
+      "loss": 0.8042,
+      "step": 3901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5529639157460095e-06,
+      "loss": 0.7524,
+      "step": 3902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548076867929331e-06,
+      "loss": 0.7497,
+      "step": 3903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5431911458421906e-06,
+      "loss": 0.7495,
+      "step": 3904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538306750939493e-06,
+      "loss": 0.7058,
+      "step": 3905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53342368467576e-06,
+      "loss": 0.7793,
+      "step": 3906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528541948505111e-06,
+      "loss": 0.7125,
+      "step": 3907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523661543881266e-06,
+      "loss": 0.707,
+      "step": 3908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518782472257547e-06,
+      "loss": 0.2696,
+      "step": 3909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5139047350868876e-06,
+      "loss": 0.726,
+      "step": 3910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509028333821824e-06,
+      "loss": 0.755,
+      "step": 3911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504153269914485e-06,
+      "loss": 0.7145,
+      "step": 3912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499279544816604e-06,
+      "loss": 0.7501,
+      "step": 3913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494407159979525e-06,
+      "loss": 0.7941,
+      "step": 3914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4895361168541775e-06,
+      "loss": 0.7992,
+      "step": 3915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484666416891109e-06,
+      "loss": 0.7378,
+      "step": 3916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47979806154045e-06,
+      "loss": 0.8029,
+      "step": 3917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474931052251946e-06,
+      "loss": 0.7952,
+      "step": 3918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4700653904749275e-06,
+      "loss": 0.8253,
+      "step": 3919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465201077658333e-06,
+      "loss": 0.77,
+      "step": 3920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460338115250705e-06,
+      "loss": 0.7937,
+      "step": 3921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455476504700161e-06,
+      "loss": 0.7513,
+      "step": 3922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450616247454437e-06,
+      "loss": 0.738,
+      "step": 3923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445757344960859e-06,
+      "loss": 0.7227,
+      "step": 3924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440899798666359e-06,
+      "loss": 0.7133,
+      "step": 3925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436043610017442e-06,
+      "loss": 0.8085,
+      "step": 3926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431188780460228e-06,
+      "loss": 0.2647,
+      "step": 3927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426335311440434e-06,
+      "loss": 0.7725,
+      "step": 3928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421483204403355e-06,
+      "loss": 0.753,
+      "step": 3929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416632460793901e-06,
+      "loss": 0.78,
+      "step": 3930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4117830820565545e-06,
+      "loss": 0.7357,
+      "step": 3931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406935069635411e-06,
+      "loss": 0.8742,
+      "step": 3932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402088424974144e-06,
+      "loss": 0.8161,
+      "step": 3933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397243149516034e-06,
+      "loss": 0.7924,
+      "step": 3934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392399244703943e-06,
+      "loss": 0.729,
+      "step": 3935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387556711980322e-06,
+      "loss": 0.736,
+      "step": 3936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3827155527872255e-06,
+      "loss": 0.7571,
+      "step": 3937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3778757685662966e-06,
+      "loss": 0.7794,
+      "step": 3938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3730373607587595e-06,
+      "loss": 0.7348,
+      "step": 3939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368200330805431e-06,
+      "loss": 0.7286,
+      "step": 3940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3633646801467255e-06,
+      "loss": 0.7263,
+      "step": 3941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3585304102226464e-06,
+      "loss": 0.8327,
+      "step": 3942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353697522472776e-06,
+      "loss": 0.7126,
+      "step": 3943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348866018336288e-06,
+      "loss": 0.7907,
+      "step": 3944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.344035899251951e-06,
+      "loss": 0.7419,
+      "step": 3945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339207166658118e-06,
+      "loss": 0.7621,
+      "step": 3946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334379821992726e-06,
+      "loss": 0.7689,
+      "step": 3947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329553866693296e-06,
+      "loss": 0.8009,
+      "step": 3948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324729302196945e-06,
+      "loss": 0.6904,
+      "step": 3949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3199061299403745e-06,
+      "loss": 0.7089,
+      "step": 3950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315084351359863e-06,
+      "loss": 0.722,
+      "step": 3951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310263967891273e-06,
+      "loss": 0.7261,
+      "step": 3952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305444980970067e-06,
+      "loss": 0.2598,
+      "step": 3953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300627392031273e-06,
+      "loss": 0.7863,
+      "step": 3954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295811202509521e-06,
+      "loss": 0.7248,
+      "step": 3955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290996413839005e-06,
+      "loss": 0.7879,
+      "step": 3956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28618302745352e-06,
+      "loss": 0.782,
+      "step": 3957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281371044786427e-06,
+      "loss": 0.8113,
+      "step": 3958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276560467270685e-06,
+      "loss": 0.7718,
+      "step": 3959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271751296338823e-06,
+      "loss": 0.7163,
+      "step": 3960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26694353342295e-06,
+      "loss": 0.8282,
+      "step": 3961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262137179954765e-06,
+      "loss": 0.7075,
+      "step": 3962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257332237365546e-06,
+      "loss": 0.7682,
+      "step": 3963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252528707086144e-06,
+      "loss": 0.7398,
+      "step": 3964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247726590546988e-06,
+      "loss": 0.8021,
+      "step": 3965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242925889178097e-06,
+      "loss": 0.7517,
+      "step": 3966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238126604409064e-06,
+      "loss": 0.7753,
+      "step": 3967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233328737669057e-06,
+      "loss": 0.7346,
+      "step": 3968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228532290386816e-06,
+      "loss": 0.728,
+      "step": 3969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223737263990674e-06,
+      "loss": 0.6656,
+      "step": 3970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218943659908535e-06,
+      "loss": 0.8046,
+      "step": 3971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214151479567873e-06,
+      "loss": 0.7932,
+      "step": 3972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209360724395735e-06,
+      "loss": 0.7381,
+      "step": 3973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204571395818764e-06,
+      "loss": 0.7105,
+      "step": 3974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199783495263153e-06,
+      "loss": 0.7212,
+      "step": 3975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194997024154689e-06,
+      "loss": 0.769,
+      "step": 3976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190211983918725e-06,
+      "loss": 0.6908,
+      "step": 3977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185428375980188e-06,
+      "loss": 0.6671,
+      "step": 3978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1806462017635775e-06,
+      "loss": 0.2582,
+      "step": 3979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175865462692967e-06,
+      "loss": 0.7586,
+      "step": 3980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171086160192014e-06,
+      "loss": 0.7226,
+      "step": 3981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166308295683922e-06,
+      "loss": 0.7934,
+      "step": 3982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161531870591489e-06,
+      "loss": 0.8414,
+      "step": 3983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15675688633708e-06,
+      "loss": 0.7635,
+      "step": 3984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151983344342628e-06,
+      "loss": 0.2334,
+      "step": 3985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147211246029635e-06,
+      "loss": 0.2534,
+      "step": 3986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1424405928191715e-06,
+      "loss": 0.7741,
+      "step": 3987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137671386131884e-06,
+      "loss": 0.7771,
+      "step": 3988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13290362738799e-06,
+      "loss": 0.7927,
+      "step": 3989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128137318007267e-06,
+      "loss": 0.7525,
+      "step": 3990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123372459409059e-06,
+      "loss": 0.813,
+      "step": 3991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118609053012296e-06,
+      "loss": 0.7169,
+      "step": 3992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113847100235452e-06,
+      "loss": 0.7513,
+      "step": 3993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109086602496589e-06,
+      "loss": 0.698,
+      "step": 3994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104327561213319e-06,
+      "loss": 0.7199,
+      "step": 3995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099569977802834e-06,
+      "loss": 0.7699,
+      "step": 3996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094813853681879e-06,
+      "loss": 0.7561,
+      "step": 3997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090059190266779e-06,
+      "loss": 0.7518,
+      "step": 3998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0853059889734115e-06,
+      "loss": 0.2467,
+      "step": 3999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080554251217218e-06,
+      "loss": 0.7442,
+      "step": 4000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075803978413216e-06,
+      "loss": 0.76,
+      "step": 4001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071055171975981e-06,
+      "loss": 0.8161,
+      "step": 4002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066307833319648e-06,
+      "loss": 0.7219,
+      "step": 4003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061561963857914e-06,
+      "loss": 0.7658,
+      "step": 4004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056817565004046e-06,
+      "loss": 0.8146,
+      "step": 4005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0520746381708745e-06,
+      "loss": 0.7919,
+      "step": 4006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.04733318477078e-06,
+      "loss": 0.6977,
+      "step": 4007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042593206215709e-06,
+      "loss": 0.7106,
+      "step": 4008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037854703917173e-06,
+      "loss": 0.7279,
+      "step": 4009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033117679286246e-06,
+      "loss": 0.7157,
+      "step": 4010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028382133733554e-06,
+      "loss": 0.7507,
+      "step": 4011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0236480686692805e-06,
+      "loss": 0.6593,
+      "step": 4012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018915485503182e-06,
+      "loss": 0.7492,
+      "step": 4013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014184385644558e-06,
+      "loss": 0.7568,
+      "step": 4014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009454770502282e-06,
+      "loss": 0.7449,
+      "step": 4015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004726641484767e-06,
+      "loss": 0.6897,
+      "step": 4016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.7388,
+      "step": 4017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9952748474555176e-06,
+      "loss": 0.736,
+      "step": 4018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990551185258414e-06,
+      "loss": 0.7282,
+      "step": 4019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985829014815339e-06,
+      "loss": 0.7345,
+      "step": 4020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981108337532495e-06,
+      "loss": 0.7766,
+      "step": 4021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976389154815644e-06,
+      "loss": 0.6694,
+      "step": 4022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971671468070105e-06,
+      "loss": 0.8051,
+      "step": 4023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966955278700756e-06,
+      "loss": 0.7615,
+      "step": 4024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9622405881120065e-06,
+      "loss": 0.7309,
+      "step": 4025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957527397707843e-06,
+      "loss": 0.7479,
+      "step": 4026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952815708891796e-06,
+      "loss": 0.7836,
+      "step": 4027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9481055230669596e-06,
+      "loss": 0.8114,
+      "step": 4028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943396841635955e-06,
+      "loss": 0.2538,
+      "step": 4029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938689666000978e-06,
+      "loss": 0.7104,
+      "step": 4030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933983997563772e-06,
+      "loss": 0.6992,
+      "step": 4031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929279837725626e-06,
+      "loss": 0.7305,
+      "step": 4032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9245771878873845e-06,
+      "loss": 0.7504,
+      "step": 4033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919876049449438e-06,
+      "loss": 0.7764,
+      "step": 4034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9151764238117325e-06,
+      "loss": 0.7877,
+      "step": 4035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910478312373757e-06,
+      "loss": 0.7186,
+      "step": 4036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905781716534556e-06,
+      "loss": 0.7982,
+      "step": 4037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9010866376927206e-06,
+      "loss": 0.2527,
+      "step": 4038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896393077246381e-06,
+      "loss": 0.2318,
+      "step": 4039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891701036593231e-06,
+      "loss": 0.8094,
+      "step": 4040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887010517130505e-06,
+      "loss": 0.7989,
+      "step": 4041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8823215202549814e-06,
+      "loss": 0.7991,
+      "step": 4042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877634047362985e-06,
+      "loss": 0.6825,
+      "step": 4043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872948099850389e-06,
+      "loss": 0.2198,
+      "step": 4044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868263679112621e-06,
+      "loss": 0.8143,
+      "step": 4045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8635807865446395e-06,
+      "loss": 0.7449,
+      "step": 4046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8588994235409494e-06,
+      "loss": 0.7694,
+      "step": 4047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8542195914956105e-06,
+      "loss": 0.7789,
+      "step": 4048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849541291802222e-06,
+      "loss": 0.7743,
+      "step": 4049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844864525853925e-06,
+      "loss": 0.7876,
+      "step": 4050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840189295043399e-06,
+      "loss": 0.7218,
+      "step": 4051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83551560076288e-06,
+      "loss": 0.7346,
+      "step": 4052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8308434444041305e-06,
+      "loss": 0.7145,
+      "step": 4053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8261728273584705e-06,
+      "loss": 0.7602,
+      "step": 4054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821503751016746e-06,
+      "loss": 0.79,
+      "step": 4055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81683621676936e-06,
+      "loss": 0.7724,
+      "step": 4056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812170226006242e-06,
+      "loss": 0.6053,
+      "step": 4057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807505780116873e-06,
+      "loss": 0.7586,
+      "step": 4058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802842880490269e-06,
+      "loss": 0.6761,
+      "step": 4059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79818152851498e-06,
+      "loss": 0.7598,
+      "step": 4060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793521725579105e-06,
+      "loss": 0.6596,
+      "step": 4061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7888634730702825e-06,
+      "loss": 0.7834,
+      "step": 4062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784206772375678e-06,
+      "loss": 0.8002,
+      "step": 4063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779551624882001e-06,
+      "loss": 0.7913,
+      "step": 4064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774898031975499e-06,
+      "loss": 0.7409,
+      "step": 4065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7702459950419635e-06,
+      "loss": 0.778,
+      "step": 4066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76559551546671e-06,
+      "loss": 0.7837,
+      "step": 4067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7609465946345935e-06,
+      "loss": 0.8121,
+      "step": 4068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756299233930009e-06,
+      "loss": 0.7119,
+      "step": 4069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751653434736888e-06,
+      "loss": 0.7686,
+      "step": 4070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747009198438694e-06,
+      "loss": 0.6769,
+      "step": 4071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742366526418417e-06,
+      "loss": 0.7016,
+      "step": 4072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737725420058597e-06,
+      "loss": 0.6917,
+      "step": 4073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733085880741301e-06,
+      "loss": 0.7312,
+      "step": 4074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728447909848126e-06,
+      "loss": 0.7809,
+      "step": 4075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7238115087602e-06,
+      "loss": 0.7834,
+      "step": 4076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719176678858196e-06,
+      "loss": 0.7161,
+      "step": 4077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714543421522303e-06,
+      "loss": 0.687,
+      "step": 4078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709911738132252e-06,
+      "loss": 0.8009,
+      "step": 4079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70528163006731e-06,
+      "loss": 0.8051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700653098706262e-06,
+      "loss": 0.7074,
+      "step": 4081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696026145427426e-06,
+      "loss": 0.6598,
+      "step": 4082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691400771608656e-06,
+      "loss": 0.6955,
+      "step": 4083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686776978627343e-06,
+      "loss": 0.7365,
+      "step": 4084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682154767860381e-06,
+      "loss": 0.7758,
+      "step": 4085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677534140684216e-06,
+      "loss": 0.2283,
+      "step": 4086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672915098474816e-06,
+      "loss": 0.7901,
+      "step": 4087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668297642607682e-06,
+      "loss": 0.78,
+      "step": 4088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663681774457831e-06,
+      "loss": 0.7459,
+      "step": 4089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659067495399812e-06,
+      "loss": 0.6999,
+      "step": 4090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6544548068077076e-06,
+      "loss": 0.8416,
+      "step": 4091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649843710055115e-06,
+      "loss": 0.7757,
+      "step": 4092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645234206515171e-06,
+      "loss": 0.7014,
+      "step": 4093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640626297560523e-06,
+      "loss": 0.835,
+      "step": 4094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636019984563359e-06,
+      "loss": 0.7496,
+      "step": 4095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631415268895376e-06,
+      "loss": 0.6863,
+      "step": 4096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626812151927812e-06,
+      "loss": 0.2568,
+      "step": 4097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622210635031416e-06,
+      "loss": 0.755,
+      "step": 4098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617610719576461e-06,
+      "loss": 0.8122,
+      "step": 4099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6130124069327485e-06,
+      "loss": 0.7217,
+      "step": 4100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608415698469608e-06,
+      "loss": 0.796,
+      "step": 4101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60382059555588e-06,
+      "loss": 0.7229,
+      "step": 4102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599227099559926e-06,
+      "loss": 0.7434,
+      "step": 4103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594635211849639e-06,
+      "loss": 0.735,
+      "step": 4104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5900449337924336e-06,
+      "loss": 0.6768,
+      "step": 4105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585456266755232e-06,
+      "loss": 0.7556,
+      "step": 4106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580869212104485e-06,
+      "loss": 0.691,
+      "step": 4107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576283771206164e-06,
+      "loss": 0.7564,
+      "step": 4108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571699945425763e-06,
+      "loss": 0.7546,
+      "step": 4109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5671177361282855e-06,
+      "loss": 0.7514,
+      "step": 4110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562537144678257e-06,
+      "loss": 0.7491,
+      "step": 4111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557958172439726e-06,
+      "loss": 0.6889,
+      "step": 4112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553380820776259e-06,
+      "loss": 0.7545,
+      "step": 4113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548805091050933e-06,
+      "loss": 0.818,
+      "step": 4114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544230984626342e-06,
+      "loss": 0.7082,
+      "step": 4115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5396585028646076e-06,
+      "loss": 0.6395,
+      "step": 4116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5350876471273546e-06,
+      "loss": 0.7597,
+      "step": 4117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.7563,
+      "step": 4118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5259508191704e-06,
+      "loss": 0.7858,
+      "step": 4119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52138484967154e-06,
+      "loss": 0.7518,
+      "step": 4120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516820511638834e-06,
+      "loss": 0.7587,
+      "step": 4121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512257806431497e-06,
+      "loss": 0.7815,
+      "step": 4122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507696735408243e-06,
+      "loss": 0.7906,
+      "step": 4123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503137299927303e-06,
+      "loss": 0.2423,
+      "step": 4124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498579501346423e-06,
+      "loss": 0.7417,
+      "step": 4125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494023341022864e-06,
+      "loss": 0.7647,
+      "step": 4126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4894688203134004e-06,
+      "loss": 0.7199,
+      "step": 4127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484915940574303e-06,
+      "loss": 0.837,
+      "step": 4128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48036470316137e-06,
+      "loss": 0.7109,
+      "step": 4129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475815109429908e-06,
+      "loss": 0.809,
+      "step": 4130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471267160734731e-06,
+      "loss": 0.7617,
+      "step": 4131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466720858430158e-06,
+      "loss": 0.7599,
+      "step": 4132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462176203870029e-06,
+      "loss": 0.7502,
+      "step": 4133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4576331984076895e-06,
+      "loss": 0.8058,
+      "step": 4134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453091843395987e-06,
+      "loss": 0.7209,
+      "step": 4135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448552140187288e-06,
+      "loss": 0.7634,
+      "step": 4136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44401409013346e-06,
+      "loss": 0.7839,
+      "step": 4137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439477694585875e-06,
+      "loss": 0.7954,
+      "step": 4138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4349429548954225e-06,
+      "loss": 0.7365,
+      "step": 4139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430409872412497e-06,
+      "loss": 0.7374,
+      "step": 4140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425878448486991e-06,
+      "loss": 0.7125,
+      "step": 4141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421348684468307e-06,
+      "loss": 0.7593,
+      "step": 4142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416820581705358e-06,
+      "loss": 0.814,
+      "step": 4143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4122941415465615e-06,
+      "loss": 0.7183,
+      "step": 4144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407769365339835e-06,
+      "loss": 0.7592,
+      "step": 4145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403246254432597e-06,
+      "loss": 0.7717,
+      "step": 4146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398724810171782e-06,
+      "loss": 0.7152,
+      "step": 4147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394205033903824e-06,
+      "loss": 0.8196,
+      "step": 4148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389686926974655e-06,
+      "loss": 0.7421,
+      "step": 4149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385170490729712e-06,
+      "loss": 0.6993,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380655726513936e-06,
+      "loss": 0.7553,
+      "step": 4151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376142635671776e-06,
+      "loss": 0.7,
+      "step": 4152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371631219547171e-06,
+      "loss": 0.2491,
+      "step": 4153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367121479483565e-06,
+      "loss": 0.7254,
+      "step": 4154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36261341682391e-06,
+      "loss": 0.7185,
+      "step": 4155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358107032910648e-06,
+      "loss": 0.7266,
+      "step": 4156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353602329085732e-06,
+      "loss": 0.711,
+      "step": 4157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349099306690603e-06,
+      "loss": 0.7465,
+      "step": 4158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344597967066213e-06,
+      "loss": 0.7204,
+      "step": 4159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340098311553001e-06,
+      "loss": 0.7771,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335600341490919e-06,
+      "loss": 0.7306,
+      "step": 4161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331104058219403e-06,
+      "loss": 0.6874,
+      "step": 4162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32660946307739e-06,
+      "loss": 0.6962,
+      "step": 4163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322116557403319e-06,
+      "loss": 0.7654,
+      "step": 4164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31762534253513e-06,
+      "loss": 0.7883,
+      "step": 4165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313135819810248e-06,
+      "loss": 0.7071,
+      "step": 4166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308647990565594e-06,
+      "loss": 0.8413,
+      "step": 4167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304161856137596e-06,
+      "loss": 0.7533,
+      "step": 4168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299677417862174e-06,
+      "loss": 0.8005,
+      "step": 4169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295194677074736e-06,
+      "loss": 0.7773,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290713635110186e-06,
+      "loss": 0.7865,
+      "step": 4171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286234293302929e-06,
+      "loss": 0.7218,
+      "step": 4172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28175665298686e-06,
+      "loss": 0.7192,
+      "step": 4173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277280715495367e-06,
+      "loss": 0.7802,
+      "step": 4174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272806482161325e-06,
+      "loss": 0.7842,
+      "step": 4175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268333954317114e-06,
+      "loss": 0.7291,
+      "step": 4176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263863133294595e-06,
+      "loss": 0.7266,
+      "step": 4177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2593940204251285e-06,
+      "loss": 0.2787,
+      "step": 4178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254926617039557e-06,
+      "loss": 0.7879,
+      "step": 4179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250460924468229e-06,
+      "loss": 0.8067,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2459969440409655e-06,
+      "loss": 0.7597,
+      "step": 4181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241534677087091e-06,
+      "loss": 0.6636,
+      "step": 4182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237074124935418e-06,
+      "loss": 0.7366,
+      "step": 4183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232615288914243e-06,
+      "loss": 0.7368,
+      "step": 4184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2281581703513485e-06,
+      "loss": 0.7695,
+      "step": 4185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223702770574018e-06,
+      "loss": 0.2615,
+      "step": 4186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2192490909090215e-06,
+      "loss": 0.7022,
+      "step": 4187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214797132682597e-06,
+      "loss": 0.7716,
+      "step": 4188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210346897220492e-06,
+      "loss": 0.72,
+      "step": 4189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205898385847933e-06,
+      "loss": 0.7481,
+      "step": 4190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2014515998896385e-06,
+      "loss": 0.7384,
+      "step": 4191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197006540669804e-06,
+      "loss": 0.7701,
+      "step": 4192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192563209512113e-06,
+      "loss": 0.7502,
+      "step": 4193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18812160773974e-06,
+      "loss": 0.7334,
+      "step": 4194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183681736675337e-06,
+      "loss": 0.7156,
+      "step": 4195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179243597641051e-06,
+      "loss": 0.7488,
+      "step": 4196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174807191958498e-06,
+      "loss": 0.7529,
+      "step": 4197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170372520948796e-06,
+      "loss": 0.7557,
+      "step": 4198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165939585932528e-06,
+      "loss": 0.2828,
+      "step": 4199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1615083882297766e-06,
+      "loss": 0.7468,
+      "step": 4200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157078929160097e-06,
+      "loss": 0.7223,
+      "step": 4201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152651210042527e-06,
+      "loss": 0.7465,
+      "step": 4202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148225232195589e-06,
+      "loss": 0.8116,
+      "step": 4203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143800996937292e-06,
+      "loss": 0.793,
+      "step": 4204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139378505585114e-06,
+      "loss": 0.7013,
+      "step": 4205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13495775945602e-06,
+      "loss": 0.7872,
+      "step": 4206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130538759866457e-06,
+      "loss": 0.736,
+      "step": 4207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126121508132355e-06,
+      "loss": 0.7887,
+      "step": 4208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121706005569114e-06,
+      "loss": 0.7502,
+      "step": 4209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117292253491615e-06,
+      "loss": 0.6799,
+      "step": 4210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112880253214225e-06,
+      "loss": 0.8025,
+      "step": 4211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108470006050787e-06,
+      "loss": 0.7613,
+      "step": 4212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1040615133146175e-06,
+      "loss": 0.772,
+      "step": 4213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099654776318509e-06,
+      "loss": 0.7294,
+      "step": 4214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095249796374742e-06,
+      "loss": 0.8309,
+      "step": 4215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0908465747950595e-06,
+      "loss": 0.7487,
+      "step": 4216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086445112890698e-06,
+      "loss": 0.6875,
+      "step": 4217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0820454119723495e-06,
+      "loss": 0.7973,
+      "step": 4218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077647473350201e-06,
+      "loss": 0.7979,
+      "step": 4219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073251298333901e-06,
+      "loss": 0.7121,
+      "step": 4220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068856888232582e-06,
+      "loss": 0.7735,
+      "step": 4221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064464244354841e-06,
+      "loss": 0.7528,
+      "step": 4222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060073368008762e-06,
+      "loss": 0.7349,
+      "step": 4223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055684260501889e-06,
+      "loss": 0.7925,
+      "step": 4224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051296923141252e-06,
+      "loss": 0.7499,
+      "step": 4225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046911357233343e-06,
+      "loss": 0.752,
+      "step": 4226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.04252756408413e-06,
+      "loss": 0.7989,
+      "step": 4227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038145544999057e-06,
+      "loss": 0.7083,
+      "step": 4228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0337653012830365e-06,
+      "loss": 0.7208,
+      "step": 4229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029386834240461e-06,
+      "loss": 0.7907,
+      "step": 4230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025010145175169e-06,
+      "loss": 0.2721,
+      "step": 4231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020635235390495e-06,
+      "loss": 0.2763,
+      "step": 4232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016262106189239e-06,
+      "loss": 0.7455,
+      "step": 4233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011890758873663e-06,
+      "loss": 0.2545,
+      "step": 4234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007521194745497e-06,
+      "loss": 0.7409,
+      "step": 4235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0031534151059504e-06,
+      "loss": 0.7185,
+      "step": 4236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998787421255698e-06,
+      "loss": 0.7707,
+      "step": 4237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994423214494874e-06,
+      "loss": 0.765,
+      "step": 4238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990060796123093e-06,
+      "loss": 0.7514,
+      "step": 4239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985700167439431e-06,
+      "loss": 0.7738,
+      "step": 4240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981341329742423e-06,
+      "loss": 0.7926,
+      "step": 4241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976984284330086e-06,
+      "loss": 0.8124,
+      "step": 4242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972629032499897e-06,
+      "loss": 0.7497,
+      "step": 4243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968275575548797e-06,
+      "loss": 0.6415,
+      "step": 4244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9639239147731865e-06,
+      "loss": 0.759,
+      "step": 4245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959574051468946e-06,
+      "loss": 0.7133,
+      "step": 4246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955225986931413e-06,
+      "loss": 0.6993,
+      "step": 4247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9508797224553874e-06,
+      "loss": 0.7102,
+      "step": 4248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946535259335131e-06,
+      "loss": 0.7328,
+      "step": 4249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942192598864375e-06,
+      "loss": 0.7536,
+      "step": 4250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9378517423363175e-06,
+      "loss": 0.7761,
+      "step": 4251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93351269104361e-06,
+      "loss": 0.6889,
+      "step": 4252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9291754462783635e-06,
+      "loss": 0.7609,
+      "step": 4253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924840009332169e-06,
+      "loss": 0.7722,
+      "step": 4254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92050638149606e-06,
+      "loss": 0.7857,
+      "step": 4255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916174564060543e-06,
+      "loss": 0.2622,
+      "step": 4256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911844558315577e-06,
+      "loss": 0.7325,
+      "step": 4257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9075163655505945e-06,
+      "loss": 0.7359,
+      "step": 4258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903189987054469e-06,
+      "loss": 0.258,
+      "step": 4259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898865424115553e-06,
+      "loss": 0.8155,
+      "step": 4260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894542678021643e-06,
+      "loss": 0.7478,
+      "step": 4261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890221750060008e-06,
+      "loss": 0.8359,
+      "step": 4262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885902641517362e-06,
+      "loss": 0.7019,
+      "step": 4263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881585353679891e-06,
+      "loss": 0.7686,
+      "step": 4264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877269887833227e-06,
+      "loss": 0.765,
+      "step": 4265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872956245262461e-06,
+      "loss": 0.8254,
+      "step": 4266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86864442725215e-06,
+      "loss": 0.282,
+      "step": 4267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864334435086302e-06,
+      "loss": 0.7921,
+      "step": 4268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86002627004838e-06,
+      "loss": 0.2771,
+      "step": 4269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855719933421301e-06,
+      "loss": 0.7628,
+      "step": 4270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851415426487443e-06,
+      "loss": 0.7337,
+      "step": 4271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847112750528639e-06,
+      "loss": 0.8261,
+      "step": 4272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842811906826173e-06,
+      "loss": 0.7242,
+      "step": 4273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83851289666078e-06,
+      "loss": 0.7624,
+      "step": 4274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83421572131266e-06,
+      "loss": 0.738,
+      "step": 4275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829920382061461e-06,
+      "loss": 0.6793,
+      "step": 4276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825626880186281e-06,
+      "loss": 0.713,
+      "step": 4277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82133521696567e-06,
+      "loss": 0.7536,
+      "step": 4278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817045393677643e-06,
+      "loss": 0.7534,
+      "step": 4279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812757411599648e-06,
+      "loss": 0.7705,
+      "step": 4280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8084712720086046e-06,
+      "loss": 0.7281,
+      "step": 4281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804186976180866e-06,
+      "loss": 0.7749,
+      "step": 4282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799904525392251e-06,
+      "loss": 0.7393,
+      "step": 4283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7956239209180157e-06,
+      "loss": 0.7881,
+      "step": 4284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.791345164032876e-06,
+      "loss": 0.2713,
+      "step": 4285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7870682560110006e-06,
+      "loss": 0.7168,
+      "step": 4286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7827931981259903e-06,
+      "loss": 0.7238,
+      "step": 4287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778519991650912e-06,
+      "loss": 0.8005,
+      "step": 4288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7742486378582756e-06,
+      "loss": 0.6318,
+      "step": 4289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7699791380200444e-06,
+      "loss": 0.7647,
+      "step": 4290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7657114934076146e-06,
+      "loss": 0.8323,
+      "step": 4291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761445705291843e-06,
+      "loss": 0.8059,
+      "step": 4292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757181774943035e-06,
+      "loss": 0.6854,
+      "step": 4293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7529197036309317e-06,
+      "loss": 0.6954,
+      "step": 4294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748659492624733e-06,
+      "loss": 0.7481,
+      "step": 4295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7444011431930737e-06,
+      "loss": 0.2378,
+      "step": 4296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7401446566040444e-06,
+      "loss": 0.7382,
+      "step": 4297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7358900341251704e-06,
+      "loss": 0.7834,
+      "step": 4298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731637277023433e-06,
+      "loss": 0.8116,
+      "step": 4299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727386386565246e-06,
+      "loss": 0.7587,
+      "step": 4300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7231373640164824e-06,
+      "loss": 0.7233,
+      "step": 4301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718890210642442e-06,
+      "loss": 0.7539,
+      "step": 4302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7146449277078824e-06,
+      "loss": 0.747,
+      "step": 4303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710401516476997e-06,
+      "loss": 0.7566,
+      "step": 4304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7061599782134172e-06,
+      "loss": 0.7799,
+      "step": 4305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7019203141802273e-06,
+      "loss": 0.7034,
+      "step": 4306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697682525639952e-06,
+      "loss": 0.7781,
+      "step": 4307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6934466138545487e-06,
+      "loss": 0.7244,
+      "step": 4308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6892125800854217e-06,
+      "loss": 0.6966,
+      "step": 4309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6849804255934164e-06,
+      "loss": 0.8039,
+      "step": 4310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680750151638821e-06,
+      "loss": 0.772,
+      "step": 4311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6765217594813584e-06,
+      "loss": 0.709,
+      "step": 4312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67229525038019e-06,
+      "loss": 0.2244,
+      "step": 4313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6680706255939223e-06,
+      "loss": 0.7205,
+      "step": 4314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6638478863806014e-06,
+      "loss": 0.7488,
+      "step": 4315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6596270339977068e-06,
+      "loss": 0.7985,
+      "step": 4316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6554080697021523e-06,
+      "loss": 0.2414,
+      "step": 4317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651190994750303e-06,
+      "loss": 0.6748,
+      "step": 4318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6469758103979468e-06,
+      "loss": 0.7079,
+      "step": 4319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6427625179003223e-06,
+      "loss": 0.7454,
+      "step": 4320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638551118512089e-06,
+      "loss": 0.7415,
+      "step": 4321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.634341613487361e-06,
+      "loss": 0.695,
+      "step": 4322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6301340040796705e-06,
+      "loss": 0.7358,
+      "step": 4323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625928291541999e-06,
+      "loss": 0.7514,
+      "step": 4324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6217244771267557e-06,
+      "loss": 0.6988,
+      "step": 4325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6175225620857824e-06,
+      "loss": 0.8016,
+      "step": 4326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6133225476703626e-06,
+      "loss": 0.7262,
+      "step": 4327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6091244351312127e-06,
+      "loss": 0.7716,
+      "step": 4328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6049282257184794e-06,
+      "loss": 0.802,
+      "step": 4329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600733920681738e-06,
+      "loss": 0.8179,
+      "step": 4330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5965415212700082e-06,
+      "loss": 0.7593,
+      "step": 4331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5923510287317377e-06,
+      "loss": 0.7939,
+      "step": 4332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588162444314803e-06,
+      "loss": 0.7269,
+      "step": 4333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58397576926651e-06,
+      "loss": 0.7014,
+      "step": 4334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5797910048336058e-06,
+      "loss": 0.7818,
+      "step": 4335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5756081522622643e-06,
+      "loss": 0.7414,
+      "step": 4336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5714272127980874e-06,
+      "loss": 0.758,
+      "step": 4337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5672481876861055e-06,
+      "loss": 0.787,
+      "step": 4338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563071078170789e-06,
+      "loss": 0.7748,
+      "step": 4339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558895885496023e-06,
+      "loss": 0.7708,
+      "step": 4340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.554722610905135e-06,
+      "loss": 0.243,
+      "step": 4341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5505512556408794e-06,
+      "loss": 0.7198,
+      "step": 4342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5463818209454325e-06,
+      "loss": 0.8111,
+      "step": 4343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542214308060399e-06,
+      "loss": 0.7607,
+      "step": 4344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5380487182268165e-06,
+      "loss": 0.759,
+      "step": 4345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5338850526851532e-06,
+      "loss": 0.7235,
+      "step": 4346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5297233126752938e-06,
+      "loss": 0.7674,
+      "step": 4347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5255634994365527e-06,
+      "loss": 0.7308,
+      "step": 4348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5214056142076756e-06,
+      "loss": 0.2525,
+      "step": 4349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5172496582268357e-06,
+      "loss": 0.7395,
+      "step": 4350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5130956327316214e-06,
+      "loss": 0.7417,
+      "step": 4351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5089435389590486e-06,
+      "loss": 0.6921,
+      "step": 4352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5047933781455666e-06,
+      "loss": 0.7608,
+      "step": 4353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5006451515270455e-06,
+      "loss": 0.7743,
+      "step": 4354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496498860338775e-06,
+      "loss": 0.7177,
+      "step": 4355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4923545058154685e-06,
+      "loss": 0.7858,
+      "step": 4356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48821208919127e-06,
+      "loss": 0.801,
+      "step": 4357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4840716116997363e-06,
+      "loss": 0.7592,
+      "step": 4358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479933074573858e-06,
+      "loss": 0.768,
+      "step": 4359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4757964790460365e-06,
+      "loss": 0.7442,
+      "step": 4360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471661826348106e-06,
+      "loss": 0.7655,
+      "step": 4361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4675291177113115e-06,
+      "loss": 0.6198,
+      "step": 4362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.46339835436633e-06,
+      "loss": 0.7562,
+      "step": 4363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4592695375432497e-06,
+      "loss": 0.2483,
+      "step": 4364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4551426684715805e-06,
+      "loss": 0.745,
+      "step": 4365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4510177483802577e-06,
+      "loss": 0.7389,
+      "step": 4366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446894778497636e-06,
+      "loss": 0.7459,
+      "step": 4367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4427737600514843e-06,
+      "loss": 0.7188,
+      "step": 4368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4386546942689893e-06,
+      "loss": 0.7831,
+      "step": 4369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434537582376761e-06,
+      "loss": 0.7567,
+      "step": 4370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4304224256008324e-06,
+      "loss": 0.7102,
+      "step": 4371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.426309225166643e-06,
+      "loss": 0.7002,
+      "step": 4372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4221979822990515e-06,
+      "loss": 0.8073,
+      "step": 4373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4180886982223404e-06,
+      "loss": 0.7979,
+      "step": 4374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413981374160209e-06,
+      "loss": 0.6449,
+      "step": 4375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4098760113357665e-06,
+      "loss": 0.7278,
+      "step": 4376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4057726109715362e-06,
+      "loss": 0.7687,
+      "step": 4377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401671174289469e-06,
+      "loss": 0.7462,
+      "step": 4378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397571702510918e-06,
+      "loss": 0.7625,
+      "step": 4379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3934741968566622e-06,
+      "loss": 0.7701,
+      "step": 4380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3893786585468834e-06,
+      "loss": 0.7473,
+      "step": 4381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3852850888011912e-06,
+      "loss": 0.7467,
+      "step": 4382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3811934888385957e-06,
+      "loss": 0.6638,
+      "step": 4383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377103859877532e-06,
+      "loss": 0.7042,
+      "step": 4384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373016203135836e-06,
+      "loss": 0.7471,
+      "step": 4385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368930519830771e-06,
+      "loss": 0.6972,
+      "step": 4386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3648468111789967e-06,
+      "loss": 0.793,
+      "step": 4387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360765078396597e-06,
+      "loss": 0.8227,
+      "step": 4388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3566853226990693e-06,
+      "loss": 0.7878,
+      "step": 4389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352607545301304e-06,
+      "loss": 0.7456,
+      "step": 4390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3485317474176193e-06,
+      "loss": 0.6754,
+      "step": 4391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3444579302617395e-06,
+      "loss": 0.7768,
+      "step": 4392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340386095046807e-06,
+      "loss": 0.7603,
+      "step": 4393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33631624298535e-06,
+      "loss": 0.2738,
+      "step": 4394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332248375289331e-06,
+      "loss": 0.742,
+      "step": 4395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3281824931701144e-06,
+      "loss": 0.7564,
+      "step": 4396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3241185978384636e-06,
+      "loss": 0.7443,
+      "step": 4397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3200566905045674e-06,
+      "loss": 0.6897,
+      "step": 4398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.315996772378004e-06,
+      "loss": 0.6931,
+      "step": 4399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311938844667777e-06,
+      "loss": 0.7515,
+      "step": 4400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3078829085822805e-06,
+      "loss": 0.6393,
+      "step": 4401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303828965329332e-06,
+      "loss": 0.6854,
+      "step": 4402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2997770161161437e-06,
+      "loss": 0.7127,
+      "step": 4403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295727062149334e-06,
+      "loss": 0.7648,
+      "step": 4404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291679104634935e-06,
+      "loss": 0.7142,
+      "step": 4405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2876331447783814e-06,
+      "loss": 0.6679,
+      "step": 4406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2835891837845114e-06,
+      "loss": 0.7831,
+      "step": 4407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.279547222857562e-06,
+      "loss": 0.7565,
+      "step": 4408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2755072632011864e-06,
+      "loss": 0.7729,
+      "step": 4409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2714693060184376e-06,
+      "loss": 0.771,
+      "step": 4410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26743335251177e-06,
+      "loss": 0.812,
+      "step": 4411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2633994038830374e-06,
+      "loss": 0.727,
+      "step": 4412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2593674613335058e-06,
+      "loss": 0.7763,
+      "step": 4413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255337526063842e-06,
+      "loss": 0.7369,
+      "step": 4414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2513095992741095e-06,
+      "loss": 0.7231,
+      "step": 4415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2472836821637744e-06,
+      "loss": 0.7213,
+      "step": 4416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243259775931712e-06,
+      "loss": 0.6725,
+      "step": 4417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239237881776187e-06,
+      "loss": 0.7675,
+      "step": 4418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2352180008948798e-06,
+      "loss": 0.8084,
+      "step": 4419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2312001344848543e-06,
+      "loss": 0.7443,
+      "step": 4420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.6826,
+      "step": 4421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2231704498639547e-06,
+      "loss": 0.721,
+      "step": 4422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219158634044225e-06,
+      "loss": 0.6575,
+      "step": 4423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215148837478065e-06,
+      "loss": 0.7633,
+      "step": 4424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2111410613595505e-06,
+      "loss": 0.7274,
+      "step": 4425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2071353068821456e-06,
+      "loss": 0.7643,
+      "step": 4426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203131575238718e-06,
+      "loss": 0.678,
+      "step": 4427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199129867621532e-06,
+      "loss": 0.7462,
+      "step": 4428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195130185222244e-06,
+      "loss": 0.7708,
+      "step": 4429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1911325292319138e-06,
+      "loss": 0.2439,
+      "step": 4430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1871369008410003e-06,
+      "loss": 0.2404,
+      "step": 4431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1831433012393486e-06,
+      "loss": 0.7333,
+      "step": 4432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1791517316162045e-06,
+      "loss": 0.7463,
+      "step": 4433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1751621931602104e-06,
+      "loss": 0.7604,
+      "step": 4434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1711746870594083e-06,
+      "loss": 0.7725,
+      "step": 4435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1671892145012262e-06,
+      "loss": 0.7828,
+      "step": 4436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1632057766724855e-06,
+      "loss": 0.7681,
+      "step": 4437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159224374759412e-06,
+      "loss": 0.7363,
+      "step": 4438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155245009947622e-06,
+      "loss": 0.7226,
+      "step": 4439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151267683422119e-06,
+      "loss": 0.7127,
+      "step": 4440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1472923963672996e-06,
+      "loss": 0.7426,
+      "step": 4441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1433191499669656e-06,
+      "loss": 0.7384,
+      "step": 4442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139347945404293e-06,
+      "loss": 0.6997,
+      "step": 4443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1353787838618644e-06,
+      "loss": 0.6393,
+      "step": 4444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131411666521651e-06,
+      "loss": 0.7559,
+      "step": 4445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12744659456501e-06,
+      "loss": 0.779,
+      "step": 4446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1234835691726904e-06,
+      "loss": 0.7172,
+      "step": 4447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119522591524835e-06,
+      "loss": 0.7467,
+      "step": 4448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1155636628009823e-06,
+      "loss": 0.2475,
+      "step": 4449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1116067841800424e-06,
+      "loss": 0.7201,
+      "step": 4450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1076519568403318e-06,
+      "loss": 0.69,
+      "step": 4451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103699181959552e-06,
+      "loss": 0.2277,
+      "step": 4452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0997484607147943e-06,
+      "loss": 0.6886,
+      "step": 4453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0957997942825337e-06,
+      "loss": 0.743,
+      "step": 4454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091853183838632e-06,
+      "loss": 0.7662,
+      "step": 4455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0879086305583504e-06,
+      "loss": 0.8449,
+      "step": 4456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083966135616323e-06,
+      "loss": 0.7888,
+      "step": 4457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080025700186583e-06,
+      "loss": 0.8054,
+      "step": 4458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0760873254425384e-06,
+      "loss": 0.8245,
+      "step": 4459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0721510125569984e-06,
+      "loss": 0.7832,
+      "step": 4460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068216762702142e-06,
+      "loss": 0.7217,
+      "step": 4461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0642845770495488e-06,
+      "loss": 0.7884,
+      "step": 4462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0603544567701706e-06,
+      "loss": 0.7205,
+      "step": 4463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056426403034356e-06,
+      "loss": 0.7758,
+      "step": 4464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052500417011827e-06,
+      "loss": 0.7584,
+      "step": 4465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0485764998717014e-06,
+      "loss": 0.7426,
+      "step": 4466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0446546527824705e-06,
+      "loss": 0.7518,
+      "step": 4467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0407348769120125e-06,
+      "loss": 0.2507,
+      "step": 4468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036817173427591e-06,
+      "loss": 0.7886,
+      "step": 4469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032901543495855e-06,
+      "loss": 0.8161,
+      "step": 4470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02898798828283e-06,
+      "loss": 0.7308,
+      "step": 4471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0250765089539225e-06,
+      "loss": 0.8048,
+      "step": 4472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021167106673928e-06,
+      "loss": 0.7396,
+      "step": 4473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0172597826070206e-06,
+      "loss": 0.7865,
+      "step": 4474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013354537916754e-06,
+      "loss": 0.7931,
+      "step": 4475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0094513737660602e-06,
+      "loss": 0.7092,
+      "step": 4476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0055502913172565e-06,
+      "loss": 0.7873,
+      "step": 4477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0016512917320427e-06,
+      "loss": 0.2624,
+      "step": 4478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9977543761714924e-06,
+      "loss": 0.6538,
+      "step": 4479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993859545796055e-06,
+      "loss": 0.7323,
+      "step": 4480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9899668017655716e-06,
+      "loss": 0.6982,
+      "step": 4481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9860761452392494e-06,
+      "loss": 0.7023,
+      "step": 4482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9821875773756847e-06,
+      "loss": 0.7888,
+      "step": 4483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978301099332841e-06,
+      "loss": 0.7647,
+      "step": 4484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974416712268071e-06,
+      "loss": 0.7984,
+      "step": 4485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9705344173380923e-06,
+      "loss": 0.7158,
+      "step": 4486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966654215699013e-06,
+      "loss": 0.7558,
+      "step": 4487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962776108506308e-06,
+      "loss": 0.7551,
+      "step": 4488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958900096914826e-06,
+      "loss": 0.7707,
+      "step": 4489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9550261820788016e-06,
+      "loss": 0.7123,
+      "step": 4490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9511543651518414e-06,
+      "loss": 0.8498,
+      "step": 4491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94728464728693e-06,
+      "loss": 0.7573,
+      "step": 4492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943417029636413e-06,
+      "loss": 0.7193,
+      "step": 4493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9395515133520247e-06,
+      "loss": 0.7233,
+      "step": 4494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935688099584875e-06,
+      "loss": 0.7876,
+      "step": 4495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931826789485438e-06,
+      "loss": 0.6714,
+      "step": 4496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927967584203564e-06,
+      "loss": 0.7593,
+      "step": 4497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9241104848884783e-06,
+      "loss": 0.6943,
+      "step": 4498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.920255492688786e-06,
+      "loss": 0.7532,
+      "step": 4499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916402608752449e-06,
+      "loss": 0.7761,
+      "step": 4500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9125518342268166e-06,
+      "loss": 0.829,
+      "step": 4501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9087031702585975e-06,
+      "loss": 0.7164,
+      "step": 4502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904856617993885e-06,
+      "loss": 0.7425,
+      "step": 4503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9010121785781285e-06,
+      "loss": 0.7899,
+      "step": 4504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8971698531561643e-06,
+      "loss": 0.7084,
+      "step": 4505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8933296428721857e-06,
+      "loss": 0.7907,
+      "step": 4506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8894915488697604e-06,
+      "loss": 0.7078,
+      "step": 4507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885655572291829e-06,
+      "loss": 0.7897,
+      "step": 4508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8818217142807036e-06,
+      "loss": 0.8023,
+      "step": 4509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877989975978056e-06,
+      "loss": 0.7238,
+      "step": 4510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8741603585249312e-06,
+      "loss": 0.7441,
+      "step": 4511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8703328630617455e-06,
+      "loss": 0.771,
+      "step": 4512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866507490728285e-06,
+      "loss": 0.7105,
+      "step": 4513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862684242663697e-06,
+      "loss": 0.7063,
+      "step": 4514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8588631200064954e-06,
+      "loss": 0.7801,
+      "step": 4515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8550441238945683e-06,
+      "loss": 0.7606,
+      "step": 4516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85122725546517e-06,
+      "loss": 0.7297,
+      "step": 4517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847412515854915e-06,
+      "loss": 0.7243,
+      "step": 4518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8435999061997855e-06,
+      "loss": 0.7315,
+      "step": 4519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8397894276351356e-06,
+      "loss": 0.7287,
+      "step": 4520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8359810812956746e-06,
+      "loss": 0.7717,
+      "step": 4521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8321748683154893e-06,
+      "loss": 0.7211,
+      "step": 4522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8283707898280176e-06,
+      "loss": 0.7459,
+      "step": 4523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.824568846966075e-06,
+      "loss": 0.7144,
+      "step": 4524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8207690408618294e-06,
+      "loss": 0.6849,
+      "step": 4525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816971372646823e-06,
+      "loss": 0.7043,
+      "step": 4526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813175843451952e-06,
+      "loss": 0.733,
+      "step": 4527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809382454407479e-06,
+      "loss": 0.6798,
+      "step": 4528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8055912066430314e-06,
+      "loss": 0.7021,
+      "step": 4529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8018021012875994e-06,
+      "loss": 0.7761,
+      "step": 4530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7980151394695333e-06,
+      "loss": 0.6701,
+      "step": 4531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79423032231654e-06,
+      "loss": 0.7521,
+      "step": 4532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7904476509556965e-06,
+      "loss": 0.7081,
+      "step": 4533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786667126513439e-06,
+      "loss": 0.8225,
+      "step": 4534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782888750115562e-06,
+      "loss": 0.7295,
+      "step": 4535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779112522887215e-06,
+      "loss": 0.682,
+      "step": 4536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7753384459529188e-06,
+      "loss": 0.8081,
+      "step": 4537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7715665204365506e-06,
+      "loss": 0.7733,
+      "step": 4538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767796747461341e-06,
+      "loss": 0.7419,
+      "step": 4539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76402912814988e-06,
+      "loss": 0.7168,
+      "step": 4540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7602636636241244e-06,
+      "loss": 0.6656,
+      "step": 4541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756500355005387e-06,
+      "loss": 0.7844,
+      "step": 4542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752739203414333e-06,
+      "loss": 0.7726,
+      "step": 4543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.748980209970985e-06,
+      "loss": 0.7682,
+      "step": 4544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7452233757947322e-06,
+      "loss": 0.8023,
+      "step": 4545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7414687020043094e-06,
+      "loss": 0.7504,
+      "step": 4546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7377161897178175e-06,
+      "loss": 0.7141,
+      "step": 4547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7339658400527115e-06,
+      "loss": 0.7528,
+      "step": 4548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7302176541257984e-06,
+      "loss": 0.6959,
+      "step": 4549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7264716330532402e-06,
+      "loss": 0.782,
+      "step": 4550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722727777950559e-06,
+      "loss": 0.7775,
+      "step": 4551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7189860899326383e-06,
+      "loss": 0.7267,
+      "step": 4552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7152465701136943e-06,
+      "loss": 0.7584,
+      "step": 4553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711509219607319e-06,
+      "loss": 0.7089,
+      "step": 4554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7077740395264484e-06,
+      "loss": 0.8745,
+      "step": 4555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7040410309833797e-06,
+      "loss": 0.7052,
+      "step": 4556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7003101950897538e-06,
+      "loss": 0.7416,
+      "step": 4557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696581532956567e-06,
+      "loss": 0.683,
+      "step": 4558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692855045694177e-06,
+      "loss": 0.6941,
+      "step": 4559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68913073441228e-06,
+      "loss": 0.7662,
+      "step": 4560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6854086002199364e-06,
+      "loss": 0.7365,
+      "step": 4561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6816886442255496e-06,
+      "loss": 0.7002,
+      "step": 4562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6779708675368833e-06,
+      "loss": 0.8178,
+      "step": 4563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674255271261039e-06,
+      "loss": 0.7919,
+      "step": 4564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670541856504486e-06,
+      "loss": 0.7217,
+      "step": 4565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6668306243730313e-06,
+      "loss": 0.7898,
+      "step": 4566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663121575971831e-06,
+      "loss": 0.7375,
+      "step": 4567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6594147124053983e-06,
+      "loss": 0.7184,
+      "step": 4568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655710034777598e-06,
+      "loss": 0.6197,
+      "step": 4569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6520075441916336e-06,
+      "loss": 0.2575,
+      "step": 4570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6483072417500613e-06,
+      "loss": 0.7535,
+      "step": 4571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644609128554788e-06,
+      "loss": 0.6894,
+      "step": 4572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640913205707073e-06,
+      "loss": 0.7009,
+      "step": 4573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637219474307514e-06,
+      "loss": 0.7819,
+      "step": 4574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633527935456056e-06,
+      "loss": 0.7686,
+      "step": 4575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6298385902520006e-06,
+      "loss": 0.7683,
+      "step": 4576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6261514397939914e-06,
+      "loss": 0.7283,
+      "step": 4577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6224664851800154e-06,
+      "loss": 0.7214,
+      "step": 4578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6187837275074055e-06,
+      "loss": 0.7589,
+      "step": 4579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6151031678728458e-06,
+      "loss": 0.7625,
+      "step": 4580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611424807372366e-06,
+      "loss": 0.7584,
+      "step": 4581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607748647101335e-06,
+      "loss": 0.7979,
+      "step": 4582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6040746881544665e-06,
+      "loss": 0.7329,
+      "step": 4583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6004029316258263e-06,
+      "loss": 0.2089,
+      "step": 4584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5967333786088157e-06,
+      "loss": 0.732,
+      "step": 4585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5930660301961885e-06,
+      "loss": 0.7406,
+      "step": 4586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5894008874800323e-06,
+      "loss": 0.7284,
+      "step": 4587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5857379515517875e-06,
+      "loss": 0.738,
+      "step": 4588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5820772235022273e-06,
+      "loss": 0.7339,
+      "step": 4589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578418704421479e-06,
+      "loss": 0.7642,
+      "step": 4590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574762395399002e-06,
+      "loss": 0.61,
+      "step": 4591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5711082975236e-06,
+      "loss": 0.7379,
+      "step": 4592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567456411883421e-06,
+      "loss": 0.7802,
+      "step": 4593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563806739565954e-06,
+      "loss": 0.2503,
+      "step": 4594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5601592816580345e-06,
+      "loss": 0.6916,
+      "step": 4595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5565140392458176e-06,
+      "loss": 0.7989,
+      "step": 4596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5528710134148203e-06,
+      "loss": 0.7002,
+      "step": 4597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.549230205249896e-06,
+      "loss": 0.7712,
+      "step": 4598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54559161583523e-06,
+      "loss": 0.7359,
+      "step": 4599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5419552462543484e-06,
+      "loss": 0.7389,
+      "step": 4600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5383210975901208e-06,
+      "loss": 0.7756,
+      "step": 4601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534689170924756e-06,
+      "loss": 0.7977,
+      "step": 4602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5310594673397926e-06,
+      "loss": 0.7507,
+      "step": 4603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52743198791612e-06,
+      "loss": 0.7339,
+      "step": 4604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5238067337339545e-06,
+      "loss": 0.7233,
+      "step": 4605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5201837058728506e-06,
+      "loss": 0.7679,
+      "step": 4606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5165629054117056e-06,
+      "loss": 0.8865,
+      "step": 4607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512944333428754e-06,
+      "loss": 0.8091,
+      "step": 4608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50932799100156e-06,
+      "loss": 0.8444,
+      "step": 4609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5057138792070246e-06,
+      "loss": 0.6959,
+      "step": 4610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502101999121389e-06,
+      "loss": 0.759,
+      "step": 4611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498492351820232e-06,
+      "loss": 0.6933,
+      "step": 4612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49488493837846e-06,
+      "loss": 0.7245,
+      "step": 4613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4912797598703155e-06,
+      "loss": 0.7579,
+      "step": 4614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4876768173693788e-06,
+      "loss": 0.7189,
+      "step": 4615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484076111948568e-06,
+      "loss": 0.6953,
+      "step": 4616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4804776446801272e-06,
+      "loss": 0.7774,
+      "step": 4617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4768814166356316e-06,
+      "loss": 0.7773,
+      "step": 4618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473287428886001e-06,
+      "loss": 0.6192,
+      "step": 4619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469695682501483e-06,
+      "loss": 0.7669,
+      "step": 4620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4661061785516537e-06,
+      "loss": 0.7787,
+      "step": 4621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4625189181054234e-06,
+      "loss": 0.7048,
+      "step": 4622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4589339022310386e-06,
+      "loss": 0.7042,
+      "step": 4623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455351131996071e-06,
+      "loss": 0.7952,
+      "step": 4624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451770608467432e-06,
+      "loss": 0.7486,
+      "step": 4625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4481923327113512e-06,
+      "loss": 0.6652,
+      "step": 4626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4446163057934035e-06,
+      "loss": 0.752,
+      "step": 4627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4410425287784823e-06,
+      "loss": 0.7722,
+      "step": 4628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437471002730819e-06,
+      "loss": 0.8481,
+      "step": 4629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433901728713971e-06,
+      "loss": 0.7158,
+      "step": 4630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4303347077908214e-06,
+      "loss": 0.2486,
+      "step": 4631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426769941023588e-06,
+      "loss": 0.7452,
+      "step": 4632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4232074294738206e-06,
+      "loss": 0.6743,
+      "step": 4633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4196471742023895e-06,
+      "loss": 0.7628,
+      "step": 4634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4160891762694926e-06,
+      "loss": 0.2871,
+      "step": 4635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412533436734663e-06,
+      "loss": 0.7619,
+      "step": 4636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4089799566567596e-06,
+      "loss": 0.3015,
+      "step": 4637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4054287370939633e-06,
+      "loss": 0.2611,
+      "step": 4638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4018797791037817e-06,
+      "loss": 0.731,
+      "step": 4639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3983330837430565e-06,
+      "loss": 0.7174,
+      "step": 4640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394788652067953e-06,
+      "loss": 0.7114,
+      "step": 4641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391246485133958e-06,
+      "loss": 0.805,
+      "step": 4642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3877065839958823e-06,
+      "loss": 0.7847,
+      "step": 4643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3841689497078746e-06,
+      "loss": 0.7294,
+      "step": 4644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3806335833233918e-06,
+      "loss": 0.7081,
+      "step": 4645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3771004858952317e-06,
+      "loss": 0.7959,
+      "step": 4646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3735696584755017e-06,
+      "loss": 0.7074,
+      "step": 4647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3700411021156454e-06,
+      "loss": 0.2398,
+      "step": 4648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36651481786642e-06,
+      "loss": 0.2579,
+      "step": 4649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3629908067779116e-06,
+      "loss": 0.6755,
+      "step": 4650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359469069899537e-06,
+      "loss": 0.7251,
+      "step": 4651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355949608280015e-06,
+      "loss": 0.7849,
+      "step": 4652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3524324229674046e-06,
+      "loss": 0.28,
+      "step": 4653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3489175150090825e-06,
+      "loss": 0.7295,
+      "step": 4654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3454048854517496e-06,
+      "loss": 0.7611,
+      "step": 4655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3418945353414167e-06,
+      "loss": 0.7743,
+      "step": 4656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338386465723428e-06,
+      "loss": 0.7808,
+      "step": 4657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334880677642445e-06,
+      "loss": 0.7405,
+      "step": 4658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3313771721424528e-06,
+      "loss": 0.74,
+      "step": 4659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32787595026675e-06,
+      "loss": 0.734,
+      "step": 4660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3243770130579566e-06,
+      "loss": 0.7251,
+      "step": 4661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32088036155802e-06,
+      "loss": 0.7395,
+      "step": 4662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317385996808195e-06,
+      "loss": 0.7847,
+      "step": 4663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3138939198490663e-06,
+      "loss": 0.7209,
+      "step": 4664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3104041317205296e-06,
+      "loss": 0.7367,
+      "step": 4665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3069166334618053e-06,
+      "loss": 0.2614,
+      "step": 4666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3034314261114256e-06,
+      "loss": 0.7033,
+      "step": 4667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2999485107072473e-06,
+      "loss": 0.7448,
+      "step": 4668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296467888286439e-06,
+      "loss": 0.7875,
+      "step": 4669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2929895598854847e-06,
+      "loss": 0.7175,
+      "step": 4670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2895135265401936e-06,
+      "loss": 0.7426,
+      "step": 4671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2860397892856877e-06,
+      "loss": 0.7106,
+      "step": 4672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282568349156403e-06,
+      "loss": 0.7097,
+      "step": 4673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2790992071860907e-06,
+      "loss": 0.7364,
+      "step": 4674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2756323644078213e-06,
+      "loss": 0.7489,
+      "step": 4675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272167821853982e-06,
+      "loss": 0.7535,
+      "step": 4676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2687055805562706e-06,
+      "loss": 0.7095,
+      "step": 4677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265245641545697e-06,
+      "loss": 0.732,
+      "step": 4678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2617880058525942e-06,
+      "loss": 0.7838,
+      "step": 4679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258332674506606e-06,
+      "loss": 0.705,
+      "step": 4680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2548796485366865e-06,
+      "loss": 0.8576,
+      "step": 4681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251428928971102e-06,
+      "loss": 0.7515,
+      "step": 4682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2479805168374436e-06,
+      "loss": 0.7334,
+      "step": 4683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244534413162599e-06,
+      "loss": 0.7122,
+      "step": 4684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2410906189727844e-06,
+      "loss": 0.7419,
+      "step": 4685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2376491352935126e-06,
+      "loss": 0.7678,
+      "step": 4686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234209963149625e-06,
+      "loss": 0.7111,
+      "step": 4687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2307731035652567e-06,
+      "loss": 0.7307,
+      "step": 4688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227338557563873e-06,
+      "loss": 0.7404,
+      "step": 4689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2239063261682346e-06,
+      "loss": 0.7844,
+      "step": 4690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2204764104004182e-06,
+      "loss": 0.676,
+      "step": 4691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2170488112818145e-06,
+      "loss": 0.7638,
+      "step": 4692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2136235298331234e-06,
+      "loss": 0.7506,
+      "step": 4693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2102005670743497e-06,
+      "loss": 0.7626,
+      "step": 4694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20677992402481e-06,
+      "loss": 0.7111,
+      "step": 4695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2033616017031312e-06,
+      "loss": 0.7629,
+      "step": 4696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1999456011272502e-06,
+      "loss": 0.6608,
+      "step": 4697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1965319233144177e-06,
+      "loss": 0.7764,
+      "step": 4698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193120569281173e-06,
+      "loss": 0.7919,
+      "step": 4699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1897115400433845e-06,
+      "loss": 0.7197,
+      "step": 4700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186304836616221e-06,
+      "loss": 0.6935,
+      "step": 4701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1829004600141555e-06,
+      "loss": 0.7649,
+      "step": 4702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17949841125097e-06,
+      "loss": 0.8167,
+      "step": 4703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176098691339754e-06,
+      "loss": 0.7331,
+      "step": 4704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1727013012929077e-06,
+      "loss": 0.7236,
+      "step": 4705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1693062421221287e-06,
+      "loss": 0.6958,
+      "step": 4706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165913514838429e-06,
+      "loss": 0.8106,
+      "step": 4707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1625231204521204e-06,
+      "loss": 0.7102,
+      "step": 4708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1591350599728178e-06,
+      "loss": 0.735,
+      "step": 4709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1557493344094494e-06,
+      "loss": 0.7288,
+      "step": 4710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152365944770246e-06,
+      "loss": 0.7249,
+      "step": 4711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148984892062739e-06,
+      "loss": 0.7702,
+      "step": 4712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1456061772937596e-06,
+      "loss": 0.7803,
+      "step": 4713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1422298014694553e-06,
+      "loss": 0.7411,
+      "step": 4714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13885576559527e-06,
+      "loss": 0.78,
+      "step": 4715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1354840706759515e-06,
+      "loss": 0.2446,
+      "step": 4716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1321147177155453e-06,
+      "loss": 0.7754,
+      "step": 4717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128747707717408e-06,
+      "loss": 0.7683,
+      "step": 4718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125383041684198e-06,
+      "loss": 0.8322,
+      "step": 4719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122020720617869e-06,
+      "loss": 0.6902,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1186607455196784e-06,
+      "loss": 0.76,
+      "step": 4721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1153031173901917e-06,
+      "loss": 0.7411,
+      "step": 4722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1119478372292646e-06,
+      "loss": 0.2776,
+      "step": 4723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.7728,
+      "step": 4724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1052443248090516e-06,
+      "loss": 0.7427,
+      "step": 4725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101896094545991e-06,
+      "loss": 0.7239,
+      "step": 4726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098550216243942e-06,
+      "loss": 0.6569,
+      "step": 4727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0952066908992718e-06,
+      "loss": 0.6854,
+      "step": 4728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0918655195076398e-06,
+      "loss": 0.7117,
+      "step": 4729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885267030640056e-06,
+      "loss": 0.6642,
+      "step": 4730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08519024256263e-06,
+      "loss": 0.6947,
+      "step": 4731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0818561389970737e-06,
+      "loss": 0.7692,
+      "step": 4732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078524393360192e-06,
+      "loss": 0.715,
+      "step": 4733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075195006644134e-06,
+      "loss": 0.7764,
+      "step": 4734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071867979840355e-06,
+      "loss": 0.77,
+      "step": 4735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068543313939606e-06,
+      "loss": 0.6749,
+      "step": 4736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0652210099319324e-06,
+      "loss": 0.7093,
+      "step": 4737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061901068806671e-06,
+      "loss": 0.7355,
+      "step": 4738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058583491552465e-06,
+      "loss": 0.7255,
+      "step": 4739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055268279157251e-06,
+      "loss": 0.725,
+      "step": 4740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051955432608258e-06,
+      "loss": 0.7233,
+      "step": 4741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0486449528920085e-06,
+      "loss": 0.7273,
+      "step": 4742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0453368409943277e-06,
+      "loss": 0.6727,
+      "step": 4743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0420310979003322e-06,
+      "loss": 0.7478,
+      "step": 4744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0387277245944327e-06,
+      "loss": 0.7745,
+      "step": 4745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035426722060331e-06,
+      "loss": 0.6814,
+      "step": 4746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321280912810302e-06,
+      "loss": 0.2515,
+      "step": 4747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0288318332388202e-06,
+      "loss": 0.7279,
+      "step": 4748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0255379489152893e-06,
+      "loss": 0.728,
+      "step": 4749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0222464392913144e-06,
+      "loss": 0.7771,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0189573053470723e-06,
+      "loss": 0.7332,
+      "step": 4751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015670548062023e-06,
+      "loss": 0.7103,
+      "step": 4752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012386168414927e-06,
+      "loss": 0.7758,
+      "step": 4753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009104167383833e-06,
+      "loss": 0.706,
+      "step": 4754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005824545946077e-06,
+      "loss": 0.69,
+      "step": 4755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002547305078295e-06,
+      "loss": 0.6239,
+      "step": 4756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9992724457564093e-06,
+      "loss": 0.7432,
+      "step": 4757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9959999689556407e-06,
+      "loss": 0.7127,
+      "step": 4758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992729875650482e-06,
+      "loss": 0.7248,
+      "step": 4759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989462166814734e-06,
+      "loss": 0.6529,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986196843421483e-06,
+      "loss": 0.2658,
+      "step": 4761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9829339064430974e-06,
+      "loss": 0.7721,
+      "step": 4762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9796733568512484e-06,
+      "loss": 0.6208,
+      "step": 4763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976415195616882e-06,
+      "loss": 0.7821,
+      "step": 4764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731594237102446e-06,
+      "loss": 0.7192,
+      "step": 4765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969906042100863e-06,
+      "loss": 0.7993,
+      "step": 4766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9666550517575577e-06,
+      "loss": 0.7604,
+      "step": 4767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634064536484343e-06,
+      "loss": 0.7449,
+      "step": 4768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9601602487408833e-06,
+      "loss": 0.6553,
+      "step": 4769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956916438001588e-06,
+      "loss": 0.7956,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9536750223965205e-06,
+      "loss": 0.7596,
+      "step": 4771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950436002890931e-06,
+      "loss": 0.7207,
+      "step": 4772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9471993804493595e-06,
+      "loss": 0.6948,
+      "step": 4773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9439651560356353e-06,
+      "loss": 0.7329,
+      "step": 4774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9407333306128752e-06,
+      "loss": 0.7585,
+      "step": 4775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9375039051434753e-06,
+      "loss": 0.7774,
+      "step": 4776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342768805891176e-06,
+      "loss": 0.7595,
+      "step": 4777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931052257910774e-06,
+      "loss": 0.7813,
+      "step": 4778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927830038068701e-06,
+      "loss": 0.7677,
+      "step": 4779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924610222022434e-06,
+      "loss": 0.7305,
+      "step": 4780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9213928107307945e-06,
+      "loss": 0.7201,
+      "step": 4781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9181778051518908e-06,
+      "loss": 0.8212,
+      "step": 4782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9149652062431155e-06,
+      "loss": 0.6439,
+      "step": 4783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9117550149611407e-06,
+      "loss": 0.7184,
+      "step": 4784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90854723226192e-06,
+      "loss": 0.7602,
+      "step": 4785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9053418591006977e-06,
+      "loss": 0.7431,
+      "step": 4786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9021388964319909e-06,
+      "loss": 0.8076,
+      "step": 4787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8989383452096076e-06,
+      "loss": 0.6826,
+      "step": 4788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8957402063866304e-06,
+      "loss": 0.7451,
+      "step": 4789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8925444809154303e-06,
+      "loss": 0.6764,
+      "step": 4790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.889351169747652e-06,
+      "loss": 0.6859,
+      "step": 4791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8861602738342299e-06,
+      "loss": 0.8415,
+      "step": 4792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8829717941253733e-06,
+      "loss": 0.7528,
+      "step": 4793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797857315705703e-06,
+      "loss": 0.7531,
+      "step": 4794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876602087118593e-06,
+      "loss": 0.7209,
+      "step": 4795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8734208617174986e-06,
+      "loss": 0.7833,
+      "step": 4796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8702420563146128e-06,
+      "loss": 0.7955,
+      "step": 4797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8670656718565449e-06,
+      "loss": 0.752,
+      "step": 4798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863891709289185e-06,
+      "loss": 0.758,
+      "step": 4799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.860720169557706e-06,
+      "loss": 0.8033,
+      "step": 4800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857551053606551e-06,
+      "loss": 0.6981,
+      "step": 4801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8543843623794422e-06,
+      "loss": 0.7511,
+      "step": 4802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.851220096819385e-06,
+      "loss": 0.7373,
+      "step": 4803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8480582578686624e-06,
+      "loss": 0.7308,
+      "step": 4804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84489884646883e-06,
+      "loss": 0.7292,
+      "step": 4805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8417418635607197e-06,
+      "loss": 0.7758,
+      "step": 4806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8385873100844498e-06,
+      "loss": 0.6899,
+      "step": 4807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8354351869794017e-06,
+      "loss": 0.7811,
+      "step": 4808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8322854951842462e-06,
+      "loss": 0.6968,
+      "step": 4809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.829138235636918e-06,
+      "loss": 0.7023,
+      "step": 4810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8259934092746401e-06,
+      "loss": 0.7235,
+      "step": 4811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8228510170338964e-06,
+      "loss": 0.7241,
+      "step": 4812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8197110598504597e-06,
+      "loss": 0.7232,
+      "step": 4813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.816573538659372e-06,
+      "loss": 0.7468,
+      "step": 4814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813438454394948e-06,
+      "loss": 0.7407,
+      "step": 4815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8103058079907753e-06,
+      "loss": 0.796,
+      "step": 4816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8071756003797203e-06,
+      "loss": 0.7417,
+      "step": 4817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.804047832493927e-06,
+      "loss": 0.6706,
+      "step": 4818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8009225052648015e-06,
+      "loss": 0.7653,
+      "step": 4819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7977996196230274e-06,
+      "loss": 0.7057,
+      "step": 4820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7946791764985661e-06,
+      "loss": 0.6884,
+      "step": 4821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.791561176820651e-06,
+      "loss": 0.2443,
+      "step": 4822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7884456215177814e-06,
+      "loss": 0.7205,
+      "step": 4823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78533251151773e-06,
+      "loss": 0.7333,
+      "step": 4824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7822218477475496e-06,
+      "loss": 0.7234,
+      "step": 4825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7791136311335532e-06,
+      "loss": 0.6993,
+      "step": 4826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7760078626013366e-06,
+      "loss": 0.7527,
+      "step": 4827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772904543075754e-06,
+      "loss": 0.7404,
+      "step": 4828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.769803673480942e-06,
+      "loss": 0.6628,
+      "step": 4829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7667052547402996e-06,
+      "loss": 0.7521,
+      "step": 4830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763609287776501e-06,
+      "loss": 0.2505,
+      "step": 4831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7605157735114885e-06,
+      "loss": 0.2906,
+      "step": 4832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757424712866469e-06,
+      "loss": 0.7431,
+      "step": 4833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7543361067619269e-06,
+      "loss": 0.7594,
+      "step": 4834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7512499561176154e-06,
+      "loss": 0.7406,
+      "step": 4835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7481662618525486e-06,
+      "loss": 0.7456,
+      "step": 4836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7450850248850138e-06,
+      "loss": 0.7709,
+      "step": 4837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7420062461325683e-06,
+      "loss": 0.7783,
+      "step": 4838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7389299265120362e-06,
+      "loss": 0.6729,
+      "step": 4839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7358560669395097e-06,
+      "loss": 0.719,
+      "step": 4840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7327846683303418e-06,
+      "loss": 0.7726,
+      "step": 4841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729715731599163e-06,
+      "loss": 0.6922,
+      "step": 4842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726649257659867e-06,
+      "loss": 0.6785,
+      "step": 4843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723585247425612e-06,
+      "loss": 0.7253,
+      "step": 4844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7205237018088205e-06,
+      "loss": 0.7367,
+      "step": 4845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7174646217211888e-06,
+      "loss": 0.7377,
+      "step": 4846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7144080080736701e-06,
+      "loss": 0.8009,
+      "step": 4847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7113538617764913e-06,
+      "loss": 0.7217,
+      "step": 4848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083021837391366e-06,
+      "loss": 0.716,
+      "step": 4849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705252974870365e-06,
+      "loss": 0.7162,
+      "step": 4850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022062360781887e-06,
+      "loss": 0.2713,
+      "step": 4851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6991619682698946e-06,
+      "loss": 0.7319,
+      "step": 4852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6961201723520248e-06,
+      "loss": 0.7341,
+      "step": 4853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.693080849230395e-06,
+      "loss": 0.7524,
+      "step": 4854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6900439998100737e-06,
+      "loss": 0.792,
+      "step": 4855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6870096249954038e-06,
+      "loss": 0.8156,
+      "step": 4856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6839777256899836e-06,
+      "loss": 0.721,
+      "step": 4857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6809483027966745e-06,
+      "loss": 0.7457,
+      "step": 4858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6779213572176033e-06,
+      "loss": 0.7445,
+      "step": 4859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6748968898541596e-06,
+      "loss": 0.7273,
+      "step": 4860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.671874901606998e-06,
+      "loss": 0.7746,
+      "step": 4861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668855393376021e-06,
+      "loss": 0.7359,
+      "step": 4862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665838366060406e-06,
+      "loss": 0.2438,
+      "step": 4863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6628238205585922e-06,
+      "loss": 0.7534,
+      "step": 4864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6598117577682715e-06,
+      "loss": 0.7493,
+      "step": 4865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6568021785863975e-06,
+      "loss": 0.8157,
+      "step": 4866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6537950839091887e-06,
+      "loss": 0.7487,
+      "step": 4867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6507904746321268e-06,
+      "loss": 0.7235,
+      "step": 4868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6477883516499416e-06,
+      "loss": 0.7224,
+      "step": 4869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6447887158566355e-06,
+      "loss": 0.7448,
+      "step": 4870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641791568145461e-06,
+      "loss": 0.7229,
+      "step": 4871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387969094089318e-06,
+      "loss": 0.7035,
+      "step": 4872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6358047405388221e-06,
+      "loss": 0.747,
+      "step": 4873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6328150624261673e-06,
+      "loss": 0.7572,
+      "step": 4874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6298278759612562e-06,
+      "loss": 0.7288,
+      "step": 4875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6268431820336327e-06,
+      "loss": 0.7969,
+      "step": 4876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6238609815321083e-06,
+      "loss": 0.741,
+      "step": 4877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208812753447466e-06,
+      "loss": 0.7226,
+      "step": 4878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6179040643588672e-06,
+      "loss": 0.7727,
+      "step": 4879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149293494610462e-06,
+      "loss": 0.6717,
+      "step": 4880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6119571315371197e-06,
+      "loss": 0.6858,
+      "step": 4881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089874114721804e-06,
+      "loss": 0.7749,
+      "step": 4882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6060201901505744e-06,
+      "loss": 0.7348,
+      "step": 4883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603055468455903e-06,
+      "loss": 0.646,
+      "step": 4884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6000932472710274e-06,
+      "loss": 0.6904,
+      "step": 4885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5971335274780574e-06,
+      "loss": 0.7638,
+      "step": 4886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5941763099583696e-06,
+      "loss": 0.6885,
+      "step": 4887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5912215955925802e-06,
+      "loss": 0.76,
+      "step": 4888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5882693852605735e-06,
+      "loss": 0.6789,
+      "step": 4889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5853196798414783e-06,
+      "loss": 0.6795,
+      "step": 4890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5823724802136863e-06,
+      "loss": 0.2352,
+      "step": 4891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5794277872548336e-06,
+      "loss": 0.7363,
+      "step": 4892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5764856018418195e-06,
+      "loss": 0.7968,
+      "step": 4893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5735459248507857e-06,
+      "loss": 0.7642,
+      "step": 4894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706087571571394e-06,
+      "loss": 0.7283,
+      "step": 4895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5676740996355299e-06,
+      "loss": 0.7736,
+      "step": 4896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5647419531598619e-06,
+      "loss": 0.6741,
+      "step": 4897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5618123186032963e-06,
+      "loss": 0.7672,
+      "step": 4898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5588851968382447e-06,
+      "loss": 0.7655,
+      "step": 4899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5559605887363672e-06,
+      "loss": 0.8254,
+      "step": 4900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.553038495168574e-06,
+      "loss": 0.7098,
+      "step": 4901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5501189170050334e-06,
+      "loss": 0.7543,
+      "step": 4902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5472018551151624e-06,
+      "loss": 0.7395,
+      "step": 4903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.544287310367625e-06,
+      "loss": 0.7105,
+      "step": 4904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5413752836303352e-06,
+      "loss": 0.7721,
+      "step": 4905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5384657757704625e-06,
+      "loss": 0.6837,
+      "step": 4906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5355587876544264e-06,
+      "loss": 0.7744,
+      "step": 4907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5326543201478895e-06,
+      "loss": 0.6804,
+      "step": 4908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5297523741157671e-06,
+      "loss": 0.8335,
+      "step": 4909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5268529504222262e-06,
+      "loss": 0.7693,
+      "step": 4910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5239560499306781e-06,
+      "loss": 0.7502,
+      "step": 4911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5210616735037897e-06,
+      "loss": 0.7297,
+      "step": 4912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5181698220034658e-06,
+      "loss": 0.679,
+      "step": 4913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5152804962908708e-06,
+      "loss": 0.7977,
+      "step": 4914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5123936972264054e-06,
+      "loss": 0.2409,
+      "step": 4915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5095094256697285e-06,
+      "loss": 0.6893,
+      "step": 4916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5066276824797444e-06,
+      "loss": 0.7144,
+      "step": 4917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5037484685145943e-06,
+      "loss": 0.8289,
+      "step": 4918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5008717846316768e-06,
+      "loss": 0.7112,
+      "step": 4919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4979976316876333e-06,
+      "loss": 0.709,
+      "step": 4920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4951260105383591e-06,
+      "loss": 0.782,
+      "step": 4921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4922569220389793e-06,
+      "loss": 0.7175,
+      "step": 4922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489390367043877e-06,
+      "loss": 0.7813,
+      "step": 4923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4865263464066814e-06,
+      "loss": 0.7046,
+      "step": 4924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4836648609802584e-06,
+      "loss": 0.843,
+      "step": 4925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4808059116167306e-06,
+      "loss": 0.6692,
+      "step": 4926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4779494991674526e-06,
+      "loss": 0.7148,
+      "step": 4927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475095624483034e-06,
+      "loss": 0.7147,
+      "step": 4928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4722442884133214e-06,
+      "loss": 0.8189,
+      "step": 4929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4693954918074137e-06,
+      "loss": 0.713,
+      "step": 4930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4665492355136423e-06,
+      "loss": 0.7017,
+      "step": 4931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4637055203795936e-06,
+      "loss": 0.7363,
+      "step": 4932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4608643472520878e-06,
+      "loss": 0.7178,
+      "step": 4933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.458025716977196e-06,
+      "loss": 0.7723,
+      "step": 4934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4551896304002279e-06,
+      "loss": 0.7764,
+      "step": 4935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4523560883657317e-06,
+      "loss": 0.8345,
+      "step": 4936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4495250917175052e-06,
+      "loss": 0.765,
+      "step": 4937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4466966412985895e-06,
+      "loss": 0.6647,
+      "step": 4938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4438707379512583e-06,
+      "loss": 0.7722,
+      "step": 4939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4410473825170312e-06,
+      "loss": 0.752,
+      "step": 4940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438226575836673e-06,
+      "loss": 0.7977,
+      "step": 4941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4354083187501865e-06,
+      "loss": 0.714,
+      "step": 4942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4325926120968137e-06,
+      "loss": 0.791,
+      "step": 4943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4297794567150359e-06,
+      "loss": 0.7107,
+      "step": 4944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4269688534425786e-06,
+      "loss": 0.7371,
+      "step": 4945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.42416080311641e-06,
+      "loss": 0.7657,
+      "step": 4946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4213553065727314e-06,
+      "loss": 0.7371,
+      "step": 4947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4185523646469822e-06,
+      "loss": 0.7371,
+      "step": 4948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4157519781738515e-06,
+      "loss": 0.7022,
+      "step": 4949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4129541479872545e-06,
+      "loss": 0.7523,
+      "step": 4950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4101588749203576e-06,
+      "loss": 0.7226,
+      "step": 4951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4073661598055555e-06,
+      "loss": 0.6965,
+      "step": 4952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4045760034744881e-06,
+      "loss": 0.6899,
+      "step": 4953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.401788406758028e-06,
+      "loss": 0.7453,
+      "step": 4954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3990033704862916e-06,
+      "loss": 0.7422,
+      "step": 4955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396220895488627e-06,
+      "loss": 0.6921,
+      "step": 4956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3934409825936213e-06,
+      "loss": 0.7374,
+      "step": 4957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3906636326291001e-06,
+      "loss": 0.7596,
+      "step": 4958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3878888464221285e-06,
+      "loss": 0.7271,
+      "step": 4959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3851166247990011e-06,
+      "loss": 0.728,
+      "step": 4960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3823469685852509e-06,
+      "loss": 0.7913,
+      "step": 4961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3795798786056503e-06,
+      "loss": 0.2925,
+      "step": 4962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376815355684209e-06,
+      "loss": 0.7462,
+      "step": 4963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3740534006441652e-06,
+      "loss": 0.6891,
+      "step": 4964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3712940143079945e-06,
+      "loss": 0.7763,
+      "step": 4965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3685371974974115e-06,
+      "loss": 0.7218,
+      "step": 4966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3657829510333653e-06,
+      "loss": 0.7357,
+      "step": 4967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3630312757360364e-06,
+      "loss": 0.8005,
+      "step": 4968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3602821724248373e-06,
+      "loss": 0.7346,
+      "step": 4969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3575356419184204e-06,
+      "loss": 0.7178,
+      "step": 4970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547916850346732e-06,
+      "loss": 0.7824,
+      "step": 4971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3520503025907073e-06,
+      "loss": 0.732,
+      "step": 4972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3493114954028796e-06,
+      "loss": 0.7623,
+      "step": 4973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3465752642867702e-06,
+      "loss": 0.7605,
+      "step": 4974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343841610057196e-06,
+      "loss": 0.7272,
+      "step": 4975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3411105335282071e-06,
+      "loss": 0.7179,
+      "step": 4976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3383820355130872e-06,
+      "loss": 0.7944,
+      "step": 4977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3356561168243499e-06,
+      "loss": 0.7832,
+      "step": 4978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332932778273739e-06,
+      "loss": 0.8113,
+      "step": 4979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3302120206722324e-06,
+      "loss": 0.6898,
+      "step": 4980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3274938448300435e-06,
+      "loss": 0.7348,
+      "step": 4981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32477825155661e-06,
+      "loss": 0.6627,
+      "step": 4982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3220652416605995e-06,
+      "loss": 0.7555,
+      "step": 4983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3193548159499181e-06,
+      "loss": 0.6753,
+      "step": 4984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.316646975231699e-06,
+      "loss": 0.8013,
+      "step": 4985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313941720312303e-06,
+      "loss": 0.7203,
+      "step": 4986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3112390519973207e-06,
+      "loss": 0.673,
+      "step": 4987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3085389710915796e-06,
+      "loss": 0.6698,
+      "step": 4988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3058414783991257e-06,
+      "loss": 0.7778,
+      "step": 4989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3031465747232453e-06,
+      "loss": 0.7871,
+      "step": 4990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300454260866444e-06,
+      "loss": 0.7312,
+      "step": 4991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297764537630465e-06,
+      "loss": 0.7649,
+      "step": 4992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2950774058162708e-06,
+      "loss": 0.7594,
+      "step": 4993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2923928662240625e-06,
+      "loss": 0.7757,
+      "step": 4994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289710919653262e-06,
+      "loss": 0.7421,
+      "step": 4995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287031566902518e-06,
+      "loss": 0.7091,
+      "step": 4996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2843548087697123e-06,
+      "loss": 0.7,
+      "step": 4997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2816806460519526e-06,
+      "loss": 0.7972,
+      "step": 4998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2790090795455724e-06,
+      "loss": 0.8579,
+      "step": 4999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2763401100461281e-06,
+      "loss": 0.8454,
+      "step": 5000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273673738348411e-06,
+      "loss": 0.7919,
+      "step": 5001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710099652464358e-06,
+      "loss": 0.7728,
+      "step": 5002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2683487915334402e-06,
+      "loss": 0.7706,
+      "step": 5003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656902180018881e-06,
+      "loss": 0.7524,
+      "step": 5004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263034245443473e-06,
+      "loss": 0.7066,
+      "step": 5005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260380874649113e-06,
+      "loss": 0.7498,
+      "step": 5006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2577301064089498e-06,
+      "loss": 0.7157,
+      "step": 5007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2550819415123483e-06,
+      "loss": 0.7567,
+      "step": 5008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2524363807479023e-06,
+      "loss": 0.7592,
+      "step": 5009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249793424903427e-06,
+      "loss": 0.7731,
+      "step": 5010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471530747659655e-06,
+      "loss": 0.2418,
+      "step": 5011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2445153311217783e-06,
+      "loss": 0.7456,
+      "step": 5012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2418801947563586e-06,
+      "loss": 0.7235,
+      "step": 5013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.239247666454415e-06,
+      "loss": 0.6385,
+      "step": 5014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366177469998853e-06,
+      "loss": 0.718,
+      "step": 5015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2339904371759259e-06,
+      "loss": 0.7457,
+      "step": 5016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313657377649212e-06,
+      "loss": 0.757,
+      "step": 5017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2287436495484727e-06,
+      "loss": 0.7192,
+      "step": 5018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.226124173307407e-06,
+      "loss": 0.768,
+      "step": 5019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2235073098217786e-06,
+      "loss": 0.7515,
+      "step": 5020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2208930598708491e-06,
+      "loss": 0.7528,
+      "step": 5021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182814242331143e-06,
+      "loss": 0.7575,
+      "step": 5022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2156724036862898e-06,
+      "loss": 0.7297,
+      "step": 5023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2130659990073146e-06,
+      "loss": 0.2416,
+      "step": 5024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2104622109723352e-06,
+      "loss": 0.6988,
+      "step": 5025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2078610403567337e-06,
+      "loss": 0.7751,
+      "step": 5026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.7457,
+      "step": 5027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2026665544812777e-06,
+      "loss": 0.7115,
+      "step": 5028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200073240768278e-06,
+      "loss": 0.8047,
+      "step": 5029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197482547568366e-06,
+      "loss": 0.6601,
+      "step": 5030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1948944756530223e-06,
+      "loss": 0.7165,
+      "step": 5031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1923090257929404e-06,
+      "loss": 0.2703,
+      "step": 5032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.18972619875804e-06,
+      "loss": 0.7242,
+      "step": 5033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1871459953174547e-06,
+      "loss": 0.2688,
+      "step": 5034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845684162395344e-06,
+      "loss": 0.8101,
+      "step": 5035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.181993462291856e-06,
+      "loss": 0.7479,
+      "step": 5036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1794211342412099e-06,
+      "loss": 0.7418,
+      "step": 5037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1768514328536041e-06,
+      "loss": 0.6669,
+      "step": 5038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1742843588942622e-06,
+      "loss": 0.7528,
+      "step": 5039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1717199131276314e-06,
+      "loss": 0.7766,
+      "step": 5040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1691580963173733e-06,
+      "loss": 0.7254,
+      "step": 5041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1665989092263642e-06,
+      "loss": 0.7905,
+      "step": 5042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640423526166987e-06,
+      "loss": 0.8071,
+      "step": 5043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1614884272496918e-06,
+      "loss": 0.7042,
+      "step": 5044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1589371338858713e-06,
+      "loss": 0.7524,
+      "step": 5045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1563884732849818e-06,
+      "loss": 0.7632,
+      "step": 5046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1538424462059805e-06,
+      "loss": 0.7276,
+      "step": 5047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.15129905340705e-06,
+      "loss": 0.7661,
+      "step": 5048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1487582956455767e-06,
+      "loss": 0.7839,
+      "step": 5049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146220173678172e-06,
+      "loss": 0.729,
+      "step": 5050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1436846882606546e-06,
+      "loss": 0.7676,
+      "step": 5051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141151840148066e-06,
+      "loss": 0.7295,
+      "step": 5052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138621630094654e-06,
+      "loss": 0.2392,
+      "step": 5053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1360940588538883e-06,
+      "loss": 0.6196,
+      "step": 5054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335691271784465e-06,
+      "loss": 0.8006,
+      "step": 5055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1310468358202265e-06,
+      "loss": 0.7216,
+      "step": 5056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1285271855303315e-06,
+      "loss": 0.7652,
+      "step": 5057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1260101770590893e-06,
+      "loss": 0.68,
+      "step": 5058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234958111560324e-06,
+      "loss": 0.7204,
+      "step": 5059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1209840885699065e-06,
+      "loss": 0.6552,
+      "step": 5060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118475010048674e-06,
+      "loss": 0.7269,
+      "step": 5061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159685763395113e-06,
+      "loss": 0.7015,
+      "step": 5062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1134647881888017e-06,
+      "loss": 0.7621,
+      "step": 5063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1109636463421425e-06,
+      "loss": 0.7087,
+      "step": 5064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1084651515443457e-06,
+      "loss": 0.7633,
+      "step": 5065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.105969304539435e-06,
+      "loss": 0.7251,
+      "step": 5066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1034761060706411e-06,
+      "loss": 0.7203,
+      "step": 5067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009855568804074e-06,
+      "loss": 0.7321,
+      "step": 5068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0984976577103924e-06,
+      "loss": 0.7782,
+      "step": 5069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0960124093014624e-06,
+      "loss": 0.7739,
+      "step": 5070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935298123936955e-06,
+      "loss": 0.7867,
+      "step": 5071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0910498677263758e-06,
+      "loss": 0.7428,
+      "step": 5072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0885725760380062e-06,
+      "loss": 0.6855,
+      "step": 5073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0860979380662894e-06,
+      "loss": 0.2713,
+      "step": 5074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0836259545481453e-06,
+      "loss": 0.2617,
+      "step": 5075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811566262197037e-06,
+      "loss": 0.7385,
+      "step": 5076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0786899538162988e-06,
+      "loss": 0.7274,
+      "step": 5077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0762259380724739e-06,
+      "loss": 0.8029,
+      "step": 5078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0737645797219865e-06,
+      "loss": 0.6943,
+      "step": 5079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0713058794978026e-06,
+      "loss": 0.7666,
+      "step": 5080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0688498381320855e-06,
+      "loss": 0.2749,
+      "step": 5081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06639645635622e-06,
+      "loss": 0.7458,
+      "step": 5082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.063945734900793e-06,
+      "loss": 0.7358,
+      "step": 5083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061497674495603e-06,
+      "loss": 0.6965,
+      "step": 5084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590522758696508e-06,
+      "loss": 0.7523,
+      "step": 5085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0566095397511445e-06,
+      "loss": 0.643,
+      "step": 5086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054169466867505e-06,
+      "loss": 0.6911,
+      "step": 5087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051732057945356e-06,
+      "loss": 0.788,
+      "step": 5088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0492973137105289e-06,
+      "loss": 0.7764,
+      "step": 5089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0468652348880604e-06,
+      "loss": 0.7194,
+      "step": 5090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0444358222021965e-06,
+      "loss": 0.7615,
+      "step": 5091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0420090763763846e-06,
+      "loss": 0.725,
+      "step": 5092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.039584998133284e-06,
+      "loss": 0.691,
+      "step": 5093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371635881947528e-06,
+      "loss": 0.7169,
+      "step": 5094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0347448472818612e-06,
+      "loss": 0.7915,
+      "step": 5095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032328776114877e-06,
+      "loss": 0.8182,
+      "step": 5096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0299153754132829e-06,
+      "loss": 0.6694,
+      "step": 5097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0275046458957582e-06,
+      "loss": 0.7578,
+      "step": 5098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250965882801867e-06,
+      "loss": 0.7805,
+      "step": 5099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.022691203283661e-06,
+      "loss": 0.6745,
+      "step": 5100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0202884916224797e-06,
+      "loss": 0.7548,
+      "step": 5101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017888454012138e-06,
+      "loss": 0.7534,
+      "step": 5102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0154910911673365e-06,
+      "loss": 0.7521,
+      "step": 5103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0130964038019853e-06,
+      "loss": 0.7865,
+      "step": 5104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107043926291926e-06,
+      "loss": 0.7405,
+      "step": 5105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083150583612711e-06,
+      "loss": 0.6795,
+      "step": 5106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0059284017097327e-06,
+      "loss": 0.6898,
+      "step": 5107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0035444233852976e-06,
+      "loss": 0.7123,
+      "step": 5108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011631240978892e-06,
+      "loss": 0.704,
+      "step": 5109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.987845045566258e-07,
+      "loss": 0.7633,
+      "step": 5110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.96408565469832e-07,
+      "loss": 0.7243,
+      "step": 5111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.940353075450371e-07,
+      "loss": 0.8281,
+      "step": 5112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.916647314889648e-07,
+      "loss": 0.6702,
+      "step": 5113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.892968380075485e-07,
+      "loss": 0.7606,
+      "step": 5114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.869316278059149e-07,
+      "loss": 0.716,
+      "step": 5115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.845691015883985e-07,
+      "loss": 0.7048,
+      "step": 5116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.822092600585287e-07,
+      "loss": 0.7272,
+      "step": 5117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.798521039190413e-07,
+      "loss": 0.7393,
+      "step": 5118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77497633871868e-07,
+      "loss": 0.7082,
+      "step": 5119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.751458506181376e-07,
+      "loss": 0.7038,
+      "step": 5120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.727967548581885e-07,
+      "loss": 0.7118,
+      "step": 5121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704503472915505e-07,
+      "loss": 0.7272,
+      "step": 5122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.681066286169604e-07,
+      "loss": 0.7118,
+      "step": 5123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.657655995323434e-07,
+      "loss": 0.6724,
+      "step": 5124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.634272607348315e-07,
+      "loss": 0.8203,
+      "step": 5125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.610916129207581e-07,
+      "loss": 0.7525,
+      "step": 5126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587586567856478e-07,
+      "loss": 0.7677,
+      "step": 5127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.564283930242258e-07,
+      "loss": 0.697,
+      "step": 5128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.541008223304181e-07,
+      "loss": 0.7321,
+      "step": 5129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517759453973507e-07,
+      "loss": 0.6779,
+      "step": 5130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494537629173395e-07,
+      "loss": 0.7294,
+      "step": 5131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471342755819068e-07,
+      "loss": 0.7057,
+      "step": 5132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448174840817637e-07,
+      "loss": 0.7575,
+      "step": 5133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425033891068269e-07,
+      "loss": 0.7778,
+      "step": 5134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.401919913462044e-07,
+      "loss": 0.7307,
+      "step": 5135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.378832914882041e-07,
+      "loss": 0.7918,
+      "step": 5136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.355772902203286e-07,
+      "loss": 0.6232,
+      "step": 5137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332739882292752e-07,
+      "loss": 0.7331,
+      "step": 5138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.309733862009418e-07,
+      "loss": 0.7121,
+      "step": 5139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.286754848204216e-07,
+      "loss": 0.803,
+      "step": 5140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.263802847720016e-07,
+      "loss": 0.7376,
+      "step": 5141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.240877867391629e-07,
+      "loss": 0.7031,
+      "step": 5142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.217979914045849e-07,
+      "loss": 0.724,
+      "step": 5143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.195108994501434e-07,
+      "loss": 0.7194,
+      "step": 5144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.172265115569046e-07,
+      "loss": 0.6631,
+      "step": 5145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14944828405132e-07,
+      "loss": 0.6806,
+      "step": 5146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.126658506742836e-07,
+      "loss": 0.7386,
+      "step": 5147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.103895790430151e-07,
+      "loss": 0.7336,
+      "step": 5148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081160141891699e-07,
+      "loss": 0.7529,
+      "step": 5149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.058451567897875e-07,
+      "loss": 0.713,
+      "step": 5150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035770075211048e-07,
+      "loss": 0.7236,
+      "step": 5151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.01311567058547e-07,
+      "loss": 0.6284,
+      "step": 5152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.99048836076738e-07,
+      "loss": 0.7765,
+      "step": 5153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.967888152494897e-07,
+      "loss": 0.8406,
+      "step": 5154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945315052498105e-07,
+      "loss": 0.7579,
+      "step": 5155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.922769067498993e-07,
+      "loss": 0.758,
+      "step": 5156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900250204211513e-07,
+      "loss": 0.7575,
+      "step": 5157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.877758469341491e-07,
+      "loss": 0.7984,
+      "step": 5158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.855293869586679e-07,
+      "loss": 0.7044,
+      "step": 5159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.832856411636793e-07,
+      "loss": 0.7189,
+      "step": 5160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.810446102173443e-07,
+      "loss": 0.7439,
+      "step": 5161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.788062947870157e-07,
+      "loss": 0.7008,
+      "step": 5162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765706955392328e-07,
+      "loss": 0.7308,
+      "step": 5163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.74337813139734e-07,
+      "loss": 0.6808,
+      "step": 5164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.721076482534452e-07,
+      "loss": 0.728,
+      "step": 5165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.698802015444829e-07,
+      "loss": 0.7709,
+      "step": 5166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.676554736761511e-07,
+      "loss": 0.7485,
+      "step": 5167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.654334653109498e-07,
+      "loss": 0.7357,
+      "step": 5168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.632141771105674e-07,
+      "loss": 0.7664,
+      "step": 5169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.609976097358797e-07,
+      "loss": 0.715,
+      "step": 5170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.587837638469531e-07,
+      "loss": 0.7454,
+      "step": 5171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56572640103045e-07,
+      "loss": 0.78,
+      "step": 5172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.54364239162605e-07,
+      "loss": 0.2704,
+      "step": 5173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52158561683265e-07,
+      "loss": 0.7589,
+      "step": 5174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.499556083218485e-07,
+      "loss": 0.7468,
+      "step": 5175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.477553797343729e-07,
+      "loss": 0.7304,
+      "step": 5176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.455578765760364e-07,
+      "loss": 0.7144,
+      "step": 5177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.433630995012287e-07,
+      "loss": 0.7595,
+      "step": 5178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.411710491635327e-07,
+      "loss": 0.7615,
+      "step": 5179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.389817262157118e-07,
+      "loss": 0.7774,
+      "step": 5180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367951313097189e-07,
+      "loss": 0.7732,
+      "step": 5181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346112650966975e-07,
+      "loss": 0.7604,
+      "step": 5182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.324301282269797e-07,
+      "loss": 0.7677,
+      "step": 5183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302517213500761e-07,
+      "loss": 0.7293,
+      "step": 5184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280760451146918e-07,
+      "loss": 0.7455,
+      "step": 5185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259031001687201e-07,
+      "loss": 0.6595,
+      "step": 5186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.237328871592376e-07,
+      "loss": 0.7454,
+      "step": 5187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.215654067325074e-07,
+      "loss": 0.8091,
+      "step": 5188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194006595339766e-07,
+      "loss": 0.7273,
+      "step": 5189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172386462082849e-07,
+      "loss": 0.6567,
+      "step": 5190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.150793673992497e-07,
+      "loss": 0.6595,
+      "step": 5191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129228237498843e-07,
+      "loss": 0.7002,
+      "step": 5192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107690159023751e-07,
+      "loss": 0.7445,
+      "step": 5193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.086179444981057e-07,
+      "loss": 0.7723,
+      "step": 5194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06469610177636e-07,
+      "loss": 0.6746,
+      "step": 5195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043240135807174e-07,
+      "loss": 0.6425,
+      "step": 5196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02181155346281e-07,
+      "loss": 0.6615,
+      "step": 5197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.000410361124434e-07,
+      "loss": 0.7447,
+      "step": 5198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.979036565165077e-07,
+      "loss": 0.7005,
+      "step": 5199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.957690171949605e-07,
+      "loss": 0.7723,
+      "step": 5200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936371187834724e-07,
+      "loss": 0.7311,
+      "step": 5201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.915079619168953e-07,
+      "loss": 0.7014,
+      "step": 5202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.893815472292665e-07,
+      "loss": 0.7289,
+      "step": 5203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.872578753538107e-07,
+      "loss": 0.7161,
+      "step": 5204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.851369469229287e-07,
+      "loss": 0.7263,
+      "step": 5205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.830187625682073e-07,
+      "loss": 0.7248,
+      "step": 5206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.809033229204178e-07,
+      "loss": 0.7201,
+      "step": 5207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.787906286095137e-07,
+      "loss": 0.6684,
+      "step": 5208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766806802646298e-07,
+      "loss": 0.7511,
+      "step": 5209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.745734785140802e-07,
+      "loss": 0.6826,
+      "step": 5210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.724690239853682e-07,
+      "loss": 0.7323,
+      "step": 5211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.703673173051752e-07,
+      "loss": 0.7316,
+      "step": 5212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.682683590993633e-07,
+      "loss": 0.7615,
+      "step": 5213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.661721499929753e-07,
+      "loss": 0.745,
+      "step": 5214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.640786906102404e-07,
+      "loss": 0.7689,
+      "step": 5215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.619879815745623e-07,
+      "loss": 0.7627,
+      "step": 5216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599000235085341e-07,
+      "loss": 0.8093,
+      "step": 5217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.578148170339205e-07,
+      "loss": 0.7724,
+      "step": 5218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.557323627716728e-07,
+      "loss": 0.7421,
+      "step": 5219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536526613419204e-07,
+      "loss": 0.7132,
+      "step": 5220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.515757133639756e-07,
+      "loss": 0.7447,
+      "step": 5221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.49501519456326e-07,
+      "loss": 0.8479,
+      "step": 5222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.474300802366418e-07,
+      "loss": 0.7757,
+      "step": 5223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453613963217743e-07,
+      "loss": 0.7508,
+      "step": 5224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.432954683277516e-07,
+      "loss": 0.6988,
+      "step": 5225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.412322968697872e-07,
+      "loss": 0.7592,
+      "step": 5226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.391718825622607e-07,
+      "loss": 0.7632,
+      "step": 5227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371142260187436e-07,
+      "loss": 0.8288,
+      "step": 5228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.350593278519824e-07,
+      "loss": 0.6991,
+      "step": 5229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.330071886739009e-07,
+      "loss": 0.7421,
+      "step": 5230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.309578090955982e-07,
+      "loss": 0.7446,
+      "step": 5231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289111897273571e-07,
+      "loss": 0.7578,
+      "step": 5232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.268673311786378e-07,
+      "loss": 0.7878,
+      "step": 5233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.248262340580747e-07,
+      "loss": 0.8137,
+      "step": 5234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227878989734849e-07,
+      "loss": 0.756,
+      "step": 5235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.207523265318594e-07,
+      "loss": 0.738,
+      "step": 5236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.187195173393647e-07,
+      "loss": 0.6956,
+      "step": 5237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.166894720013485e-07,
+      "loss": 0.7115,
+      "step": 5238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146621911223361e-07,
+      "loss": 0.6487,
+      "step": 5239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.126376753060261e-07,
+      "loss": 0.7948,
+      "step": 5240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106159251552935e-07,
+      "loss": 0.7257,
+      "step": 5241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.085969412721927e-07,
+      "loss": 0.7013,
+      "step": 5242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065807242579548e-07,
+      "loss": 0.7548,
+      "step": 5243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.045672747129839e-07,
+      "loss": 0.8025,
+      "step": 5244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025565932368595e-07,
+      "loss": 0.721,
+      "step": 5245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.00548680428339e-07,
+      "loss": 0.769,
+      "step": 5246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985435368853577e-07,
+      "loss": 0.7643,
+      "step": 5247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965411632050212e-07,
+      "loss": 0.7785,
+      "step": 5248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.945415599836114e-07,
+      "loss": 0.7223,
+      "step": 5249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925447278165864e-07,
+      "loss": 0.2339,
+      "step": 5250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905506672985829e-07,
+      "loss": 0.2586,
+      "step": 5251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.885593790234057e-07,
+      "loss": 0.7694,
+      "step": 5252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.86570863584034e-07,
+      "loss": 0.24,
+      "step": 5253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.845851215726285e-07,
+      "loss": 0.7157,
+      "step": 5254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.826021535805161e-07,
+      "loss": 0.7358,
+      "step": 5255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.806219601982034e-07,
+      "loss": 0.6367,
+      "step": 5256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.786445420153653e-07,
+      "loss": 0.8074,
+      "step": 5257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.766698996208576e-07,
+      "loss": 0.7046,
+      "step": 5258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.746980336027009e-07,
+      "loss": 0.7488,
+      "step": 5259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727289445480967e-07,
+      "loss": 0.7611,
+      "step": 5260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.707626330434148e-07,
+      "loss": 0.8085,
+      "step": 5261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.687990996741977e-07,
+      "loss": 0.7064,
+      "step": 5262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.668383450251636e-07,
+      "loss": 0.7324,
+      "step": 5263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.648803696802042e-07,
+      "loss": 0.7707,
+      "step": 5264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629251742223785e-07,
+      "loss": 0.8213,
+      "step": 5265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.609727592339199e-07,
+      "loss": 0.7037,
+      "step": 5266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590231252962342e-07,
+      "loss": 0.738,
+      "step": 5267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.570762729899028e-07,
+      "loss": 0.8005,
+      "step": 5268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.551322028946716e-07,
+      "loss": 0.7788,
+      "step": 5269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5319091558946e-07,
+      "loss": 0.7515,
+      "step": 5270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512524116523633e-07,
+      "loss": 0.7358,
+      "step": 5271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.493166916606452e-07,
+      "loss": 0.7104,
+      "step": 5272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.473837561907381e-07,
+      "loss": 0.7619,
+      "step": 5273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.45453605818247e-07,
+      "loss": 0.7696,
+      "step": 5274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435262411179499e-07,
+      "loss": 0.7281,
+      "step": 5275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416016626637888e-07,
+      "loss": 0.8091,
+      "step": 5276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.396798710288854e-07,
+      "loss": 0.7823,
+      "step": 5277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.377608667855217e-07,
+      "loss": 0.7891,
+      "step": 5278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.358446505051574e-07,
+      "loss": 0.6933,
+      "step": 5279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339312227584172e-07,
+      "loss": 0.7693,
+      "step": 5280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320205841150962e-07,
+      "loss": 0.7445,
+      "step": 5281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.301127351441649e-07,
+      "loss": 0.7378,
+      "step": 5282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282076764137512e-07,
+      "loss": 0.7278,
+      "step": 5283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.263054084911612e-07,
+      "loss": 0.7338,
+      "step": 5284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.244059319428686e-07,
+      "loss": 0.7611,
+      "step": 5285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225092473345162e-07,
+      "loss": 0.7151,
+      "step": 5286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206153552309102e-07,
+      "loss": 0.246,
+      "step": 5287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.187242561960305e-07,
+      "loss": 0.7425,
+      "step": 5288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.168359507930244e-07,
+      "loss": 0.6943,
+      "step": 5289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149504395842087e-07,
+      "loss": 0.7248,
+      "step": 5290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.130677231310633e-07,
+      "loss": 0.7195,
+      "step": 5291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111878019942386e-07,
+      "loss": 0.7203,
+      "step": 5292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093106767335544e-07,
+      "loss": 0.754,
+      "step": 5293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.074363479079936e-07,
+      "loss": 0.7412,
+      "step": 5294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055648160757122e-07,
+      "loss": 0.7711,
+      "step": 5295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.036960817940263e-07,
+      "loss": 0.7464,
+      "step": 5296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.018301456194276e-07,
+      "loss": 0.755,
+      "step": 5297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999670081075637e-07,
+      "loss": 0.7382,
+      "step": 5298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.981066698132609e-07,
+      "loss": 0.7283,
+      "step": 5299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962491312905016e-07,
+      "loss": 0.7085,
+      "step": 5300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.943943930924378e-07,
+      "loss": 0.7143,
+      "step": 5301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.92542455771391e-07,
+      "loss": 0.7618,
+      "step": 5302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906933198788456e-07,
+      "loss": 0.8108,
+      "step": 5303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888469859654533e-07,
+      "loss": 0.7774,
+      "step": 5304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.870034545810255e-07,
+      "loss": 0.6631,
+      "step": 5305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.851627262745485e-07,
+      "loss": 0.7238,
+      "step": 5306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833248015941684e-07,
+      "loss": 0.7553,
+      "step": 5307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814896810871973e-07,
+      "loss": 0.741,
+      "step": 5308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.796573653001091e-07,
+      "loss": 0.6198,
+      "step": 5309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.778278547785488e-07,
+      "loss": 0.7702,
+      "step": 5310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.760011500673224e-07,
+      "loss": 0.7384,
+      "step": 5311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.741772517104005e-07,
+      "loss": 0.7912,
+      "step": 5312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.723561602509165e-07,
+      "loss": 0.6785,
+      "step": 5313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.70537876231173e-07,
+      "loss": 0.8052,
+      "step": 5314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.687224001926305e-07,
+      "loss": 0.7961,
+      "step": 5315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.669097326759176e-07,
+      "loss": 0.7637,
+      "step": 5316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.650998742208236e-07,
+      "loss": 0.8063,
+      "step": 5317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63292825366305e-07,
+      "loss": 0.7468,
+      "step": 5318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.61488586650476e-07,
+      "loss": 0.7054,
+      "step": 5319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.596871586106212e-07,
+      "loss": 0.7641,
+      "step": 5320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.578885417831814e-07,
+      "loss": 0.7519,
+      "step": 5321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.560927367037639e-07,
+      "loss": 0.7611,
+      "step": 5322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542997439071385e-07,
+      "loss": 0.8045,
+      "step": 5323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.525095639272383e-07,
+      "loss": 0.7115,
+      "step": 5324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.507221972971555e-07,
+      "loss": 0.7627,
+      "step": 5325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489376445491467e-07,
+      "loss": 0.6727,
+      "step": 5326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471559062146304e-07,
+      "loss": 0.6934,
+      "step": 5327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.453769828241872e-07,
+      "loss": 0.7189,
+      "step": 5328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.436008749075617e-07,
+      "loss": 0.7974,
+      "step": 5329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.6598,
+      "step": 5330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.400571076105299e-07,
+      "loss": 0.8263,
+      "step": 5331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.382894492854185e-07,
+      "loss": 0.6202,
+      "step": 5332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365246085447052e-07,
+      "loss": 0.6732,
+      "step": 5333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347625859139372e-07,
+      "loss": 0.7682,
+      "step": 5334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.330033819178249e-07,
+      "loss": 0.7071,
+      "step": 5335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.312469970802403e-07,
+      "loss": 0.8573,
+      "step": 5336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.294934319242106e-07,
+      "loss": 0.2449,
+      "step": 5337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277426869719304e-07,
+      "loss": 0.7344,
+      "step": 5338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259947627447482e-07,
+      "loss": 0.7488,
+      "step": 5339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24249659763173e-07,
+      "loss": 0.7923,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.225073785468793e-07,
+      "loss": 0.7326,
+      "step": 5341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207679196146964e-07,
+      "loss": 0.6772,
+      "step": 5342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.190312834846134e-07,
+      "loss": 0.7687,
+      "step": 5343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.172974706737799e-07,
+      "loss": 0.6809,
+      "step": 5344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.155664816985062e-07,
+      "loss": 0.8078,
+      "step": 5345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138383170742589e-07,
+      "loss": 0.7494,
+      "step": 5346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121129773156663e-07,
+      "loss": 0.7045,
+      "step": 5347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.103904629365108e-07,
+      "loss": 0.7824,
+      "step": 5348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086707744497388e-07,
+      "loss": 0.7437,
+      "step": 5349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.069539123674539e-07,
+      "loss": 0.7251,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.05239877200916e-07,
+      "loss": 0.8209,
+      "step": 5351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035286694605435e-07,
+      "loss": 0.6324,
+      "step": 5352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018202896559154e-07,
+      "loss": 0.7471,
+      "step": 5353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001147382957639e-07,
+      "loss": 0.7343,
+      "step": 5354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984120158879858e-07,
+      "loss": 0.2758,
+      "step": 5355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967121229396277e-07,
+      "loss": 0.7455,
+      "step": 5356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950150599569015e-07,
+      "loss": 0.2574,
+      "step": 5357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933208274451673e-07,
+      "loss": 0.6829,
+      "step": 5358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916294259089516e-07,
+      "loss": 0.6864,
+      "step": 5359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899408558519325e-07,
+      "loss": 0.7509,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.882551177769435e-07,
+      "loss": 0.8071,
+      "step": 5361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.865722121859794e-07,
+      "loss": 0.7423,
+      "step": 5362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848921395801909e-07,
+      "loss": 0.7841,
+      "step": 5363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.832149004598807e-07,
+      "loss": 0.7933,
+      "step": 5364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.815404953245118e-07,
+      "loss": 0.7868,
+      "step": 5365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798689246727006e-07,
+      "loss": 0.6922,
+      "step": 5366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.782001890022247e-07,
+      "loss": 0.6921,
+      "step": 5367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.765342888100111e-07,
+      "loss": 0.746,
+      "step": 5368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7487122459214274e-07,
+      "loss": 0.6308,
+      "step": 5369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.732109968438636e-07,
+      "loss": 0.7822,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7155360605956935e-07,
+      "loss": 0.7498,
+      "step": 5371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6989905273281065e-07,
+      "loss": 0.7129,
+      "step": 5372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6824733735629234e-07,
+      "loss": 0.2479,
+      "step": 5373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.665984604218765e-07,
+      "loss": 0.7417,
+      "step": 5374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.649524224205815e-07,
+      "loss": 0.8018,
+      "step": 5375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6330922384257514e-07,
+      "loss": 0.764,
+      "step": 5376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616688651771806e-07,
+      "loss": 0.736,
+      "step": 5377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6003134691288143e-07,
+      "loss": 0.7449,
+      "step": 5378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5839666953730764e-07,
+      "loss": 0.6549,
+      "step": 5379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.567648335372488e-07,
+      "loss": 0.7576,
+      "step": 5380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551358393986438e-07,
+      "loss": 0.7616,
+      "step": 5381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5350968760659033e-07,
+      "loss": 0.6481,
+      "step": 5382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5188637864533423e-07,
+      "loss": 0.6687,
+      "step": 5383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5026591299828007e-07,
+      "loss": 0.6123,
+      "step": 5384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4864829114798394e-07,
+      "loss": 0.7768,
+      "step": 5385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.470335135761494e-07,
+      "loss": 0.689,
+      "step": 5386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4542158076364085e-07,
+      "loss": 0.6449,
+      "step": 5387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4381249319047303e-07,
+      "loss": 0.7846,
+      "step": 5388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4220625133581605e-07,
+      "loss": 0.8159,
+      "step": 5389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406028556779829e-07,
+      "loss": 0.7882,
+      "step": 5390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3900230669444933e-07,
+      "loss": 0.7415,
+      "step": 5391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.374046048618397e-07,
+      "loss": 0.6721,
+      "step": 5392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3580975065593e-07,
+      "loss": 0.7587,
+      "step": 5393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3421774455165043e-07,
+      "loss": 0.7118,
+      "step": 5394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.326285870230773e-07,
+      "loss": 0.7127,
+      "step": 5395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.310422785434476e-07,
+      "loss": 0.6621,
+      "step": 5396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2945881958514125e-07,
+      "loss": 0.7442,
+      "step": 5397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2787821061969657e-07,
+      "loss": 0.7515,
+      "step": 5398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2630045211779825e-07,
+      "loss": 0.6506,
+      "step": 5399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.247255445492826e-07,
+      "loss": 0.7501,
+      "step": 5400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.231534883831401e-07,
+      "loss": 0.6829,
+      "step": 5401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.215842840875106e-07,
+      "loss": 0.7449,
+      "step": 5402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.200179321296849e-07,
+      "loss": 0.7516,
+      "step": 5403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184544329761009e-07,
+      "loss": 0.2554,
+      "step": 5404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1689378709235173e-07,
+      "loss": 0.6815,
+      "step": 5405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.153359949431801e-07,
+      "loss": 0.7553,
+      "step": 5406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1378105699247825e-07,
+      "loss": 0.7862,
+      "step": 5407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1222897370328475e-07,
+      "loss": 0.7959,
+      "step": 5408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1067974553779424e-07,
+      "loss": 0.6801,
+      "step": 5409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.091333729573488e-07,
+      "loss": 0.7493,
+      "step": 5410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0758985642243896e-07,
+      "loss": 0.8033,
+      "step": 5411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060491963927027e-07,
+      "loss": 0.8286,
+      "step": 5412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.04511393326934e-07,
+      "loss": 0.819,
+      "step": 5413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029764476830722e-07,
+      "loss": 0.7197,
+      "step": 5414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0144435991820384e-07,
+      "loss": 0.7463,
+      "step": 5415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9991513048856623e-07,
+      "loss": 0.7436,
+      "step": 5416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9838875984954727e-07,
+      "loss": 0.6976,
+      "step": 5417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9686524845568117e-07,
+      "loss": 0.7887,
+      "step": 5418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.953445967606517e-07,
+      "loss": 0.7244,
+      "step": 5419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9382680521728886e-07,
+      "loss": 0.7889,
+      "step": 5420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.923118742775756e-07,
+      "loss": 0.7429,
+      "step": 5421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9079980439263864e-07,
+      "loss": 0.6745,
+      "step": 5422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8929059601275463e-07,
+      "loss": 0.6924,
+      "step": 5423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.877842495873496e-07,
+      "loss": 0.6854,
+      "step": 5424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8628076556499137e-07,
+      "loss": 0.6818,
+      "step": 5425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8478014439340294e-07,
+      "loss": 0.6973,
+      "step": 5426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8328238651945127e-07,
+      "loss": 0.7561,
+      "step": 5427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.817874923891496e-07,
+      "loss": 0.7846,
+      "step": 5428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.802954624476585e-07,
+      "loss": 0.7261,
+      "step": 5429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7880629713928917e-07,
+      "loss": 0.7129,
+      "step": 5430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.773199969074959e-07,
+      "loss": 0.8143,
+      "step": 5431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.758365621948823e-07,
+      "loss": 0.789,
+      "step": 5432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7435599344319617e-07,
+      "loss": 0.8474,
+      "step": 5433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7287829109333264e-07,
+      "loss": 0.7278,
+      "step": 5434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7140345558533517e-07,
+      "loss": 0.7239,
+      "step": 5435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.699314873583937e-07,
+      "loss": 0.7106,
+      "step": 5436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.684623868508386e-07,
+      "loss": 0.7099,
+      "step": 5437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.669961545001532e-07,
+      "loss": 0.8075,
+      "step": 5438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.65532790742964e-07,
+      "loss": 0.6417,
+      "step": 5439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.640722960150411e-07,
+      "loss": 0.733,
+      "step": 5440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.626146707513067e-07,
+      "loss": 0.7295,
+      "step": 5441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.611599153858214e-07,
+      "loss": 0.7797,
+      "step": 5442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5970803035179193e-07,
+      "loss": 0.7431,
+      "step": 5443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5825901608157574e-07,
+      "loss": 0.7574,
+      "step": 5444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.568128730066711e-07,
+      "loss": 0.7396,
+      "step": 5445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5536960155772124e-07,
+      "loss": 0.2608,
+      "step": 5446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.539292021645158e-07,
+      "loss": 0.7844,
+      "step": 5447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5249167525598817e-07,
+      "loss": 0.6976,
+      "step": 5448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.510570212602171e-07,
+      "loss": 0.7413,
+      "step": 5449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.496252406044265e-07,
+      "loss": 0.8248,
+      "step": 5450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481963337149807e-07,
+      "loss": 0.6566,
+      "step": 5451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467703010173928e-07,
+      "loss": 0.7107,
+      "step": 5452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4534714293631976e-07,
+      "loss": 0.6259,
+      "step": 5453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4392685989555806e-07,
+      "loss": 0.7212,
+      "step": 5454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4250945231805276e-07,
+      "loss": 0.7171,
+      "step": 5455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4109492062589175e-07,
+      "loss": 0.2488,
+      "step": 5456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.396832652403026e-07,
+      "loss": 0.6837,
+      "step": 5457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.382744865816634e-07,
+      "loss": 0.7337,
+      "step": 5458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.368685850694875e-07,
+      "loss": 0.7657,
+      "step": 5459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3546556112244e-07,
+      "loss": 0.6599,
+      "step": 5460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3406541515832e-07,
+      "loss": 0.7296,
+      "step": 5461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.326681475940785e-07,
+      "loss": 0.7177,
+      "step": 5462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.312737588458037e-07,
+      "loss": 0.819,
+      "step": 5463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.298822493287257e-07,
+      "loss": 0.7546,
+      "step": 5464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.284936194572208e-07,
+      "loss": 0.7908,
+      "step": 5465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2710786964480825e-07,
+      "loss": 0.7548,
+      "step": 5466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2572500030414567e-07,
+      "loss": 0.7,
+      "step": 5467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.243450118470348e-07,
+      "loss": 0.2346,
+      "step": 5468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.229679046844203e-07,
+      "loss": 0.7183,
+      "step": 5469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2159367922639075e-07,
+      "loss": 0.7119,
+      "step": 5470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20222335882171e-07,
+      "loss": 0.7278,
+      "step": 5471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1885387506012997e-07,
+      "loss": 0.7121,
+      "step": 5472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1748829716778175e-07,
+      "loss": 0.674,
+      "step": 5473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161256026117776e-07,
+      "loss": 0.7561,
+      "step": 5474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.14765791797913e-07,
+      "loss": 0.7237,
+      "step": 5475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1340886513112066e-07,
+      "loss": 0.7945,
+      "step": 5476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1205482301547965e-07,
+      "loss": 0.7517,
+      "step": 5477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.107036658542062e-07,
+      "loss": 0.6902,
+      "step": 5478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0935539404966073e-07,
+      "loss": 0.2646,
+      "step": 5479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.080100080033388e-07,
+      "loss": 0.7419,
+      "step": 5480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.066675081158843e-07,
+      "loss": 0.6813,
+      "step": 5481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053278947870753e-07,
+      "loss": 0.7278,
+      "step": 5482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0399116841583363e-07,
+      "loss": 0.7248,
+      "step": 5483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0265732940021887e-07,
+      "loss": 0.2393,
+      "step": 5484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013263781374354e-07,
+      "loss": 0.6997,
+      "step": 5485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.999983150238206e-07,
+      "loss": 0.814,
+      "step": 5486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.986731404548582e-07,
+      "loss": 0.763,
+      "step": 5487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9735085482517243e-07,
+      "loss": 0.678,
+      "step": 5488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9603145852851736e-07,
+      "loss": 0.6362,
+      "step": 5489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.947149519577974e-07,
+      "loss": 0.6052,
+      "step": 5490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9340133550505244e-07,
+      "loss": 0.7656,
+      "step": 5491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9209060956146283e-07,
+      "loss": 0.6508,
+      "step": 5492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9078277451734307e-07,
+      "loss": 0.6773,
+      "step": 5493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.894778307621537e-07,
+      "loss": 0.7537,
+      "step": 5494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.88175778684493e-07,
+      "loss": 0.7326,
+      "step": 5495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.868766186720928e-07,
+      "loss": 0.7448,
+      "step": 5496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.855803511118305e-07,
+      "loss": 0.6925,
+      "step": 5497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8428697638971605e-07,
+      "loss": 0.698,
+      "step": 5498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.829964948909048e-07,
+      "loss": 0.6901,
+      "step": 5499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.817089069996837e-07,
+      "loss": 0.7436,
+      "step": 5500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8042421309948164e-07,
+      "loss": 0.6973,
+      "step": 5501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.791424135728671e-07,
+      "loss": 0.7945,
+      "step": 5502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7786350880154134e-07,
+      "loss": 0.7447,
+      "step": 5503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7658749916634864e-07,
+      "loss": 0.7314,
+      "step": 5504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.753143850472706e-07,
+      "loss": 0.8127,
+      "step": 5505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.740441668234239e-07,
+      "loss": 0.6938,
+      "step": 5506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727768448730639e-07,
+      "loss": 0.6976,
+      "step": 5507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715124195735852e-07,
+      "loss": 0.7373,
+      "step": 5508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.702508913015178e-07,
+      "loss": 0.7023,
+      "step": 5509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6899226043253125e-07,
+      "loss": 0.7461,
+      "step": 5510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6773652734142896e-07,
+      "loss": 0.7166,
+      "step": 5511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66483692402153e-07,
+      "loss": 0.7723,
+      "step": 5512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.652337559877849e-07,
+      "loss": 0.7509,
+      "step": 5513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639867184705391e-07,
+      "loss": 0.7392,
+      "step": 5514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627425802217676e-07,
+      "loss": 0.2832,
+      "step": 5515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.615013416119627e-07,
+      "loss": 0.7807,
+      "step": 5516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602630030107478e-07,
+      "loss": 0.7793,
+      "step": 5517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5902756478688674e-07,
+      "loss": 0.7515,
+      "step": 5518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5779502730827743e-07,
+      "loss": 0.7473,
+      "step": 5519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.565653909419563e-07,
+      "loss": 0.7829,
+      "step": 5520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.553386560540927e-07,
+      "loss": 0.7929,
+      "step": 5521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.541148230099955e-07,
+      "loss": 0.7213,
+      "step": 5522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.528938921741042e-07,
+      "loss": 0.7924,
+      "step": 5523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5167586391000145e-07,
+      "loss": 0.6966,
+      "step": 5524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.504607385803992e-07,
+      "loss": 0.7629,
+      "step": 5525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4924851654714697e-07,
+      "loss": 0.7338,
+      "step": 5526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480391981712327e-07,
+      "loss": 0.6922,
+      "step": 5527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4683278381277267e-07,
+      "loss": 0.243,
+      "step": 5528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.456292738310251e-07,
+      "loss": 0.6371,
+      "step": 5529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.444286685843822e-07,
+      "loss": 0.7026,
+      "step": 5530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.432309684303691e-07,
+      "loss": 0.6764,
+      "step": 5531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.420361737256438e-07,
+      "loss": 0.6859,
+      "step": 5532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.408442848260051e-07,
+      "loss": 0.7625,
+      "step": 5533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3965530208638235e-07,
+      "loss": 0.7594,
+      "step": 5534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.384692258608412e-07,
+      "loss": 0.6938,
+      "step": 5535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3728605650257807e-07,
+      "loss": 0.7069,
+      "step": 5536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3610579436392999e-07,
+      "loss": 0.7269,
+      "step": 5537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.349284397963636e-07,
+      "loss": 0.7719,
+      "step": 5538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3375399315048064e-07,
+      "loss": 0.7408,
+      "step": 5539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3258245477601804e-07,
+      "loss": 0.7313,
+      "step": 5540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3141382502184563e-07,
+      "loss": 0.7102,
+      "step": 5541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3024810423596722e-07,
+      "loss": 0.7225,
+      "step": 5542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290852927655196e-07,
+      "loss": 0.7811,
+      "step": 5543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.279253909567769e-07,
+      "loss": 0.6899,
+      "step": 5544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.267683991551417e-07,
+      "loss": 0.7543,
+      "step": 5545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2561431770515285e-07,
+      "loss": 0.7414,
+      "step": 5546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244631469504821e-07,
+      "loss": 0.7797,
+      "step": 5547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2331488723393634e-07,
+      "loss": 0.7064,
+      "step": 5548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2216953889744875e-07,
+      "loss": 0.7542,
+      "step": 5549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2102710228209422e-07,
+      "loss": 0.7189,
+      "step": 5550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1988757772807624e-07,
+      "loss": 0.6253,
+      "step": 5551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1875096557473108e-07,
+      "loss": 0.2613,
+      "step": 5552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1761726616052915e-07,
+      "loss": 0.7312,
+      "step": 5553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1648647982307037e-07,
+      "loss": 0.7147,
+      "step": 5554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1535860689909206e-07,
+      "loss": 0.7392,
+      "step": 5555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1423364772445886e-07,
+      "loss": 0.7202,
+      "step": 5556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1311160263417285e-07,
+      "loss": 0.7338,
+      "step": 5557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.119924719623634e-07,
+      "loss": 0.7493,
+      "step": 5558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.108762560422961e-07,
+      "loss": 0.798,
+      "step": 5559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0976295520636403e-07,
+      "loss": 0.7688,
+      "step": 5560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0865256978609748e-07,
+      "loss": 0.6815,
+      "step": 5561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0754510011215422e-07,
+      "loss": 0.6964,
+      "step": 5562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.064405465143282e-07,
+      "loss": 0.808,
+      "step": 5563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0533890932153855e-07,
+      "loss": 0.7519,
+      "step": 5564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0424018886184282e-07,
+      "loss": 0.784,
+      "step": 5565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0314438546242488e-07,
+      "loss": 0.7704,
+      "step": 5566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0205149944960257e-07,
+      "loss": 0.7133,
+      "step": 5567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0096153114882445e-07,
+      "loss": 0.7783,
+      "step": 5568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9987448088466977e-07,
+      "loss": 0.761,
+      "step": 5569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9879034898084958e-07,
+      "loss": 0.7052,
+      "step": 5570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9770913576020455e-07,
+      "loss": 0.7333,
+      "step": 5571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9663084154470713e-07,
+      "loss": 0.742,
+      "step": 5572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9555546665546155e-07,
+      "loss": 0.2877,
+      "step": 5573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9448301141270166e-07,
+      "loss": 0.7471,
+      "step": 5574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9341347613579086e-07,
+      "loss": 0.6616,
+      "step": 5575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.923468611432233e-07,
+      "loss": 0.7403,
+      "step": 5576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9128316675262604e-07,
+      "loss": 0.7625,
+      "step": 5577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9022239328075343e-07,
+      "loss": 0.6334,
+      "step": 5578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8916454104349058e-07,
+      "loss": 0.2608,
+      "step": 5579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.881096103558555e-07,
+      "loss": 0.6829,
+      "step": 5580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705760153199137e-07,
+      "loss": 0.6699,
+      "step": 5581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8600851488517647e-07,
+      "loss": 0.7787,
+      "step": 5582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.849623507278131e-07,
+      "loss": 0.7453,
+      "step": 5583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8391910937143876e-07,
+      "loss": 0.7081,
+      "step": 5584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8287879112671712e-07,
+      "loss": 0.7188,
+      "step": 5585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8184139630344378e-07,
+      "loss": 0.842,
+      "step": 5586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8080692521054267e-07,
+      "loss": 0.6994,
+      "step": 5587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7977537815606405e-07,
+      "loss": 0.8157,
+      "step": 5588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7874675544719334e-07,
+      "loss": 0.7182,
+      "step": 5589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7772105739024102e-07,
+      "loss": 0.7282,
+      "step": 5590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7669828429064838e-07,
+      "loss": 0.6921,
+      "step": 5591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.75678436452984e-07,
+      "loss": 0.7148,
+      "step": 5592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.74661514180946e-07,
+      "loss": 0.7862,
+      "step": 5593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7364751777736334e-07,
+      "loss": 0.7261,
+      "step": 5594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7263644754419108e-07,
+      "loss": 0.7536,
+      "step": 5595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7162830378251394e-07,
+      "loss": 0.6525,
+      "step": 5596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7062308679254514e-07,
+      "loss": 0.7997,
+      "step": 5597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6962079687362632e-07,
+      "loss": 0.75,
+      "step": 5598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6862143432422761e-07,
+      "loss": 0.7533,
+      "step": 5599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6762499944194655e-07,
+      "loss": 0.8116,
+      "step": 5600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6663149252351019e-07,
+      "loss": 0.8223,
+      "step": 5601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.65640913864773e-07,
+      "loss": 0.6936,
+      "step": 5602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.646532637607179e-07,
+      "loss": 0.635,
+      "step": 5603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6366854250545406e-07,
+      "loss": 0.7702,
+      "step": 5604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6268675039222136e-07,
+      "loss": 0.6822,
+      "step": 5605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6170788771338374e-07,
+      "loss": 0.74,
+      "step": 5606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607319547604358e-07,
+      "loss": 0.8238,
+      "step": 5607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5975895182399948e-07,
+      "loss": 0.7815,
+      "step": 5608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5878887919382302e-07,
+      "loss": 0.7567,
+      "step": 5609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.578217371587809e-07,
+      "loss": 0.7467,
+      "step": 5610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5685752600687832e-07,
+      "loss": 0.6992,
+      "step": 5611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5589624602524555e-07,
+      "loss": 0.6566,
+      "step": 5612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5493789750014032e-07,
+      "loss": 0.7392,
+      "step": 5613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5398248071694656e-07,
+      "loss": 0.7597,
+      "step": 5614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5302999596017664e-07,
+      "loss": 0.7485,
+      "step": 5615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.520804435134693e-07,
+      "loss": 0.687,
+      "step": 5616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.511338236595894e-07,
+      "loss": 0.6878,
+      "step": 5617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5019013668042926e-07,
+      "loss": 0.7749,
+      "step": 5618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.492493828570085e-07,
+      "loss": 0.6795,
+      "step": 5619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4831156246946977e-07,
+      "loss": 0.7954,
+      "step": 5620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4737667579708735e-07,
+      "loss": 0.7721,
+      "step": 5621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4644472311825862e-07,
+      "loss": 0.7451,
+      "step": 5622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4551570471050712e-07,
+      "loss": 0.789,
+      "step": 5623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4458962085048378e-07,
+      "loss": 0.7527,
+      "step": 5624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4366647181396687e-07,
+      "loss": 0.6356,
+      "step": 5625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4274625787585762e-07,
+      "loss": 0.7333,
+      "step": 5626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4182897931018457e-07,
+      "loss": 0.7217,
+      "step": 5627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409146363901026e-07,
+      "loss": 0.7655,
+      "step": 5628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4000322938789278e-07,
+      "loss": 0.7526,
+      "step": 5629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3909475857496134e-07,
+      "loss": 0.7422,
+      "step": 5630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3818922422183855e-07,
+      "loss": 0.7781,
+      "step": 5631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728662659818205e-07,
+      "loss": 0.7447,
+      "step": 5632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.7169,
+      "step": 5633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.354902426135285e-07,
+      "loss": 0.7318,
+      "step": 5634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3459645678747135e-07,
+      "loss": 0.6579,
+      "step": 5635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3370560876076465e-07,
+      "loss": 0.6893,
+      "step": 5636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3281769879869178e-07,
+      "loss": 0.663,
+      "step": 5637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.319327271656623e-07,
+      "loss": 0.7239,
+      "step": 5638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3105069412520876e-07,
+      "loss": 0.7229,
+      "step": 5639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3017159993999218e-07,
+      "loss": 0.7715,
+      "step": 5640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2929544487179425e-07,
+      "loss": 0.7345,
+      "step": 5641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2842222918152625e-07,
+      "loss": 0.8053,
+      "step": 5642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2755195312921796e-07,
+      "loss": 0.7907,
+      "step": 5643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.266846169740288e-07,
+      "loss": 0.6766,
+      "step": 5644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2582022097424097e-07,
+      "loss": 0.7938,
+      "step": 5645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2495876538726192e-07,
+      "loss": 0.7607,
+      "step": 5646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2410025046962316e-07,
+      "loss": 0.71,
+      "step": 5647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2324467647697903e-07,
+      "loss": 0.7705,
+      "step": 5648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2239204366410906e-07,
+      "loss": 0.7214,
+      "step": 5649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2154235228491684e-07,
+      "loss": 0.7383,
+      "step": 5650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.206956025924333e-07,
+      "loss": 0.7033,
+      "step": 5651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1985179483880672e-07,
+      "loss": 0.6717,
+      "step": 5652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19010929275315e-07,
+      "loss": 0.8146,
+      "step": 5653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1817300615235672e-07,
+      "loss": 0.702,
+      "step": 5654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733802571945674e-07,
+      "loss": 0.7029,
+      "step": 5655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.165059882252606e-07,
+      "loss": 0.7399,
+      "step": 5656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1567689391754122e-07,
+      "loss": 0.7578,
+      "step": 5657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.148507430431911e-07,
+      "loss": 0.7271,
+      "step": 5658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.140275358482279e-07,
+      "loss": 0.6678,
+      "step": 5659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1320727257779551e-07,
+      "loss": 0.7615,
+      "step": 5660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1238995347615522e-07,
+      "loss": 0.6754,
+      "step": 5661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1157557878669789e-07,
+      "loss": 0.2782,
+      "step": 5662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1076414875193287e-07,
+      "loss": 0.805,
+      "step": 5663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0995566361349464e-07,
+      "loss": 0.778,
+      "step": 5664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0915012361214062e-07,
+      "loss": 0.7408,
+      "step": 5665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0834752898775002e-07,
+      "loss": 0.6654,
+      "step": 5666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0754787997932614e-07,
+      "loss": 0.6799,
+      "step": 5667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0675117682499625e-07,
+      "loss": 0.704,
+      "step": 5668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0595741976200946e-07,
+      "loss": 0.7912,
+      "step": 5669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0516660902673448e-07,
+      "loss": 0.7377,
+      "step": 5670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0437874485466626e-07,
+      "loss": 0.7058,
+      "step": 5671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0359382748042379e-07,
+      "loss": 0.6977,
+      "step": 5672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0281185713774344e-07,
+      "loss": 0.7529,
+      "step": 5673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0203283405948672e-07,
+      "loss": 0.7657,
+      "step": 5674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0125675847763805e-07,
+      "loss": 0.7367,
+      "step": 5675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0048363062330368e-07,
+      "loss": 0.7149,
+      "step": 5676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.971345072671168e-08,
+      "loss": 0.7845,
+      "step": 5677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.894621901721191e-08,
+      "loss": 0.7774,
+      "step": 5678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.818193572327827e-08,
+      "loss": 0.7312,
+      "step": 5679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.742060107250318e-08,
+      "loss": 0.7968,
+      "step": 5680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.666221529160525e-08,
+      "loss": 0.7323,
+      "step": 5681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.590677860642051e-08,
+      "loss": 0.7409,
+      "step": 5682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.515429124191011e-08,
+      "loss": 0.7449,
+      "step": 5683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.440475342215482e-08,
+      "loss": 0.2646,
+      "step": 5684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.365816537036054e-08,
+      "loss": 0.6949,
+      "step": 5685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.291452730884943e-08,
+      "loss": 0.6792,
+      "step": 5686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.217383945906987e-08,
+      "loss": 0.6617,
+      "step": 5687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.143610204158881e-08,
+      "loss": 0.6459,
+      "step": 5688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.070131527609604e-08,
+      "loss": 0.7586,
+      "step": 5689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.996947938140211e-08,
+      "loss": 0.723,
+      "step": 5690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.924059457543932e-08,
+      "loss": 0.8026,
+      "step": 5691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.851466107526074e-08,
+      "loss": 0.254,
+      "step": 5692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.779167909704011e-08,
+      "loss": 0.7306,
+      "step": 5693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.70716488560741e-08,
+      "loss": 0.6922,
+      "step": 5694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.635457056677788e-08,
+      "loss": 0.7368,
+      "step": 5695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.564044444268949e-08,
+      "loss": 0.7695,
+      "step": 5696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.492927069646772e-08,
+      "loss": 0.7181,
+      "step": 5697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.422104953988986e-08,
+      "loss": 0.6679,
+      "step": 5698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.351578118385717e-08,
+      "loss": 0.7375,
+      "step": 5699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.281346583838945e-08,
+      "loss": 0.7997,
+      "step": 5700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.211410371262829e-08,
+      "loss": 0.7768,
+      "step": 5701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.141769501483487e-08,
+      "loss": 0.7455,
+      "step": 5702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.072423995239221e-08,
+      "loss": 0.7197,
+      "step": 5703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.003373873180287e-08,
+      "loss": 0.7301,
+      "step": 5704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.934619155868794e-08,
+      "loss": 0.788,
+      "step": 5705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.866159863779255e-08,
+      "loss": 0.6549,
+      "step": 5706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.797996017298026e-08,
+      "loss": 0.7652,
+      "step": 5707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.730127636723539e-08,
+      "loss": 0.7675,
+      "step": 5708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.662554742265959e-08,
+      "loss": 0.71,
+      "step": 5709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.595277354047858e-08,
+      "loss": 0.7855,
+      "step": 5710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.528295492103765e-08,
+      "loss": 0.7313,
+      "step": 5711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.461609176379837e-08,
+      "loss": 0.8342,
+      "step": 5712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.395218426734519e-08,
+      "loss": 0.6896,
+      "step": 5713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.32912326293822e-08,
+      "loss": 0.7555,
+      "step": 5714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.263323704673308e-08,
+      "loss": 0.7274,
+      "step": 5715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.197819771534109e-08,
+      "loss": 0.7951,
+      "step": 5716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.132611483026797e-08,
+      "loss": 0.7821,
+      "step": 5717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.067698858569838e-08,
+      "loss": 0.7664,
+      "step": 5718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.003081917493215e-08,
+      "loss": 0.8022,
+      "step": 5719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.938760679039203e-08,
+      "loss": 0.8309,
+      "step": 5720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.874735162361924e-08,
+      "loss": 0.7685,
+      "step": 5721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.811005386527347e-08,
+      "loss": 0.654,
+      "step": 5722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.747571370513406e-08,
+      "loss": 0.7142,
+      "step": 5723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.684433133209988e-08,
+      "loss": 0.7458,
+      "step": 5724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.621590693419055e-08,
+      "loss": 0.7932,
+      "step": 5725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.559044069854192e-08,
+      "loss": 0.7206,
+      "step": 5726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.496793281141056e-08,
+      "loss": 0.6889,
+      "step": 5727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.434838345817374e-08,
+      "loss": 0.7427,
+      "step": 5728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.373179282332275e-08,
+      "loss": 0.7426,
+      "step": 5729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.311816109047298e-08,
+      "loss": 0.7997,
+      "step": 5730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.250748844235489e-08,
+      "loss": 0.691,
+      "step": 5731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.18997750608219e-08,
+      "loss": 0.7421,
+      "step": 5732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.12950211268426e-08,
+      "loss": 0.7028,
+      "step": 5733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.069322682050516e-08,
+      "loss": 0.7224,
+      "step": 5734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.009439232101732e-08,
+      "loss": 0.6594,
+      "step": 5735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.949851780670424e-08,
+      "loss": 0.721,
+      "step": 5736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8905603455010615e-08,
+      "loss": 0.7744,
+      "step": 5737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8315649442499675e-08,
+      "loss": 0.7357,
+      "step": 5738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7728655944852e-08,
+      "loss": 0.7631,
+      "step": 5739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7144623136866636e-08,
+      "loss": 0.6849,
+      "step": 5740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.656355119246337e-08,
+      "loss": 0.7019,
+      "step": 5741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5985440284677116e-08,
+      "loss": 0.6737,
+      "step": 5742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.541029058566349e-08,
+      "loss": 0.7702,
+      "step": 5743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4838102266693284e-08,
+      "loss": 0.6995,
+      "step": 5744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.426887549816018e-08,
+      "loss": 0.6761,
+      "step": 5745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.37026104495697e-08,
+      "loss": 0.7327,
+      "step": 5746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.313930728955141e-08,
+      "loss": 0.7715,
+      "step": 5747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.257896618585001e-08,
+      "loss": 0.78,
+      "step": 5748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.202158730532647e-08,
+      "loss": 0.7451,
+      "step": 5749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.146717081396468e-08,
+      "loss": 0.6555,
+      "step": 5750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0915716876860365e-08,
+      "loss": 0.737,
+      "step": 5751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.036722565823104e-08,
+      "loss": 0.7253,
+      "step": 5752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.982169732141162e-08,
+      "loss": 0.6736,
+      "step": 5753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.927913202885326e-08,
+      "loss": 0.7717,
+      "step": 5754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8739529942124496e-08,
+      "loss": 0.7093,
+      "step": 5755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.820289122191346e-08,
+      "loss": 0.8056,
+      "step": 5756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7669216028024546e-08,
+      "loss": 0.7137,
+      "step": 5757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.713850451937951e-08,
+      "loss": 0.7799,
+      "step": 5758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6610756854019726e-08,
+      "loss": 0.652,
+      "step": 5759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.608597318909835e-08,
+      "loss": 0.2675,
+      "step": 5760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.556415368089373e-08,
+      "loss": 0.7247,
+      "step": 5761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5045298484794887e-08,
+      "loss": 0.7035,
+      "step": 5762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.452940775531267e-08,
+      "loss": 0.7589,
+      "step": 5763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4016481646070866e-08,
+      "loss": 0.7657,
+      "step": 5764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.350652030981395e-08,
+      "loss": 0.7424,
+      "step": 5765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2999523898402675e-08,
+      "loss": 0.7394,
+      "step": 5766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.249549256281404e-08,
+      "loss": 0.7309,
+      "step": 5767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.199442645314356e-08,
+      "loss": 0.6483,
+      "step": 5768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.149632571860074e-08,
+      "loss": 0.7744,
+      "step": 5769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.100119050751472e-08,
+      "loss": 0.7732,
+      "step": 5770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0509020967332e-08,
+      "loss": 0.7187,
+      "step": 5771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0019817244614236e-08,
+      "loss": 0.7551,
+      "step": 5772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9533579485040444e-08,
+      "loss": 0.2513,
+      "step": 5773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.905030783340591e-08,
+      "loss": 0.7183,
+      "step": 5774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.85700024336233e-08,
+      "loss": 0.6762,
+      "step": 5775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.809266342872264e-08,
+      "loss": 0.7327,
+      "step": 5776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7618290960848016e-08,
+      "loss": 0.7619,
+      "step": 5777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7146885171264194e-08,
+      "loss": 0.2696,
+      "step": 5778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.667844620034777e-08,
+      "loss": 0.6906,
+      "step": 5779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.621297418759717e-08,
+      "loss": 0.7228,
+      "step": 5780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5750469271621516e-08,
+      "loss": 0.704,
+      "step": 5781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.529093159015062e-08,
+      "loss": 0.7226,
+      "step": 5782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.483436128002948e-08,
+      "loss": 0.7444,
+      "step": 5783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.438075847721933e-08,
+      "loss": 0.7422,
+      "step": 5784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393012331679657e-08,
+      "loss": 0.7048,
+      "step": 5785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.348245593295607e-08,
+      "loss": 0.7667,
+      "step": 5786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3037756459007865e-08,
+      "loss": 0.6741,
+      "step": 5787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.259602502737713e-08,
+      "loss": 0.7854,
+      "step": 5788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.215726176960754e-08,
+      "loss": 0.6953,
+      "step": 5789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1721466816356795e-08,
+      "loss": 0.7656,
+      "step": 5790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.128864029739887e-08,
+      "loss": 0.7105,
+      "step": 5791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.085878234162509e-08,
+      "loss": 0.747,
+      "step": 5792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0431893077041974e-08,
+      "loss": 0.7613,
+      "step": 5793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0007972630772265e-08,
+      "loss": 0.7186,
+      "step": 5794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9587021129053873e-08,
+      "loss": 0.7533,
+      "step": 5795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.916903869724097e-08,
+      "loss": 0.7412,
+      "step": 5796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.87540254598051e-08,
+      "loss": 0.7144,
+      "step": 5797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.834198154033074e-08,
+      "loss": 0.708,
+      "step": 5798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7932907061519744e-08,
+      "loss": 0.6809,
+      "step": 5799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7526802145190214e-08,
+      "loss": 0.7586,
+      "step": 5800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7123666912276525e-08,
+      "loss": 0.7126,
+      "step": 5801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6723501482824875e-08,
+      "loss": 0.7656,
+      "step": 5802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6326305976001054e-08,
+      "loss": 0.6686,
+      "step": 5803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5932080510084888e-08,
+      "loss": 0.2595,
+      "step": 5804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5540825202472473e-08,
+      "loss": 0.7231,
+      "step": 5805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5152540169675055e-08,
+      "loss": 0.7312,
+      "step": 5806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4767225527319026e-08,
+      "loss": 0.6946,
+      "step": 5807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.438488139014594e-08,
+      "loss": 0.7381,
+      "step": 5808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4005507872014723e-08,
+      "loss": 0.6923,
+      "step": 5809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.362910508589722e-08,
+      "loss": 0.6906,
+      "step": 5810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.325567314388155e-08,
+      "loss": 0.7343,
+      "step": 5811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2885212157170988e-08,
+      "loss": 0.7946,
+      "step": 5812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.251772223608506e-08,
+      "loss": 0.727,
+      "step": 5813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.215320349005845e-08,
+      "loss": 0.796,
+      "step": 5814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1791656027639883e-08,
+      "loss": 0.7143,
+      "step": 5815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.143307995649324e-08,
+      "loss": 0.7204,
+      "step": 5816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.107747538339977e-08,
+      "loss": 0.6851,
+      "step": 5817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0724842414252545e-08,
+      "loss": 0.8005,
+      "step": 5818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0375181154062006e-08,
+      "loss": 0.7093,
+      "step": 5819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0028491706953755e-08,
+      "loss": 0.7574,
+      "step": 5820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.968477417616632e-08,
+      "loss": 0.8207,
+      "step": 5821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9344028664056715e-08,
+      "loss": 0.7062,
+      "step": 5822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9006255272093767e-08,
+      "loss": 0.772,
+      "step": 5823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8671454100861462e-08,
+      "loss": 0.8216,
+      "step": 5824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8339625250061166e-08,
+      "loss": 0.7599,
+      "step": 5825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8010768818507164e-08,
+      "loss": 0.7469,
+      "step": 5826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7684884904127787e-08,
+      "loss": 0.7272,
+      "step": 5827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.736197360396985e-08,
+      "loss": 0.7022,
+      "step": 5828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7042035014189773e-08,
+      "loss": 0.7253,
+      "step": 5829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.672506923006245e-08,
+      "loss": 0.6891,
+      "step": 5830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.641107634597683e-08,
+      "loss": 0.7534,
+      "step": 5831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6100056455436996e-08,
+      "loss": 0.7561,
+      "step": 5832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.579200965105887e-08,
+      "loss": 0.7096,
+      "step": 5833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5486936024576847e-08,
+      "loss": 0.7536,
+      "step": 5834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.518483566683826e-08,
+      "loss": 0.6895,
+      "step": 5835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4885708667803368e-08,
+      "loss": 0.73,
+      "step": 5836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4589555116549804e-08,
+      "loss": 0.7472,
+      "step": 5837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4296375101268134e-08,
+      "loss": 0.7439,
+      "step": 5838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.400616870926519e-08,
+      "loss": 0.693,
+      "step": 5839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3718936026959618e-08,
+      "loss": 0.7257,
+      "step": 5840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3434677139885222e-08,
+      "loss": 0.7446,
+      "step": 5841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3153392132692067e-08,
+      "loss": 0.6918,
+      "step": 5842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.287508108914426e-08,
+      "loss": 0.7687,
+      "step": 5843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2599744092117728e-08,
+      "loss": 0.7564,
+      "step": 5844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2327381223604662e-08,
+      "loss": 0.7584,
+      "step": 5845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2057992564713516e-08,
+      "loss": 0.7019,
+      "step": 5846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1791578195662346e-08,
+      "loss": 0.7244,
+      "step": 5847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1528138195787685e-08,
+      "loss": 0.7602,
+      "step": 5848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1267672643539007e-08,
+      "loss": 0.7533,
+      "step": 5849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.101018161647871e-08,
+      "loss": 0.7067,
+      "step": 5850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0755665191286568e-08,
+      "loss": 0.7325,
+      "step": 5851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0504123443751957e-08,
+      "loss": 0.7473,
+      "step": 5852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0255556448782734e-08,
+      "loss": 0.7492,
+      "step": 5853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0009964280398576e-08,
+      "loss": 0.2439,
+      "step": 5854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.767347011734318e-09,
+      "loss": 0.768,
+      "step": 5855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.527704715037723e-09,
+      "loss": 0.6673,
+      "step": 5856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.291037461671704e-09,
+      "loss": 0.6945,
+      "step": 5857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.05734532211322e-09,
+      "loss": 0.6769,
+      "step": 5858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.826628365953272e-09,
+      "loss": 0.7487,
+      "step": 5859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.59888666189579e-09,
+      "loss": 0.6888,
+      "step": 5860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.374120277759856e-09,
+      "loss": 0.72,
+      "step": 5861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.152329280479709e-09,
+      "loss": 0.7911,
+      "step": 5862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.933513736099185e-09,
+      "loss": 0.6856,
+      "step": 5863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.717673709780604e-09,
+      "loss": 0.7191,
+      "step": 5864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.504809265799217e-09,
+      "loss": 0.783,
+      "step": 5865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.294920467542099e-09,
+      "loss": 0.7187,
+      "step": 5866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.088007377511474e-09,
+      "loss": 0.7187,
+      "step": 5867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8840700573258314e-09,
+      "loss": 0.7677,
+      "step": 5868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.683108567712149e-09,
+      "loss": 0.6508,
+      "step": 5869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.485122968516999e-09,
+      "loss": 0.2682,
+      "step": 5870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.290113318696556e-09,
+      "loss": 0.6917,
+      "step": 5871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0980796763232545e-09,
+      "loss": 0.7708,
+      "step": 5872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.909022098582462e-09,
+      "loss": 0.6728,
+      "step": 5873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.722940641772479e-09,
+      "loss": 0.7885,
+      "step": 5874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.539835361306756e-09,
+      "loss": 0.6195,
+      "step": 5875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.359706311710567e-09,
+      "loss": 0.7617,
+      "step": 5876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.182553546626556e-09,
+      "loss": 0.7445,
+      "step": 5877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.008377118806973e-09,
+      "loss": 0.7252,
+      "step": 5878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.837177080119215e-09,
+      "loss": 0.697,
+      "step": 5879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.668953481545835e-09,
+      "loss": 0.7788,
+      "step": 5880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503706373181205e-09,
+      "loss": 0.7157,
+      "step": 5881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.34143580423485e-09,
+      "loss": 0.7047,
+      "step": 5882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182141823027008e-09,
+      "loss": 0.2619,
+      "step": 5883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0258244769952885e-09,
+      "loss": 0.7083,
+      "step": 5884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.87248381268801e-09,
+      "loss": 0.7425,
+      "step": 5885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7221198757697585e-09,
+      "loss": 0.7219,
+      "step": 5886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.574732711014717e-09,
+      "loss": 0.712,
+      "step": 5887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4303223623155523e-09,
+      "loss": 0.7682,
+      "step": 5888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2888888726734237e-09,
+      "loss": 0.6731,
+      "step": 5889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1504322842079717e-09,
+      "loss": 0.7425,
+      "step": 5890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0149526381484385e-09,
+      "loss": 0.7811,
+      "step": 5891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.882449974840329e-09,
+      "loss": 0.6648,
+      "step": 5892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.752924333738749e-09,
+      "loss": 0.2356,
+      "step": 5893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6263757534183976e-09,
+      "loss": 0.7438,
+      "step": 5894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.502804271560244e-09,
+      "loss": 0.7961,
+      "step": 5895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3822099249659613e-09,
+      "loss": 0.729,
+      "step": 5896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.264592749544603e-09,
+      "loss": 0.7574,
+      "step": 5897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.149952780321485e-09,
+      "loss": 0.7533,
+      "step": 5898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.038290051435965e-09,
+      "loss": 0.7098,
+      "step": 5899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.929604596140333e-09,
+      "loss": 0.2685,
+      "step": 5900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8238964467975905e-09,
+      "loss": 0.6717,
+      "step": 5901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7211656348881113e-09,
+      "loss": 0.6928,
+      "step": 5902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6214121910040903e-09,
+      "loss": 0.7041,
+      "step": 5903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.524636144849545e-09,
+      "loss": 0.7623,
+      "step": 5904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.430837525244755e-09,
+      "loss": 0.7111,
+      "step": 5905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3400163601196004e-09,
+      "loss": 0.69,
+      "step": 5906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2521726765224452e-09,
+      "loss": 0.6722,
+      "step": 5907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1673065006101436e-09,
+      "loss": 0.7701,
+      "step": 5908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0854178576547026e-09,
+      "loss": 0.6914,
+      "step": 5909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0065067720432808e-09,
+      "loss": 0.647,
+      "step": 5910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.305732672726386e-10,
+      "loss": 0.6739,
+      "step": 5911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.576173659557984e-10,
+      "loss": 0.7422,
+      "step": 5912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.876390898187147e-10,
+      "loss": 0.6686,
+      "step": 5913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.206384596991634e-10,
+      "loss": 0.7397,
+      "step": 5914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.566154955500726e-10,
+      "loss": 0.7029,
+      "step": 5915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.955702164350818e-10,
+      "loss": 0.7331,
+      "step": 5916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.375026405352035e-10,
+      "loss": 0.7037,
+      "step": 5917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.824127851399408e-10,
+      "loss": 0.6405,
+      "step": 5918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3030066665505907e-10,
+      "loss": 0.7655,
+      "step": 5919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.811663006003663e-10,
+      "loss": 0.7344,
+      "step": 5920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3500970160527113e-10,
+      "loss": 0.7459,
+      "step": 5921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9183088341544483e-10,
+      "loss": 0.7479,
+      "step": 5922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.516298588894905e-10,
+      "loss": 0.6684,
+      "step": 5923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1440663999894306e-10,
+      "loss": 0.7479,
+      "step": 5924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8016123782826912e-10,
+      "loss": 0.7008,
+      "step": 5925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4889366257486715e-10,
+      "loss": 0.6871,
+      "step": 5926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.206039235501777e-10,
+      "loss": 0.6399,
+      "step": 5927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.529202917857306e-11,
+      "loss": 0.7232,
+      "step": 5928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.295798699735735e-11,
+      "loss": 0.783,
+      "step": 5929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3601803657876705e-11,
+      "loss": 0.7746,
+      "step": 5930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.722348492329886e-11,
+      "loss": 0.742,
+      "step": 5931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.382303567194377e-11,
+      "loss": 0.7074,
+      "step": 5932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3400459892842777e-11,
+      "loss": 0.7541,
+      "step": 5933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9557606912896695e-12,
+      "loss": 0.7052,
+      "step": 5934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4889402832896084e-12,
+      "loss": 0.6963,
+      "step": 5935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.2808,
+      "step": 5936
+    },
+    {
+      "epoch": 1.0,
+      "step": 5936,
+      "total_flos": 2798549630189568.0,
+      "train_loss": 0.17301319674852322,
+      "train_runtime": 23979.0097,
+      "train_samples_per_second": 31.686,
+      "train_steps_per_second": 0.248
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5936,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1500,
+  "total_flos": 2798549630189568.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}