diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8382932349735938,
+  "global_step": 5000,
+  "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": 1.3965,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 1.6401,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 1.5083,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 1.437,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 1.4478,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.4858,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 1.5767,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 0.9179,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.5225,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 1.5337,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 1.5601,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 1.4561,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 1.4609,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 0.8776,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 1.418,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 1.4297,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 1.3599,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.3145,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 1.3096,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 1.3159,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 0.8228,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 1.2876,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 1.2681,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 1.2275,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 1.2979,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 1.2644,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 1.1946,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 1.3394,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 1.21,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 1.2227,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 1.2395,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 1.1997,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 1.1387,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 1.23,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 1.144,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 0.7685,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.7113,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 1.2593,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 1.1636,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 1.1938,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 1.1299,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 1.0947,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.1335,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 1.2063,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 1.2195,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 1.1895,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 1.1694,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 1.1582,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 1.1624,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 1.1099,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 0.7074,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 1.1587,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 0.739,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 1.1802,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 0.7576,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 1.1191,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 1.1062,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 1.0662,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 1.1785,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 1.166,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 1.0884,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 1.1294,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 1.1333,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 1.1392,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 1.1277,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 1.0369,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 1.1086,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 1.1292,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 1.0435,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 1.1675,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 1.0979,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 1.0823,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 1.0161,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 1.064,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 1.113,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 0.9771,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 1.1418,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 1.0527,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 1.0291,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 1.0681,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 1.0769,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 1.0842,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 1.0583,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 1.125,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 1.0771,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 1.0579,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 1.0854,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 0.6856,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.7183,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.005586592178771e-05,
+      "loss": 0.7516,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0167597765363129e-05,
+      "loss": 1.048,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0279329608938548e-05,
+      "loss": 1.095,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0391061452513967e-05,
+      "loss": 0.7249,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 1.0425,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0614525139664806e-05,
+      "loss": 1.1333,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0726256983240225e-05,
+      "loss": 1.1306,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0837988826815644e-05,
+      "loss": 0.9962,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0949720670391063e-05,
+      "loss": 1.0566,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 1.0615,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.11731843575419e-05,
+      "loss": 1.0457,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1284916201117319e-05,
+      "loss": 1.1123,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1396648044692738e-05,
+      "loss": 0.991,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1508379888268157e-05,
+      "loss": 1.0898,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 0.6447,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1731843575418994e-05,
+      "loss": 1.1233,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1843575418994415e-05,
+      "loss": 1.0549,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1955307262569834e-05,
+      "loss": 1.0474,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2067039106145253e-05,
+      "loss": 1.0334,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 1.0378,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.229050279329609e-05,
+      "loss": 1.0715,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2402234636871509e-05,
+      "loss": 1.1006,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2513966480446928e-05,
+      "loss": 0.7164,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625698324022347e-05,
+      "loss": 1.0559,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 1.0779,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2849162011173186e-05,
+      "loss": 1.1016,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2960893854748603e-05,
+      "loss": 0.9961,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3072625698324024e-05,
+      "loss": 1.0547,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3184357541899443e-05,
+      "loss": 1.1147,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 1.1252,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.340782122905028e-05,
+      "loss": 1.0442,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3519553072625699e-05,
+      "loss": 1.1357,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3631284916201118e-05,
+      "loss": 1.0549,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3743016759776537e-05,
+      "loss": 1.0911,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 1.0449,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3966480446927374e-05,
+      "loss": 1.0879,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4078212290502795e-05,
+      "loss": 1.1355,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4189944134078212e-05,
+      "loss": 0.2795,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4301675977653633e-05,
+      "loss": 1.0469,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 1.042,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.452513966480447e-05,
+      "loss": 1.0938,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.463687150837989e-05,
+      "loss": 1.0715,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4748603351955308e-05,
+      "loss": 1.0815,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4860335195530729e-05,
+      "loss": 1.0532,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 1.0701,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083798882681566e-05,
+      "loss": 0.7221,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5195530726256983e-05,
+      "loss": 1.074,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5307262569832404e-05,
+      "loss": 1.0583,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5418994413407823e-05,
+      "loss": 1.054,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 1.0054,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564245810055866e-05,
+      "loss": 1.0708,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575418994413408e-05,
+      "loss": 1.0056,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5865921787709498e-05,
+      "loss": 0.6391,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5977653631284917e-05,
+      "loss": 1.02,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 1.0432,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6201117318435755e-05,
+      "loss": 1.0522,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6312849162011177e-05,
+      "loss": 1.0745,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6424581005586592e-05,
+      "loss": 1.0256,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6536312849162014e-05,
+      "loss": 1.033,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 1.0574,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6759776536312852e-05,
+      "loss": 1.0178,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.687150837988827e-05,
+      "loss": 1.0232,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698324022346369e-05,
+      "loss": 0.9949,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.709497206703911e-05,
+      "loss": 1.0417,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 1.0864,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7318435754189946e-05,
+      "loss": 1.1296,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7430167597765365e-05,
+      "loss": 1.0635,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7541899441340784e-05,
+      "loss": 1.1008,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7653631284916203e-05,
+      "loss": 1.1282,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 1.031,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.787709497206704e-05,
+      "loss": 1.0564,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.798882681564246e-05,
+      "loss": 1.0188,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8100558659217878e-05,
+      "loss": 1.1133,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8212290502793297e-05,
+      "loss": 1.0581,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 1.082,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8435754189944135e-05,
+      "loss": 1.0596,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8547486033519553e-05,
+      "loss": 1.0608,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8659217877094972e-05,
+      "loss": 1.0781,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8770949720670394e-05,
+      "loss": 1.0571,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 1.0747,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8994413407821232e-05,
+      "loss": 1.063,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.910614525139665e-05,
+      "loss": 1.1191,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.921787709497207e-05,
+      "loss": 0.64,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.932960893854749e-05,
+      "loss": 0.6683,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 1.0903,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9553072625698326e-05,
+      "loss": 1.0615,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9664804469273745e-05,
+      "loss": 1.0513,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776536312849164e-05,
+      "loss": 1.0398,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9888268156424583e-05,
+      "loss": 1.1448,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.0552,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998525438088e-05,
+      "loss": 0.6816,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999410175278e-05,
+      "loss": 1.0698,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986728945384e-05,
+      "loss": 1.1191,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976407018074e-05,
+      "loss": 1.0398,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963135973894e-05,
+      "loss": 1.0171,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999946915816756e-05,
+      "loss": 0.6702,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999927746551447e-05,
+      "loss": 1.0134,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999905628183617e-05,
+      "loss": 1.0081,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988056071979e-05,
+      "loss": 1.0142,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999985254416736e-05,
+      "loss": 1.0327,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999821578534588e-05,
+      "loss": 1.0569,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999787663830606e-05,
+      "loss": 1.0508,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999750800065415e-05,
+      "loss": 1.0989,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999710987249892e-05,
+      "loss": 0.9934,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999668225395772e-05,
+      "loss": 1.0564,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999622514515667e-05,
+      "loss": 1.0496,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999957385462306e-05,
+      "loss": 1.054,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999522245732302e-05,
+      "loss": 1.1375,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999946768785861e-05,
+      "loss": 0.6911,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999410181018074e-05,
+      "loss": 1.0383,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999934972522766e-05,
+      "loss": 1.0562,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999928632050519e-05,
+      "loss": 0.7196,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999921996686936e-05,
+      "loss": 1.0652,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999150664339745e-05,
+      "loss": 1.0498,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999078412936784e-05,
+      "loss": 0.9976,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999900321268178e-05,
+      "loss": 1.0657,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998925063596915e-05,
+      "loss": 1.1108,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998843965705235e-05,
+      "loss": 0.687,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998759919030653e-05,
+      "loss": 1.0933,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999867292359796e-05,
+      "loss": 1.0547,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999858297943281e-05,
+      "loss": 1.0208,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998490086561727e-05,
+      "loss": 1.0295,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839424501211e-05,
+      "loss": 1.0439,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998295454812218e-05,
+      "loss": 1.0391,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998193715991192e-05,
+      "loss": 1.0417,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998089028579035e-05,
+      "loss": 1.0125,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997981392606617e-05,
+      "loss": 1.0759,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997870808105683e-05,
+      "loss": 0.9878,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997757275108847e-05,
+      "loss": 1.054,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999764079364959e-05,
+      "loss": 1.0464,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997521363762264e-05,
+      "loss": 1.0415,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999739898548209e-05,
+      "loss": 1.0916,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999727365884516e-05,
+      "loss": 1.0723,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999714538388843e-05,
+      "loss": 1.0535,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997014160649736e-05,
+      "loss": 0.9819,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996879989167777e-05,
+      "loss": 1.0769,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999674286948212e-05,
+      "loss": 0.7207,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996602801633197e-05,
+      "loss": 1.0234,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999645978566233e-05,
+      "loss": 1.0522,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996313821611682e-05,
+      "loss": 1.0591,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996164909524306e-05,
+      "loss": 1.0107,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996013049444117e-05,
+      "loss": 1.0508,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995858241415904e-05,
+      "loss": 1.074,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999570048548532e-05,
+      "loss": 1.0002,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995539781698882e-05,
+      "loss": 1.0327,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995376130103992e-05,
+      "loss": 1.0564,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995209530748913e-05,
+      "loss": 1.094,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995039983682774e-05,
+      "loss": 1.0596,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994867488955573e-05,
+      "loss": 1.0308,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994692046618188e-05,
+      "loss": 1.0566,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994513656722356e-05,
+      "loss": 1.0527,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994332319320688e-05,
+      "loss": 1.0249,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994148034466655e-05,
+      "loss": 0.9893,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993960802214617e-05,
+      "loss": 1.0667,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993770622619784e-05,
+      "loss": 0.9702,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993577495738245e-05,
+      "loss": 0.6655,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999338142162695e-05,
+      "loss": 1.0684,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993182400343727e-05,
+      "loss": 0.9873,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992980431947276e-05,
+      "loss": 1.0254,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992775516497148e-05,
+      "loss": 1.0615,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992567654053785e-05,
+      "loss": 1.0051,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992356844678485e-05,
+      "loss": 1.1011,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992143088433417e-05,
+      "loss": 1.0359,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991926385381623e-05,
+      "loss": 0.6763,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991706735587006e-05,
+      "loss": 1.0222,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999148413911435e-05,
+      "loss": 1.1313,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991258596029297e-05,
+      "loss": 1.061,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991030106398367e-05,
+      "loss": 1.0444,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999079867028894e-05,
+      "loss": 1.0503,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999056428776927e-05,
+      "loss": 1.074,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999032695890848e-05,
+      "loss": 0.6311,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990086683776564e-05,
+      "loss": 1.0747,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989843462444377e-05,
+      "loss": 1.0862,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998959729498365e-05,
+      "loss": 1.1108,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989348181466987e-05,
+      "loss": 0.9778,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989096121967842e-05,
+      "loss": 1.0063,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988841116560566e-05,
+      "loss": 1.0454,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988583165320347e-05,
+      "loss": 1.0562,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998832226832327e-05,
+      "loss": 1.0066,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988058425646268e-05,
+      "loss": 0.9983,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987791637367157e-05,
+      "loss": 0.6696,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998752190356462e-05,
+      "loss": 1.0278,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99872492243182e-05,
+      "loss": 0.6782,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986973599708305e-05,
+      "loss": 1.1042,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986695029816237e-05,
+      "loss": 1.0198,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986413514724136e-05,
+      "loss": 0.9702,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986129054515035e-05,
+      "loss": 1.0559,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985841649272814e-05,
+      "loss": 0.9819,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998555129908224e-05,
+      "loss": 0.7016,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985258004028938e-05,
+      "loss": 1.0339,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998496176419941e-05,
+      "loss": 1.0188,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998466257968101e-05,
+      "loss": 1.0271,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998436045056198e-05,
+      "loss": 1.0535,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984055376931414e-05,
+      "loss": 1.0222,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998374735887929e-05,
+      "loss": 1.0342,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983436396496442e-05,
+      "loss": 1.0327,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983122489874582e-05,
+      "loss": 1.0374,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982805639106276e-05,
+      "loss": 1.0618,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982485844284972e-05,
+      "loss": 1.0164,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982163105504983e-05,
+      "loss": 0.7015,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981837422861487e-05,
+      "loss": 1.0247,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981508796450532e-05,
+      "loss": 1.0083,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998117722636903e-05,
+      "loss": 0.9976,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980842712714774e-05,
+      "loss": 1.001,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980505255586408e-05,
+      "loss": 1.0242,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980164855083454e-05,
+      "loss": 1.0786,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979821511306308e-05,
+      "loss": 0.9976,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979475224356214e-05,
+      "loss": 1.0066,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979125994335306e-05,
+      "loss": 1.0488,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997877382134657e-05,
+      "loss": 1.064,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978418705493868e-05,
+      "loss": 0.6763,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978060646881933e-05,
+      "loss": 1.0471,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977699645616353e-05,
+      "loss": 1.0122,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977335701803597e-05,
+      "loss": 1.0691,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976968815550994e-05,
+      "loss": 1.0435,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976598986966746e-05,
+      "loss": 0.9756,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976226216159916e-05,
+      "loss": 1.02,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975850503240444e-05,
+      "loss": 0.6697,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997547184831913e-05,
+      "loss": 1.021,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975090251507637e-05,
+      "loss": 0.991,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974705712918515e-05,
+      "loss": 1.0039,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974318232665164e-05,
+      "loss": 1.1057,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973927810861855e-05,
+      "loss": 1.0043,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973534447623727e-05,
+      "loss": 0.9932,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973138143066793e-05,
+      "loss": 1.0078,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997273889730792e-05,
+      "loss": 1.0073,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997233671046486e-05,
+      "loss": 1.0159,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997193158265622e-05,
+      "loss": 0.9412,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971523514001473e-05,
+      "loss": 1.084,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971112504620968e-05,
+      "loss": 1.0071,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970698554635913e-05,
+      "loss": 1.0566,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997028166416839e-05,
+      "loss": 1.0405,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996986183334134e-05,
+      "loss": 0.9978,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969439062278588e-05,
+      "loss": 1.0654,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969013351104803e-05,
+      "loss": 1.0179,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968584699945534e-05,
+      "loss": 0.6786,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968153108927198e-05,
+      "loss": 1.0693,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996771857817708e-05,
+      "loss": 1.0337,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967281107823326e-05,
+      "loss": 1.032,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966840697994945e-05,
+      "loss": 1.0735,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996639734882183e-05,
+      "loss": 1.0234,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996595106043472e-05,
+      "loss": 0.7045,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965501832965238e-05,
+      "loss": 0.624,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965049666545863e-05,
+      "loss": 0.682,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964594561309946e-05,
+      "loss": 1.0469,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964136517391708e-05,
+      "loss": 1.0664,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963675534926224e-05,
+      "loss": 0.6454,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963211614049443e-05,
+      "loss": 1.0156,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996274475489819e-05,
+      "loss": 1.0735,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962274957610136e-05,
+      "loss": 1.012,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996180222232384e-05,
+      "loss": 0.9714,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961326549178713e-05,
+      "loss": 0.6838,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960847938315037e-05,
+      "loss": 0.6712,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996036638987396e-05,
+      "loss": 1.0474,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99598819039975e-05,
+      "loss": 0.9456,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995939448082853e-05,
+      "loss": 0.9597,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958904120510807e-05,
+      "loss": 0.6865,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995841082318894e-05,
+      "loss": 1.0366,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957914589008405e-05,
+      "loss": 0.948,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957415418115552e-05,
+      "loss": 1.0142,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995691331065759e-05,
+      "loss": 1.0002,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99564082667826e-05,
+      "loss": 0.9875,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955900286639524e-05,
+      "loss": 1.0667,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955389370378172e-05,
+      "loss": 1.0181,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954875518149216e-05,
+      "loss": 1.0303,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99543587301042e-05,
+      "loss": 0.9773,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953839006395534e-05,
+      "loss": 1.0059,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995331634717649e-05,
+      "loss": 1.113,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99527907526012e-05,
+      "loss": 1.0339,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995226222282468e-05,
+      "loss": 0.6426,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951730758002787e-05,
+      "loss": 1.0457,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951196358292266e-05,
+      "loss": 1.0095,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950659023850716e-05,
+      "loss": 1.0598,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99501187548366e-05,
+      "loss": 0.6615,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949575551409258e-05,
+      "loss": 0.9937,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949029413728877e-05,
+      "loss": 1.0483,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948480341956524e-05,
+      "loss": 1.0393,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947928336254132e-05,
+      "loss": 1.0225,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947373396784487e-05,
+      "loss": 1.0562,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946815523711253e-05,
+      "loss": 0.9717,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946254717198954e-05,
+      "loss": 1.093,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945690977412972e-05,
+      "loss": 1.0024,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945124304519568e-05,
+      "loss": 1.0132,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944554698685853e-05,
+      "loss": 1.0286,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943982160079823e-05,
+      "loss": 1.0708,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943406688870316e-05,
+      "loss": 1.0251,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942828285227047e-05,
+      "loss": 1.0818,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942246949320598e-05,
+      "loss": 1.0061,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941662681322412e-05,
+      "loss": 1.0684,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941075481404795e-05,
+      "loss": 0.9868,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994048534974092e-05,
+      "loss": 1.0925,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939892286504822e-05,
+      "loss": 0.9727,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993929629187141e-05,
+      "loss": 1.04,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993869736601644e-05,
+      "loss": 1.0518,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938095509116547e-05,
+      "loss": 1.0212,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937490721349232e-05,
+      "loss": 0.7165,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936883002892843e-05,
+      "loss": 1.0369,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936272353926616e-05,
+      "loss": 0.6577,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935658774630632e-05,
+      "loss": 0.6784,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993504226518584e-05,
+      "loss": 1.0332,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934422825774064e-05,
+      "loss": 1.0229,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933800456577978e-05,
+      "loss": 1.0381,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993317515778113e-05,
+      "loss": 0.9768,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932546929567924e-05,
+      "loss": 0.9897,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993191577212364e-05,
+      "loss": 1.0974,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993128168563441e-05,
+      "loss": 0.6458,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930644670287233e-05,
+      "loss": 1.0239,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993000472626997e-05,
+      "loss": 1.1238,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929361853771355e-05,
+      "loss": 1.0085,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928716052980977e-05,
+      "loss": 1.0049,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928067324089286e-05,
+      "loss": 1.0635,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927415667287605e-05,
+      "loss": 1.0525,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926761082768116e-05,
+      "loss": 1.0107,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992610357072386e-05,
+      "loss": 1.0027,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992544313134875e-05,
+      "loss": 1.0586,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924779764837553e-05,
+      "loss": 1.0479,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992411347138591e-05,
+      "loss": 1.0679,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923444251190314e-05,
+      "loss": 0.9802,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992277210444813e-05,
+      "loss": 1.0784,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922097031357575e-05,
+      "loss": 1.0918,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921419032117752e-05,
+      "loss": 0.9968,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920738106928593e-05,
+      "loss": 1.0361,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992005425599092e-05,
+      "loss": 0.708,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919367479506413e-05,
+      "loss": 1.0173,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918677777677605e-05,
+      "loss": 0.9934,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99179851507079e-05,
+      "loss": 0.9814,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917289598801556e-05,
+      "loss": 0.9834,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991659112216371e-05,
+      "loss": 0.6285,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915889721000347e-05,
+      "loss": 1.0364,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991518539551832e-05,
+      "loss": 1.0078,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991447814592534e-05,
+      "loss": 1.0408,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913767972429984e-05,
+      "loss": 1.0442,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913054875241694e-05,
+      "loss": 1.0405,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991233885457077e-05,
+      "loss": 1.0374,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911619910628374e-05,
+      "loss": 1.0554,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910898043626534e-05,
+      "loss": 0.9856,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910173253778136e-05,
+      "loss": 0.6823,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990944554129693e-05,
+      "loss": 1.0601,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908714906397525e-05,
+      "loss": 1.0178,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99079813492954e-05,
+      "loss": 0.9968,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990724487020688e-05,
+      "loss": 1.0305,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906505469349176e-05,
+      "loss": 1.0447,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905763146940332e-05,
+      "loss": 0.6098,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990501790319928e-05,
+      "loss": 0.666,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904269738345793e-05,
+      "loss": 1.0027,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903518652600518e-05,
+      "loss": 0.9712,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990276464618496e-05,
+      "loss": 0.9736,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902007719321483e-05,
+      "loss": 1.009,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901247872233316e-05,
+      "loss": 1.0227,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900485105144544e-05,
+      "loss": 1.0525,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989971941828012e-05,
+      "loss": 0.9949,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898950811865854e-05,
+      "loss": 1.0664,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898179286128415e-05,
+      "loss": 1.0449,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897404841295337e-05,
+      "loss": 1.0452,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896627477595017e-05,
+      "loss": 1.0349,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895847195256702e-05,
+      "loss": 1.0437,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895063994510512e-05,
+      "loss": 1.0781,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989427787558742e-05,
+      "loss": 0.9771,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893488838719267e-05,
+      "loss": 1.0271,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892696884138743e-05,
+      "loss": 0.989,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989190201207941e-05,
+      "loss": 1.0388,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891104222775683e-05,
+      "loss": 1.0425,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890303516462842e-05,
+      "loss": 1.0425,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889499893377024e-05,
+      "loss": 1.0071,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988869335375523e-05,
+      "loss": 1.1042,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988788389783531e-05,
+      "loss": 1.0923,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887071525855993e-05,
+      "loss": 1.0603,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886256238056854e-05,
+      "loss": 1.001,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885438034678327e-05,
+      "loss": 1.0322,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884616915961717e-05,
+      "loss": 1.0063,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883792882149177e-05,
+      "loss": 0.981,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988296593348373e-05,
+      "loss": 1.0789,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988213607020925e-05,
+      "loss": 0.9897,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988130329257047e-05,
+      "loss": 1.0242,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880467600812992e-05,
+      "loss": 0.9792,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879628995183274e-05,
+      "loss": 1.0103,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878787475928624e-05,
+      "loss": 0.9866,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877943043297223e-05,
+      "loss": 1.0491,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98770956975381e-05,
+      "loss": 0.9966,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987624543890115e-05,
+      "loss": 0.69,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875392267637123e-05,
+      "loss": 1.0779,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874536183997636e-05,
+      "loss": 1.0291,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987367718823515e-05,
+      "loss": 1.0171,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872815280602996e-05,
+      "loss": 0.988,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871950461355365e-05,
+      "loss": 1.0471,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871082730747298e-05,
+      "loss": 1.0264,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870212089034707e-05,
+      "loss": 1.0168,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869338536474345e-05,
+      "loss": 1.0552,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986846207332384e-05,
+      "loss": 0.9009,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986758269984167e-05,
+      "loss": 1.0471,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866700416287174e-05,
+      "loss": 0.9736,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986581522292055e-05,
+      "loss": 1.001,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864927120002845e-05,
+      "loss": 1.0137,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864036107795978e-05,
+      "loss": 1.0273,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986314218656272e-05,
+      "loss": 1.0007,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98622453565667e-05,
+      "loss": 1.0149,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861345618072395e-05,
+      "loss": 0.9834,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986044297134516e-05,
+      "loss": 0.6696,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859537416651197e-05,
+      "loss": 0.9326,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858628954257554e-05,
+      "loss": 0.6896,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985771758443216e-05,
+      "loss": 0.6857,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856803307443782e-05,
+      "loss": 1.0278,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855886123562055e-05,
+      "loss": 1.1196,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854966033057462e-05,
+      "loss": 0.9717,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985404303620136e-05,
+      "loss": 0.6649,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985311713326594e-05,
+      "loss": 0.9863,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852188324524272e-05,
+      "loss": 1.064,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851256610250268e-05,
+      "loss": 1.0383,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850321990718703e-05,
+      "loss": 1.0134,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984938446620521e-05,
+      "loss": 1.0088,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984844403698627e-05,
+      "loss": 1.0466,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847500703339238e-05,
+      "loss": 1.0425,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846554465542306e-05,
+      "loss": 0.9834,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845605323874532e-05,
+      "loss": 0.9827,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844653278615836e-05,
+      "loss": 1.0364,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984369833004698e-05,
+      "loss": 0.9788,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842740478449594e-05,
+      "loss": 1.0508,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841779724106165e-05,
+      "loss": 1.0251,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840816067300022e-05,
+      "loss": 1.0291,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839849508315365e-05,
+      "loss": 0.7133,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838880047437243e-05,
+      "loss": 0.9797,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983790768495156e-05,
+      "loss": 1.0574,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836932421145084e-05,
+      "loss": 1.0964,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835954256305423e-05,
+      "loss": 1.0137,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834973190721057e-05,
+      "loss": 1.0801,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833989224681314e-05,
+      "loss": 1.0308,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833002358476375e-05,
+      "loss": 1.0046,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983201259239728e-05,
+      "loss": 1.0605,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831019926735924e-05,
+      "loss": 1.0337,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830024361785055e-05,
+      "loss": 0.666,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982902589783828e-05,
+      "loss": 1.0164,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828024535190056e-05,
+      "loss": 1.0222,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98270202741357e-05,
+      "loss": 1.0198,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826013114971375e-05,
+      "loss": 0.6498,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825003057994112e-05,
+      "loss": 1.043,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823990103501783e-05,
+      "loss": 0.6598,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822974251793126e-05,
+      "loss": 1.0457,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821955503167726e-05,
+      "loss": 1.0447,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820933857926024e-05,
+      "loss": 0.6392,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819909316369314e-05,
+      "loss": 0.6658,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818881878799755e-05,
+      "loss": 0.6357,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981785154552034e-05,
+      "loss": 1.0027,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816818316834933e-05,
+      "loss": 1.0637,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981578219304824e-05,
+      "loss": 1.0349,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981474317446584e-05,
+      "loss": 0.6295,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813701261394136e-05,
+      "loss": 1.0322,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981265645414041e-05,
+      "loss": 1.075,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811608753012792e-05,
+      "loss": 0.9688,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981055815832025e-05,
+      "loss": 1.1584,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809504670372634e-05,
+      "loss": 0.99,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808448289480614e-05,
+      "loss": 1.0369,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807389015955742e-05,
+      "loss": 1.0071,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806326850110406e-05,
+      "loss": 0.67,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980526179225785e-05,
+      "loss": 0.6703,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980419384271217e-05,
+      "loss": 1.0278,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803123001788327e-05,
+      "loss": 1.0186,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980204926980212e-05,
+      "loss": 1.0251,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980097264707021e-05,
+      "loss": 1.0388,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97998931339101e-05,
+      "loss": 0.9736,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798810730640153e-05,
+      "loss": 1.031,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797725437579586e-05,
+      "loss": 0.9822,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796637255048462e-05,
+      "loss": 1.0271,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795546183367708e-05,
+      "loss": 1.0044,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979445222285908e-05,
+      "loss": 1.106,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793355373845214e-05,
+      "loss": 1.0032,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792255636649574e-05,
+      "loss": 1.0874,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791153011596497e-05,
+      "loss": 1.0068,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790047499011152e-05,
+      "loss": 0.9717,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788939099219572e-05,
+      "loss": 1.0632,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787827812548634e-05,
+      "loss": 1.0271,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786713639326076e-05,
+      "loss": 1.0125,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978559657988048e-05,
+      "loss": 0.9658,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784476634541276e-05,
+      "loss": 1.0322,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783353803638757e-05,
+      "loss": 1.0383,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978222808750405e-05,
+      "loss": 1.0334,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978109948646915e-05,
+      "loss": 1.0366,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97799680008669e-05,
+      "loss": 1.0327,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977883363103098e-05,
+      "loss": 0.9832,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777696377295933e-05,
+      "loss": 1.0269,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776556239997146e-05,
+      "loss": 1.0078,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775413219470864e-05,
+      "loss": 1.04,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774267316054177e-05,
+      "loss": 1.0154,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773118530085024e-05,
+      "loss": 1.0618,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97719668619022e-05,
+      "loss": 0.9551,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770812311845344e-05,
+      "loss": 1.0376,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769654880254944e-05,
+      "loss": 1.0488,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768494567472348e-05,
+      "loss": 1.0642,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976733137383974e-05,
+      "loss": 0.9648,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766165299700164e-05,
+      "loss": 1.0369,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764996345397505e-05,
+      "loss": 1.021,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976382451127651e-05,
+      "loss": 0.9875,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976264979768276e-05,
+      "loss": 0.6968,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97614722049627e-05,
+      "loss": 1.0449,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976029173346361e-05,
+      "loss": 0.9971,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975910838353363e-05,
+      "loss": 1.0205,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757922155521737e-05,
+      "loss": 1.001,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756733049777772e-05,
+      "loss": 1.042,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755541066652416e-05,
+      "loss": 1.0723,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754346206497197e-05,
+      "loss": 1.0237,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753148469664496e-05,
+      "loss": 1.0474,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975194785650754e-05,
+      "loss": 0.9634,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975074436738041e-05,
+      "loss": 1.0417,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749538002638015e-05,
+      "loss": 1.0645,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748328762636138e-05,
+      "loss": 1.0105,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747116647731404e-05,
+      "loss": 0.9834,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745901658281267e-05,
+      "loss": 0.676,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744683794644052e-05,
+      "loss": 0.9758,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974346305717892e-05,
+      "loss": 1.0112,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974223944624588e-05,
+      "loss": 1.0059,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974101296220579e-05,
+      "loss": 0.969,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739783605420357e-05,
+      "loss": 0.9492,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738551376252134e-05,
+      "loss": 1.0239,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973731627506452e-05,
+      "loss": 1.0928,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973607830222176e-05,
+      "loss": 0.9951,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734837458088947e-05,
+      "loss": 0.9421,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733593743032024e-05,
+      "loss": 1.0691,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732347157417776e-05,
+      "loss": 0.9966,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731097701613843e-05,
+      "loss": 1.0029,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729845375988694e-05,
+      "loss": 0.9829,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972859018091166e-05,
+      "loss": 0.981,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727332116752918e-05,
+      "loss": 0.9893,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726071183883482e-05,
+      "loss": 1.0688,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724807382675215e-05,
+      "loss": 1.0225,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723540713500834e-05,
+      "loss": 1.0247,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972227117673389e-05,
+      "loss": 0.6411,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972099877274879e-05,
+      "loss": 1.0188,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971972350192077e-05,
+      "loss": 0.9988,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971844536462594e-05,
+      "loss": 1.0405,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717164361241225e-05,
+      "loss": 0.946,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715880492144415e-05,
+      "loss": 0.9965,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714593757714143e-05,
+      "loss": 1.0623,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713304158329873e-05,
+      "loss": 1.0569,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712011694371932e-05,
+      "loss": 1.1006,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971071636622148e-05,
+      "loss": 1.0022,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.9885,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708117118871918e-05,
+      "loss": 0.9683,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706813200439357e-05,
+      "loss": 1.0337,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970550641934739e-05,
+      "loss": 0.9548,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704196775981397e-05,
+      "loss": 1.0459,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970288427072761e-05,
+      "loss": 0.9778,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701568903973104e-05,
+      "loss": 1.0298,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700250676105793e-05,
+      "loss": 0.7368,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698929587514444e-05,
+      "loss": 0.9971,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697605638588655e-05,
+      "loss": 1.0818,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696278829718882e-05,
+      "loss": 0.9919,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694949161296416e-05,
+      "loss": 1.043,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969361663371339e-05,
+      "loss": 1.0391,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692281247362785e-05,
+      "loss": 0.991,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690943002638425e-05,
+      "loss": 1.0413,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968960189993497e-05,
+      "loss": 0.9812,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968825793964793e-05,
+      "loss": 0.9856,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686911122173657e-05,
+      "loss": 1.0154,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685561447909342e-05,
+      "loss": 1.0068,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684208917253018e-05,
+      "loss": 0.9985,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968285353060357e-05,
+      "loss": 1.0356,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681495288360716e-05,
+      "loss": 0.9694,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680134190925015e-05,
+      "loss": 0.9778,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678770238697876e-05,
+      "loss": 1.0061,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677403432081543e-05,
+      "loss": 1.0415,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676033771479102e-05,
+      "loss": 1.0552,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674661257294487e-05,
+      "loss": 0.6369,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673285889932468e-05,
+      "loss": 0.9436,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671907669798655e-05,
+      "loss": 0.9712,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670526597299505e-05,
+      "loss": 1.0728,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966914267284232e-05,
+      "loss": 0.9949,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966775589683522e-05,
+      "loss": 0.9739,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96663662696872e-05,
+      "loss": 1.042,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664973791808064e-05,
+      "loss": 0.9741,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966357846360848e-05,
+      "loss": 0.9663,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662180285499944e-05,
+      "loss": 1.0061,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96607792578948e-05,
+      "loss": 0.9619,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965937538120622e-05,
+      "loss": 1.0298,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657968655848232e-05,
+      "loss": 0.9368,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656559082235697e-05,
+      "loss": 1.0134,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965514666078431e-05,
+      "loss": 1.0286,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653731391910614e-05,
+      "loss": 0.9585,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652313276031995e-05,
+      "loss": 1.0151,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650892313566667e-05,
+      "loss": 1.0254,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649468504933685e-05,
+      "loss": 1.0139,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964804185055296e-05,
+      "loss": 0.6592,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646612350845223e-05,
+      "loss": 1.0142,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645180006232046e-05,
+      "loss": 0.6238,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643744817135856e-05,
+      "loss": 0.6639,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642306783979902e-05,
+      "loss": 0.9756,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640865907188273e-05,
+      "loss": 1.0454,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639422187185913e-05,
+      "loss": 0.6786,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637975624398583e-05,
+      "loss": 0.9575,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636526219252898e-05,
+      "loss": 1.1035,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635073972176303e-05,
+      "loss": 1.0457,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633618883597087e-05,
+      "loss": 0.9958,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963216095394437e-05,
+      "loss": 0.9539,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630700183648112e-05,
+      "loss": 1.0256,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629236573139118e-05,
+      "loss": 0.9785,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627770122849017e-05,
+      "loss": 0.9761,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626300833210293e-05,
+      "loss": 0.9541,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962482870465625e-05,
+      "loss": 1.0271,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623353737621035e-05,
+      "loss": 0.9993,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621875932539643e-05,
+      "loss": 0.9148,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620395289847894e-05,
+      "loss": 0.9639,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618911809982445e-05,
+      "loss": 1.0493,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617425493380796e-05,
+      "loss": 0.9553,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615936340481276e-05,
+      "loss": 1.0642,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614444351723056e-05,
+      "loss": 0.9363,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612949527546147e-05,
+      "loss": 1.0283,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611451868391384e-05,
+      "loss": 0.9531,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609951374700443e-05,
+      "loss": 1.1069,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608448046915852e-05,
+      "loss": 0.9902,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606941885480947e-05,
+      "loss": 0.9468,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960543289083992e-05,
+      "loss": 0.9905,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603921063437795e-05,
+      "loss": 0.7028,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960240640372042e-05,
+      "loss": 1.0166,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600888912134495e-05,
+      "loss": 0.9929,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599368589127542e-05,
+      "loss": 0.9551,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597845435147924e-05,
+      "loss": 0.9707,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959631945064484e-05,
+      "loss": 1.0559,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594790636068323e-05,
+      "loss": 1.0737,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593258991869235e-05,
+      "loss": 0.99,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959172451849928e-05,
+      "loss": 1.0278,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590187216410987e-05,
+      "loss": 0.9758,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588647086057734e-05,
+      "loss": 1.0437,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958710412789372e-05,
+      "loss": 1.0356,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585558342373986e-05,
+      "loss": 0.9998,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584009729954395e-05,
+      "loss": 1.0383,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582458291091664e-05,
+      "loss": 1.0132,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580904026243322e-05,
+      "loss": 0.998,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579346935867743e-05,
+      "loss": 1.0615,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577787020424135e-05,
+      "loss": 1.0215,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576224280372537e-05,
+      "loss": 1.0881,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574658716173817e-05,
+      "loss": 0.9717,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573090328289675e-05,
+      "loss": 0.9746,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957151911718266e-05,
+      "loss": 0.6371,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569945083316133e-05,
+      "loss": 1.0498,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568368227154297e-05,
+      "loss": 0.9939,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566788549162188e-05,
+      "loss": 1.0442,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565206049805673e-05,
+      "loss": 1.0232,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563620729551448e-05,
+      "loss": 0.9666,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562032588867047e-05,
+      "loss": 0.6547,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956044162822083e-05,
+      "loss": 0.9583,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955884784808199e-05,
+      "loss": 0.9744,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557251248920556e-05,
+      "loss": 1.0266,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555651831207383e-05,
+      "loss": 1.0457,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955404959541416e-05,
+      "loss": 0.6693,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95524445420134e-05,
+      "loss": 1.031,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550836671478465e-05,
+      "loss": 0.9729,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549225984283528e-05,
+      "loss": 1.0889,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547612480903598e-05,
+      "loss": 1.0347,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545996161814526e-05,
+      "loss": 1.0007,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544377027492974e-05,
+      "loss": 0.9492,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954275507841646e-05,
+      "loss": 0.9758,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541130315063297e-05,
+      "loss": 1.0525,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539502737912664e-05,
+      "loss": 1.0325,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537872347444544e-05,
+      "loss": 1.0085,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536239144139767e-05,
+      "loss": 1.0378,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534603128479976e-05,
+      "loss": 0.9954,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953296430094766e-05,
+      "loss": 0.9688,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953132266202613e-05,
+      "loss": 0.9485,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529678212199517e-05,
+      "loss": 1.0757,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528030951952796e-05,
+      "loss": 1.0024,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526380881771766e-05,
+      "loss": 1.0093,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952472800214305e-05,
+      "loss": 1.0164,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95230723135541e-05,
+      "loss": 0.9624,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521413816493206e-05,
+      "loss": 1.0183,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519752511449476e-05,
+      "loss": 0.6243,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518088398912846e-05,
+      "loss": 1.0034,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516421479374088e-05,
+      "loss": 1.0068,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514751753324795e-05,
+      "loss": 0.9854,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951307922125739e-05,
+      "loss": 1.0117,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511403883665126e-05,
+      "loss": 1.0359,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509725741042077e-05,
+      "loss": 0.9727,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508044793883152e-05,
+      "loss": 1.0845,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506361042684077e-05,
+      "loss": 1.0117,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950467448794142e-05,
+      "loss": 1.0447,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502985130152557e-05,
+      "loss": 0.6653,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950129296981571e-05,
+      "loss": 0.9946,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949959800742991e-05,
+      "loss": 0.9753,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949790024349503e-05,
+      "loss": 1.0417,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949619967851175e-05,
+      "loss": 0.9863,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94944963129816e-05,
+      "loss": 0.6669,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492790147406918e-05,
+      "loss": 0.9617,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491081182290877e-05,
+      "loss": 0.9462,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489369418137464e-05,
+      "loss": 0.9814,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487654855451507e-05,
+      "loss": 1.0291,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948593749473865e-05,
+      "loss": 0.665,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484217336505362e-05,
+      "loss": 0.9932,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948249438125894e-05,
+      "loss": 0.936,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480768629507508e-05,
+      "loss": 1.0352,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479040081760005e-05,
+      "loss": 1.0344,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477308738526207e-05,
+      "loss": 1.0393,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475574600316704e-05,
+      "loss": 0.9827,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947383766764292e-05,
+      "loss": 0.9673,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472097941017092e-05,
+      "loss": 1.033,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947035542095229e-05,
+      "loss": 1.0227,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468610107962405e-05,
+      "loss": 0.988,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466862002562154e-05,
+      "loss": 1.0098,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946511110526707e-05,
+      "loss": 1.0371,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463357416593517e-05,
+      "loss": 0.8904,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946160093705868e-05,
+      "loss": 0.978,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945984166718056e-05,
+      "loss": 0.9949,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458079607478e-05,
+      "loss": 0.9993,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456314758470642e-05,
+      "loss": 0.9636,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454547120678966e-05,
+      "loss": 1.0083,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945277669462427e-05,
+      "loss": 0.7117,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451003480828677e-05,
+      "loss": 1.0208,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449227479815126e-05,
+      "loss": 0.9846,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447448692107383e-05,
+      "loss": 1.0281,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445667118230037e-05,
+      "loss": 0.9919,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944388275870849e-05,
+      "loss": 0.9863,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442095614068977e-05,
+      "loss": 0.991,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944030568483855e-05,
+      "loss": 0.6616,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943851297154508e-05,
+      "loss": 1.0693,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436717474717253e-05,
+      "loss": 0.9961,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434919194884593e-05,
+      "loss": 1.0281,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433118132577432e-05,
+      "loss": 0.6518,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431314288326925e-05,
+      "loss": 1.0444,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942950766266505e-05,
+      "loss": 1.0532,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427698256124598e-05,
+      "loss": 1.0684,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425886069239192e-05,
+      "loss": 0.6542,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424071102543262e-05,
+      "loss": 1.0178,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942225335657207e-05,
+      "loss": 1.052,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420432831861688e-05,
+      "loss": 0.9707,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941860952894901e-05,
+      "loss": 1.0085,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416783448371758e-05,
+      "loss": 0.9976,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941495459066846e-05,
+      "loss": 0.9639,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413122956378466e-05,
+      "loss": 1.0029,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941128854604195e-05,
+      "loss": 0.679,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409451360199907e-05,
+      "loss": 0.9294,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407611399394145e-05,
+      "loss": 1.0166,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405768664167284e-05,
+      "loss": 1.0027,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403923155062776e-05,
+      "loss": 0.7063,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402074872624883e-05,
+      "loss": 1.0098,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400223817398684e-05,
+      "loss": 1.0422,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939836998993008e-05,
+      "loss": 1.0225,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939651339076579e-05,
+      "loss": 0.653,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939465402045335e-05,
+      "loss": 0.9941,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392791879541096e-05,
+      "loss": 1.0791,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390926968578213e-05,
+      "loss": 0.9626,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389059288114684e-05,
+      "loss": 1.0791,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387188838701304e-05,
+      "loss": 1.03,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93853156208897e-05,
+      "loss": 1.0571,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383439635232296e-05,
+      "loss": 0.9875,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381560882282354e-05,
+      "loss": 0.9497,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379679362593937e-05,
+      "loss": 0.9802,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937779507672193e-05,
+      "loss": 1.0083,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375908025222033e-05,
+      "loss": 1.0703,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374018208650755e-05,
+      "loss": 0.681,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372125627565435e-05,
+      "loss": 1.0452,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937023028252421e-05,
+      "loss": 0.9993,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368332174086046e-05,
+      "loss": 1.0737,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936643130281072e-05,
+      "loss": 1.0234,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364527669258815e-05,
+      "loss": 1.0103,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362621273991747e-05,
+      "loss": 1.019,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936071211757173e-05,
+      "loss": 1.1392,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935880020056179e-05,
+      "loss": 0.9172,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935688552352579e-05,
+      "loss": 0.9685,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354968087028383e-05,
+      "loss": 1.0581,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353047891635044e-05,
+      "loss": 0.9617,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351124937912065e-05,
+      "loss": 0.9778,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349199226426552e-05,
+      "loss": 0.9421,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347270757746416e-05,
+      "loss": 1.041,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345339532440387e-05,
+      "loss": 1.0503,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343405551078008e-05,
+      "loss": 0.9788,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341468814229635e-05,
+      "loss": 0.9663,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339529322466432e-05,
+      "loss": 0.9807,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337587076360387e-05,
+      "loss": 0.9634,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335642076484283e-05,
+      "loss": 0.6792,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333694323411732e-05,
+      "loss": 0.9554,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933174381771715e-05,
+      "loss": 1.0037,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932979055997576e-05,
+      "loss": 0.9763,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327834550763605e-05,
+      "loss": 0.9382,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325875790657537e-05,
+      "loss": 0.9983,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323914280235218e-05,
+      "loss": 0.9829,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321950020075117e-05,
+      "loss": 0.9949,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319983010756527e-05,
+      "loss": 0.981,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931801325285954e-05,
+      "loss": 1.0427,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931604074696506e-05,
+      "loss": 1.0039,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314065493654806e-05,
+      "loss": 1.0161,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312087493511304e-05,
+      "loss": 0.9753,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931010674711789e-05,
+      "loss": 1.0566,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930812325505871e-05,
+      "loss": 0.9941,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930613701791872e-05,
+      "loss": 1.0183,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930414803628369e-05,
+      "loss": 1.0144,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930215631074019e-05,
+      "loss": 1.0664,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930016184187561e-05,
+      "loss": 1.0671,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298164630278137e-05,
+      "loss": 0.682,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929616467653678e-05,
+      "loss": 0.9827,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929416198124135e-05,
+      "loss": 0.9648,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292156544982456e-05,
+      "loss": 0.98,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929014836835154e-05,
+      "loss": 0.3102,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928813745194083e-05,
+      "loss": 0.9829,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286123796343373e-05,
+      "loss": 0.9492,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284107402153022e-05,
+      "loss": 0.8875,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928208826996443e-05,
+      "loss": 1.0027,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280066400373073e-05,
+      "loss": 0.9905,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278041793975216e-05,
+      "loss": 0.6597,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927601445136795e-05,
+      "loss": 1.0132,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927398437314916e-05,
+      "loss": 0.9672,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271951559917537e-05,
+      "loss": 0.957,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269916012272586e-05,
+      "loss": 1.0803,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267877730814615e-05,
+      "loss": 0.6747,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265836716144744e-05,
+      "loss": 1.0332,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263792968864882e-05,
+      "loss": 0.9956,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261746489577767e-05,
+      "loss": 1.0776,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259697278886926e-05,
+      "loss": 0.9397,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257645337396692e-05,
+      "loss": 0.9152,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255590665712214e-05,
+      "loss": 0.9634,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253533264439436e-05,
+      "loss": 1.0269,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251473134185116e-05,
+      "loss": 1.0386,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924941027555681e-05,
+      "loss": 0.9717,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924734468916288e-05,
+      "loss": 0.9963,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245276375612493e-05,
+      "loss": 1.0317,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924320533551562e-05,
+      "loss": 0.9929,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241131569483035e-05,
+      "loss": 1.0576,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923905507812632e-05,
+      "loss": 1.0181,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923697586205786e-05,
+      "loss": 0.9656,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234893921890834e-05,
+      "loss": 0.9255,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923280925823924e-05,
+      "loss": 0.9463,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230721871717862e-05,
+      "loss": 1.0264,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228631762942307e-05,
+      "loss": 0.9856,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226538932528967e-05,
+      "loss": 0.9673,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224443381095047e-05,
+      "loss": 1.0115,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222345109258546e-05,
+      "loss": 0.9739,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220244117638277e-05,
+      "loss": 1.0496,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218140406853846e-05,
+      "loss": 1.0481,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216033977525658e-05,
+      "loss": 0.9939,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213924830274934e-05,
+      "loss": 0.6902,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921181296572368e-05,
+      "loss": 1.05,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209698384494714e-05,
+      "loss": 1.0479,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207581087211654e-05,
+      "loss": 0.9465,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205461074498917e-05,
+      "loss": 0.9861,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203338346981718e-05,
+      "loss": 1.0188,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201212905286074e-05,
+      "loss": 0.9932,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919908475003881e-05,
+      "loss": 1.0095,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196953881867544e-05,
+      "loss": 1.0095,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194820301400695e-05,
+      "loss": 0.9543,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919268400926748e-05,
+      "loss": 0.9617,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190545006097917e-05,
+      "loss": 0.6991,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918840329252283e-05,
+      "loss": 0.9373,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186258869173834e-05,
+      "loss": 0.957,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184111736683343e-05,
+      "loss": 1.0459,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918196189568458e-05,
+      "loss": 1.04,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917980934681155e-05,
+      "loss": 0.9824,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177654090699074e-05,
+      "loss": 0.7065,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175496127982764e-05,
+      "loss": 0.9976,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173335459299025e-05,
+      "loss": 0.9858,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171172085285064e-05,
+      "loss": 0.6492,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169006006578897e-05,
+      "loss": 1.0271,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166837223819317e-05,
+      "loss": 1.009,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164665737645927e-05,
+      "loss": 0.9741,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162491548699128e-05,
+      "loss": 1.0642,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160314657620116e-05,
+      "loss": 1.0559,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158135065050878e-05,
+      "loss": 1.0669,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155952771634207e-05,
+      "loss": 0.9617,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153767778013686e-05,
+      "loss": 1.0886,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91515800848337e-05,
+      "loss": 0.9941,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149389692739424e-05,
+      "loss": 1.0061,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147196602376833e-05,
+      "loss": 0.9932,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145000814392696e-05,
+      "loss": 1.0234,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142802329434574e-05,
+      "loss": 0.9971,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140601148150833e-05,
+      "loss": 0.6814,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913839727119063e-05,
+      "loss": 0.6815,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136190699203913e-05,
+      "loss": 0.9019,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133981432841424e-05,
+      "loss": 1.0042,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913176947275471e-05,
+      "loss": 0.9692,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129554819596098e-05,
+      "loss": 0.9698,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912733747401872e-05,
+      "loss": 0.9746,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91251174366765e-05,
+      "loss": 1.0129,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122894708224152e-05,
+      "loss": 0.929,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912066928931719e-05,
+      "loss": 1.0132,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118441180611912e-05,
+      "loss": 0.9563,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911621038276542e-05,
+      "loss": 0.9512,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113976896435603e-05,
+      "loss": 0.9771,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111740722281137e-05,
+      "loss": 1.002,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910950186096151e-05,
+      "loss": 0.9562,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910726031313698e-05,
+      "loss": 0.9924,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910501607946861e-05,
+      "loss": 1.0107,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910276916061825e-05,
+      "loss": 1.0122,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910051955724855e-05,
+      "loss": 0.9441,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098267270022943e-05,
+      "loss": 0.97,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096012299605654e-05,
+      "loss": 0.9807,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90937546466617e-05,
+      "loss": 0.9873,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091494311856903e-05,
+      "loss": 0.947,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089231295857846e-05,
+      "loss": 0.9502,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086965599331938e-05,
+      "loss": 0.6799,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084697222947344e-05,
+      "loss": 0.9924,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082426167373053e-05,
+      "loss": 0.9346,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080152433278816e-05,
+      "loss": 1.0027,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077876021335188e-05,
+      "loss": 1.0432,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907559693221351e-05,
+      "loss": 1.0554,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073315166585916e-05,
+      "loss": 0.9641,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907103072512533e-05,
+      "loss": 0.9692,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068743608505454e-05,
+      "loss": 1.0259,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066453817400793e-05,
+      "loss": 0.9319,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064161352486638e-05,
+      "loss": 0.958,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061866214439054e-05,
+      "loss": 1.0176,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905956840393491e-05,
+      "loss": 0.9502,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057267921651865e-05,
+      "loss": 0.6644,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054964768268354e-05,
+      "loss": 1.0198,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905265894446361e-05,
+      "loss": 0.6602,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905035045091764e-05,
+      "loss": 1.0149,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048039288311256e-05,
+      "loss": 0.9829,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904572545732605e-05,
+      "loss": 0.9617,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904340895864439e-05,
+      "loss": 0.7117,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041089792949446e-05,
+      "loss": 1.0449,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038767960925168e-05,
+      "loss": 1.0144,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036443463256295e-05,
+      "loss": 0.9836,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034116300628346e-05,
+      "loss": 0.988,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031786473727633e-05,
+      "loss": 1.0264,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902945398324125e-05,
+      "loss": 0.9976,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902711882985708e-05,
+      "loss": 1.0105,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902478101426378e-05,
+      "loss": 1.041,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902244053715081e-05,
+      "loss": 0.9707,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020097399208403e-05,
+      "loss": 0.9661,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017751601127575e-05,
+      "loss": 1.0227,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901540314360014e-05,
+      "loss": 0.9714,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013052027318677e-05,
+      "loss": 0.9868,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010698252976566e-05,
+      "loss": 1.031,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008341821267965e-05,
+      "loss": 1.0288,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005982732887812e-05,
+      "loss": 0.9812,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003620988531827e-05,
+      "loss": 1.0129,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001256588896526e-05,
+      "loss": 0.8926,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998889534679196e-05,
+      "loss": 0.9539,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996519826577907e-05,
+      "loss": 0.908,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994147465291526e-05,
+      "loss": 0.9404,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991772451519678e-05,
+      "loss": 1.0454,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989394785962795e-05,
+      "loss": 0.9976,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987014469322075e-05,
+      "loss": 0.9707,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89846315022995e-05,
+      "loss": 0.9973,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898224588559784e-05,
+      "loss": 0.969,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979857619920646e-05,
+      "loss": 0.9956,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977466705972244e-05,
+      "loss": 0.9766,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975073144457742e-05,
+      "loss": 1.0273,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972676936083035e-05,
+      "loss": 1.0227,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970278081554792e-05,
+      "loss": 1.001,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967876581580468e-05,
+      "loss": 0.9387,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965472436868288e-05,
+      "loss": 0.9841,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896306564812727e-05,
+      "loss": 1.0269,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960656216067204e-05,
+      "loss": 1.02,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958244141398664e-05,
+      "loss": 0.6042,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955829424832995e-05,
+      "loss": 0.907,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895341206708233e-05,
+      "loss": 0.9761,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950992068859584e-05,
+      "loss": 0.9653,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948569430878435e-05,
+      "loss": 1.0779,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894614415385335e-05,
+      "loss": 0.9917,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943716238499575e-05,
+      "loss": 1.0186,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941285685533136e-05,
+      "loss": 1.0308,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893885249567083e-05,
+      "loss": 0.9688,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936416669630236e-05,
+      "loss": 0.9714,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933978208129705e-05,
+      "loss": 1.0117,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931537111888378e-05,
+      "loss": 0.6435,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929093381626153e-05,
+      "loss": 0.9929,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926647018063724e-05,
+      "loss": 0.9634,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924198021922553e-05,
+      "loss": 1.0471,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921746393924877e-05,
+      "loss": 1.0725,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919292134793718e-05,
+      "loss": 0.9382,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916835245252854e-05,
+      "loss": 1.0237,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891437572602687e-05,
+      "loss": 0.9204,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911913577841096e-05,
+      "loss": 0.6614,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909448801421653e-05,
+      "loss": 0.6781,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906981397495434e-05,
+      "loss": 0.9861,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904511366790106e-05,
+      "loss": 1.051,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902038710034113e-05,
+      "loss": 1.0288,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899563427956674e-05,
+      "loss": 0.9592,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889708552128778e-05,
+      "loss": 1.0056,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889460499075819e-05,
+      "loss": 1.0278,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892121837099452e-05,
+      "loss": 1.0076,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889636061043874e-05,
+      "loss": 1.0115,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887147663324545e-05,
+      "loss": 1.0493,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884656644675318e-05,
+      "loss": 0.9299,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888216300583083e-05,
+      "loss": 1.0151,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887966674752649e-05,
+      "loss": 0.9729,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877167870498466e-05,
+      "loss": 0.9958,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874666375483713e-05,
+      "loss": 1.0154,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872162263219955e-05,
+      "loss": 0.946,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886965553444568e-05,
+      "loss": 1.0337,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886714618990016e-05,
+      "loss": 1.0024,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864634230323426e-05,
+      "loss": 1.0054,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886211965645629e-05,
+      "loss": 0.9209,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859602469040327e-05,
+      "loss": 1.0071,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885708266881789e-05,
+      "loss": 1.0566,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 1.0002,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852035232926848e-05,
+      "loss": 0.981,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884950759874679e-05,
+      "loss": 0.991,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846977354737356e-05,
+      "loss": 0.9814,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844444501644755e-05,
+      "loss": 1.0085,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841909040215947e-05,
+      "loss": 0.957,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883937097119868e-05,
+      "loss": 0.9448,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883683029534145e-05,
+      "loss": 0.9896,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834287013393547e-05,
+      "loss": 0.9712,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883174112610501e-05,
+      "loss": 1.0173,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882919263422665e-05,
+      "loss": 0.9644,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826641538510055e-05,
+      "loss": 0.9563,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824087839707572e-05,
+      "loss": 0.9612,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821531538572317e-05,
+      "loss": 0.9624,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818972635858174e-05,
+      "loss": 0.9722,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88164111323198e-05,
+      "loss": 1.033,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813847028712612e-05,
+      "loss": 0.6619,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811280325792788e-05,
+      "loss": 1.0093,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808711024317295e-05,
+      "loss": 0.9683,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880613912504384e-05,
+      "loss": 0.6665,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803564628730916e-05,
+      "loss": 0.6571,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800987536137767e-05,
+      "loss": 1.0879,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798407848024415e-05,
+      "loss": 0.9714,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795825565151637e-05,
+      "loss": 0.9722,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793240688280986e-05,
+      "loss": 1.0156,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790653218174766e-05,
+      "loss": 0.9905,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788063155596062e-05,
+      "loss": 1.0,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785470501308714e-05,
+      "loss": 0.9613,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782875256077322e-05,
+      "loss": 0.9844,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780277420667262e-05,
+      "loss": 0.9805,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877767699584467e-05,
+      "loss": 0.9854,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775073982376434e-05,
+      "loss": 0.9932,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772468381030222e-05,
+      "loss": 0.9319,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876986019257446e-05,
+      "loss": 0.9905,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876724941777833e-05,
+      "loss": 1.0186,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876463605741178e-05,
+      "loss": 1.0581,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876202011224553e-05,
+      "loss": 0.9871,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759401583051048e-05,
+      "loss": 0.71,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756780470600574e-05,
+      "loss": 0.9429,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754156775667108e-05,
+      "loss": 0.9016,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751530499024403e-05,
+      "loss": 0.6307,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748901641446987e-05,
+      "loss": 0.9565,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874627020371014e-05,
+      "loss": 0.6489,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874363618658991e-05,
+      "loss": 0.98,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87409995908631e-05,
+      "loss": 1.0654,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738360417307265e-05,
+      "loss": 0.967,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873571866670074e-05,
+      "loss": 1.0942,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873307433982261e-05,
+      "loss": 0.9272,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730427437452717e-05,
+      "loss": 0.9534,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727777960371662e-05,
+      "loss": 0.9653,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725125909360816e-05,
+      "loss": 0.97,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722471285202295e-05,
+      "loss": 0.9888,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871981408867898e-05,
+      "loss": 0.9717,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717154320574522e-05,
+      "loss": 1.0127,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714491981673307e-05,
+      "loss": 0.9951,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87118270727605e-05,
+      "loss": 0.6757,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870915959462201e-05,
+      "loss": 0.9867,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706489548044508e-05,
+      "loss": 0.9973,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703816933815433e-05,
+      "loss": 0.9904,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701141752722966e-05,
+      "loss": 0.6289,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698464005556047e-05,
+      "loss": 1.0154,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695783693104386e-05,
+      "loss": 0.7023,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693100816158436e-05,
+      "loss": 0.9819,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869041537550941e-05,
+      "loss": 0.6249,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687727371949275e-05,
+      "loss": 1.0012,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868503680627076e-05,
+      "loss": 0.9937,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868234367926735e-05,
+      "loss": 0.9973,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679647991733268e-05,
+      "loss": 0.9587,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676949744463523e-05,
+      "loss": 0.6292,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867424893825385e-05,
+      "loss": 1.0627,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671545573900755e-05,
+      "loss": 0.9077,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866883965220149e-05,
+      "loss": 0.7141,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866613117395407e-05,
+      "loss": 0.9487,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663420139957252e-05,
+      "loss": 0.9763,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866070655101056e-05,
+      "loss": 0.9241,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657990407914265e-05,
+      "loss": 0.9937,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865527171146938e-05,
+      "loss": 1.0117,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652550462477697e-05,
+      "loss": 0.6688,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864982666174174e-05,
+      "loss": 0.6713,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864710031006479e-05,
+      "loss": 1.0029,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864437140825088e-05,
+      "loss": 0.9246,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641639957104805e-05,
+      "loss": 0.9353,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638905957432097e-05,
+      "loss": 1.022,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636169410039046e-05,
+      "loss": 1.0134,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86334303157327e-05,
+      "loss": 0.9951,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630688675320844e-05,
+      "loss": 0.9539,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627944489612025e-05,
+      "loss": 1.0044,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862519775941554e-05,
+      "loss": 1.0596,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862244848554143e-05,
+      "loss": 0.9956,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619696668800494e-05,
+      "loss": 1.033,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616942310004273e-05,
+      "loss": 0.9536,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614185409965062e-05,
+      "loss": 0.9341,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611425969495908e-05,
+      "loss": 0.9541,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86086639894106e-05,
+      "loss": 1.0388,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8605899470523683e-05,
+      "loss": 1.0142,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603132413650443e-05,
+      "loss": 0.9758,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600362819606927e-05,
+      "loss": 0.6533,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859759068920992e-05,
+      "loss": 0.9167,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594816023276954e-05,
+      "loss": 0.9731,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592038822626314e-05,
+      "loss": 0.9827,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589259088077033e-05,
+      "loss": 0.9832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858647682044889e-05,
+      "loss": 0.9868,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583692020562404e-05,
+      "loss": 0.9927,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580904689238855e-05,
+      "loss": 1.0232,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578114827300253e-05,
+      "loss": 0.9531,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575322435569368e-05,
+      "loss": 1.0222,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857252751486971e-05,
+      "loss": 1.0295,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856973006602554e-05,
+      "loss": 1.0112,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566930089861853e-05,
+      "loss": 0.9377,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85641275872044e-05,
+      "loss": 0.5848,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561322558879672e-05,
+      "loss": 1.0242,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855851500571491e-05,
+      "loss": 1.0364,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555704928538093e-05,
+      "loss": 0.6614,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552892328177953e-05,
+      "loss": 0.979,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550077205463953e-05,
+      "loss": 1.042,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547259561226313e-05,
+      "loss": 0.9814,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544439396295987e-05,
+      "loss": 0.9011,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541616711504676e-05,
+      "loss": 0.9922,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853879150768483e-05,
+      "loss": 0.6601,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535963785669637e-05,
+      "loss": 0.6467,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533133546293026e-05,
+      "loss": 1.0012,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530300790389663e-05,
+      "loss": 0.9951,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527465518794967e-05,
+      "loss": 0.9646,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85246277323451e-05,
+      "loss": 0.9617,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521787431876954e-05,
+      "loss": 0.9475,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851894461822817e-05,
+      "loss": 0.9275,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516099292237132e-05,
+      "loss": 1.0264,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513251454742957e-05,
+      "loss": 1.0012,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510401106585508e-05,
+      "loss": 1.0645,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507548248605393e-05,
+      "loss": 0.637,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850469288164395e-05,
+      "loss": 0.6043,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501835006543264e-05,
+      "loss": 1.0088,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498974624146164e-05,
+      "loss": 0.9707,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84961117352962e-05,
+      "loss": 0.9333,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849324634083768e-05,
+      "loss": 0.9521,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490378441615645e-05,
+      "loss": 0.9973,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487508038475873e-05,
+      "loss": 1.0107,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848463513226488e-05,
+      "loss": 0.9998,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481759723829922e-05,
+      "loss": 0.6742,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478881814019e-05,
+      "loss": 0.9797,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476001403680832e-05,
+      "loss": 0.6571,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473118493664897e-05,
+      "loss": 1.041,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470233084821393e-05,
+      "loss": 0.98,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467345178001272e-05,
+      "loss": 0.9873,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84644547740562e-05,
+      "loss": 0.926,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461561873838612e-05,
+      "loss": 1.0186,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458666478201643e-05,
+      "loss": 0.9567,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845576858799919e-05,
+      "loss": 0.9919,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452868204085873e-05,
+      "loss": 0.9827,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449965327317053e-05,
+      "loss": 0.9001,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447059958548822e-05,
+      "loss": 0.9885,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444152098638014e-05,
+      "loss": 1.0242,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441241748442186e-05,
+      "loss": 0.9949,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438328908819643e-05,
+      "loss": 1.0103,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435413580629412e-05,
+      "loss": 1.0105,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432495764731265e-05,
+      "loss": 1.019,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429575461985693e-05,
+      "loss": 1.0076,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426652673253943e-05,
+      "loss": 0.6392,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423727399397968e-05,
+      "loss": 0.9846,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420799641280478e-05,
+      "loss": 1.0071,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417869399764898e-05,
+      "loss": 0.9875,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414936675715395e-05,
+      "loss": 0.9539,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412001469996867e-05,
+      "loss": 0.9739,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840906378347494e-05,
+      "loss": 1.0356,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840612361701598e-05,
+      "loss": 0.9819,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403180971487067e-05,
+      "loss": 0.9771,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400235847756035e-05,
+      "loss": 0.6553,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839728824669143e-05,
+      "loss": 0.9712,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839433816916254e-05,
+      "loss": 1.0151,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839138561603938e-05,
+      "loss": 0.6757,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388430588192693e-05,
+      "loss": 0.9514,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838547308649395e-05,
+      "loss": 0.6671,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838251311181536e-05,
+      "loss": 0.9883,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379550665029856e-05,
+      "loss": 0.6414,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376585747011094e-05,
+      "loss": 0.9712,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837361835863347e-05,
+      "loss": 0.9568,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370648500772107e-05,
+      "loss": 1.0481,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367676174302848e-05,
+      "loss": 0.9954,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364701380102267e-05,
+      "loss": 0.9868,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361724119047675e-05,
+      "loss": 1.0322,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358744392017095e-05,
+      "loss": 0.9963,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835576219988929e-05,
+      "loss": 1.0208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352777543543745e-05,
+      "loss": 0.6047,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834979042386067e-05,
+      "loss": 0.7235,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346800841721006e-05,
+      "loss": 0.9739,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343808798006417e-05,
+      "loss": 0.9648,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340814293599293e-05,
+      "loss": 1.0256,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337817329382754e-05,
+      "loss": 1.0,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334817906240638e-05,
+      "loss": 0.9424,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331816025057508e-05,
+      "loss": 0.611,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328811686718667e-05,
+      "loss": 0.9788,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832580489211012e-05,
+      "loss": 1.0063,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322795642118618e-05,
+      "loss": 0.9551,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319783937631622e-05,
+      "loss": 1.0078,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316769779537316e-05,
+      "loss": 0.9644,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831375316872462e-05,
+      "loss": 0.999,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310734106083166e-05,
+      "loss": 0.9648,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830771259250331e-05,
+      "loss": 0.9915,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304688628876147e-05,
+      "loss": 0.9937,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301662216093464e-05,
+      "loss": 0.9304,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82986333550478e-05,
+      "loss": 0.957,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295602046632396e-05,
+      "loss": 0.9924,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292568291741228e-05,
+      "loss": 0.9924,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289532091268985e-05,
+      "loss": 1.0037,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286493446111077e-05,
+      "loss": 1.0088,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283452357163644e-05,
+      "loss": 0.6797,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828040882532354e-05,
+      "loss": 1.0117,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827736285148834e-05,
+      "loss": 0.998,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274314436556336e-05,
+      "loss": 0.9049,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827126358142654e-05,
+      "loss": 0.9565,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82682102869987e-05,
+      "loss": 0.9916,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265154554173257e-05,
+      "loss": 0.896,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262096383851396e-05,
+      "loss": 1.0085,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259035776935003e-05,
+      "loss": 0.9623,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255972734326685e-05,
+      "loss": 1.0127,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252907256929777e-05,
+      "loss": 1.0442,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249839345648324e-05,
+      "loss": 0.9536,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246769001387092e-05,
+      "loss": 0.9534,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243696225051564e-05,
+      "loss": 0.6079,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240621017547938e-05,
+      "loss": 0.9473,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237543379783133e-05,
+      "loss": 0.6567,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234463312664782e-05,
+      "loss": 0.9673,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231380817101234e-05,
+      "loss": 1.0027,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228295894001554e-05,
+      "loss": 0.3209,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225208544275526e-05,
+      "loss": 0.9968,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822211876883365e-05,
+      "loss": 1.01,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821902656858713e-05,
+      "loss": 0.9546,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215931944447904e-05,
+      "loss": 1.03,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212834897328614e-05,
+      "loss": 0.9612,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820973542814261e-05,
+      "loss": 0.989,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820663353780397e-05,
+      "loss": 0.6595,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820352922722748e-05,
+      "loss": 0.9243,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200422497328635e-05,
+      "loss": 1.0215,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197313349023655e-05,
+      "loss": 0.9773,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819420178322946e-05,
+      "loss": 0.9797,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191087800863692e-05,
+      "loss": 0.957,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187971402844706e-05,
+      "loss": 0.9375,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818485259009156e-05,
+      "loss": 0.9631,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818173136352403e-05,
+      "loss": 0.9714,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178607724062613e-05,
+      "loss": 0.6624,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175481672628505e-05,
+      "loss": 0.9548,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172353210143613e-05,
+      "loss": 0.9949,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169222337530567e-05,
+      "loss": 1.0654,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816608905571269e-05,
+      "loss": 1.0071,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162953365614037e-05,
+      "loss": 0.9688,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159815268159348e-05,
+      "loss": 0.9727,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156674764274102e-05,
+      "loss": 0.9573,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153531854884465e-05,
+      "loss": 1.0125,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815038654091732e-05,
+      "loss": 0.9661,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814723882330026e-05,
+      "loss": 0.9648,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814408870296158e-05,
+      "loss": 1.0696,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81409361808303e-05,
+      "loss": 0.9829,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813778125783613e-05,
+      "loss": 1.0244,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81346239349095e-05,
+      "loss": 1.0732,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813146421298154e-05,
+      "loss": 0.9587,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128302092984095e-05,
+      "loss": 0.9724,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812513757584971e-05,
+      "loss": 0.9341,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121970662511646e-05,
+      "loss": 0.9868,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118801353903856e-05,
+      "loss": 1.0386,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115629650961014e-05,
+      "loss": 1.042,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112455554618495e-05,
+      "loss": 0.9795,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810927906581238e-05,
+      "loss": 0.9871,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106100185479446e-05,
+      "loss": 1.0974,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102918914557196e-05,
+      "loss": 0.936,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809973525398382e-05,
+      "loss": 0.6401,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096549204698223e-05,
+      "loss": 0.9841,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093360767640003e-05,
+      "loss": 0.9956,
+      "step": 1335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0334,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808697673396765e-05,
+      "loss": 0.9988,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808378113923625e-05,
+      "loss": 0.9285,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080583160497695e-05,
+      "loss": 0.9346,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80773827986951e-05,
+      "loss": 0.9731,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80741800547723e-05,
+      "loss": 0.9749,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8070974929673818e-05,
+      "loss": 1.0474,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806776742434489e-05,
+      "loss": 1.0581,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064557539731446e-05,
+      "loss": 0.9536,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061345276780124e-05,
+      "loss": 1.0193,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805813063643825e-05,
+      "loss": 0.969,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805491361965388e-05,
+      "loss": 0.9246,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051694227375734e-05,
+      "loss": 0.97,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804847246055326e-05,
+      "loss": 0.9639,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045248320136593e-05,
+      "loss": 0.9585,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042021807076575e-05,
+      "loss": 0.9631,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038792922324743e-05,
+      "loss": 0.9629,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035561666833336e-05,
+      "loss": 1.0991,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803232804155529e-05,
+      "loss": 1.0286,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029092047444238e-05,
+      "loss": 0.9692,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025853685454522e-05,
+      "loss": 0.9941,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022612956541174e-05,
+      "loss": 1.0366,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801936986165992e-05,
+      "loss": 0.9526,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801612440176719e-05,
+      "loss": 0.9849,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012876577820113e-05,
+      "loss": 0.9512,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800962639077651e-05,
+      "loss": 0.9529,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006373841594905e-05,
+      "loss": 0.916,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800311893123451e-05,
+      "loss": 0.991,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999861660655247e-05,
+      "loss": 1.0134,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996602030817713e-05,
+      "loss": 0.6909,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993340042683222e-05,
+      "loss": 1.0457,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990075697213775e-05,
+      "loss": 1.0278,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798680899537206e-05,
+      "loss": 1.031,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983539938121476e-05,
+      "loss": 0.7036,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980268526426105e-05,
+      "loss": 0.9739,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976994761250727e-05,
+      "loss": 1.001,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973718643560817e-05,
+      "loss": 0.9187,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797044017432254e-05,
+      "loss": 0.6151,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967159354502764e-05,
+      "loss": 0.9812,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963876185069032e-05,
+      "loss": 1.0168,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960590666989604e-05,
+      "loss": 1.0039,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795730280123341e-05,
+      "loss": 0.9658,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795401258877009e-05,
+      "loss": 0.9553,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795072003056996e-05,
+      "loss": 1.0608,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947425127604045e-05,
+      "loss": 0.9897,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944127880844043e-05,
+      "loss": 0.9521,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794082829126236e-05,
+      "loss": 0.7233,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937526359832088e-05,
+      "loss": 0.9563,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934222087526996e-05,
+      "loss": 0.9629,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793091547532157e-05,
+      "loss": 0.9695,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792760652419096e-05,
+      "loss": 0.9592,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792429523511102e-05,
+      "loss": 1.0383,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792098160905829e-05,
+      "loss": 0.958,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917665647010003e-05,
+      "loss": 0.9883,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791434734994407e-05,
+      "loss": 0.603,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79110267188391e-05,
+      "loss": 0.9381,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907703754674392e-05,
+      "loss": 1.0173,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790437845842993e-05,
+      "loss": 0.9917,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901050831086377e-05,
+      "loss": 0.9434,
+      "step": 1394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78977208736251e-05,
+      "loss": 0.9082,
+      "step": 1395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789438858702814e-05,
+      "loss": 0.6505,
+      "step": 1396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789105397227823e-05,
+      "loss": 0.9946,
+      "step": 1397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887717030358794e-05,
+      "loss": 0.9946,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788437776225393e-05,
+      "loss": 0.9944,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881036168948435e-05,
+      "loss": 1.0413,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877692251427783e-05,
+      "loss": 0.9346,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787434601067814e-05,
+      "loss": 0.9744,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787099744768635e-05,
+      "loss": 0.9558,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867646563439947e-05,
+      "loss": 0.9805,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786429335892715e-05,
+      "loss": 0.6695,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786093783513686e-05,
+      "loss": 0.9539,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857579993058662e-05,
+      "loss": 0.6486,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854219833682825e-05,
+      "loss": 0.9973,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78508573580003e-05,
+      "loss": 0.9883,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847492567002723e-05,
+      "loss": 0.9443,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784412546168242e-05,
+      "loss": 1.0186,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784075604303238e-05,
+      "loss": 0.9985,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837384312046297e-05,
+      "loss": 0.9492,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834010269718526e-05,
+      "loss": 1.0015,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830633917044125e-05,
+      "loss": 0.927,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827255255018813e-05,
+      "loss": 0.9978,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823874284639004e-05,
+      "loss": 0.9807,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782049100690179e-05,
+      "loss": 0.9543,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781710542280493e-05,
+      "loss": 1.0496,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781371753334689e-05,
+      "loss": 1.0242,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810327339526792e-05,
+      "loss": 0.9836,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806934842344444e-05,
+      "loss": 0.9524,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803540042800347e-05,
+      "loss": 1.0078,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800142941895658e-05,
+      "loss": 0.9482,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796743540632226e-05,
+      "loss": 0.9907,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779334184001258e-05,
+      "loss": 1.0256,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778993784103992e-05,
+      "loss": 0.6733,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778653154471813e-05,
+      "loss": 0.9976,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778312295205177e-05,
+      "loss": 1.0103,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779712064046073e-05,
+      "loss": 0.9585,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776298881706955e-05,
+      "loss": 1.0129,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772883406041002e-05,
+      "loss": 0.9419,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769465638055483e-05,
+      "loss": 0.6802,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776604557875834e-05,
+      "loss": 0.9724,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762623229158192e-05,
+      "loss": 0.8453,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759198590264327e-05,
+      "loss": 0.9587,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775577166308672e-05,
+      "loss": 1.0112,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775234244863601e-05,
+      "loss": 1.0359,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748910947923515e-05,
+      "loss": 0.9851,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774547716196123e-05,
+      "loss": 0.6084,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742041091761823e-05,
+      "loss": 0.9128,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738602738338625e-05,
+      "loss": 0.9771,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773516210270565e-05,
+      "loss": 1.0046,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731719185877593e-05,
+      "loss": 0.9546,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728273988869807e-05,
+      "loss": 1.0239,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724826512698324e-05,
+      "loss": 1.0247,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721376758379847e-05,
+      "loss": 0.9355,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717924726931754e-05,
+      "loss": 0.9675,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714470419372086e-05,
+      "loss": 0.9797,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711013836719565e-05,
+      "loss": 0.9854,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770755497999358e-05,
+      "loss": 0.9618,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770409385021419e-05,
+      "loss": 0.9648,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700630448402125e-05,
+      "loss": 1.0085,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697164775578785e-05,
+      "loss": 0.9778,
+      "step": 1454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769369683276624e-05,
+      "loss": 0.9397,
+      "step": 1455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690226620987233e-05,
+      "loss": 1.012,
+      "step": 1456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686754141265162e-05,
+      "loss": 0.9402,
+      "step": 1457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768327939462412e-05,
+      "loss": 1.0059,
+      "step": 1458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679802382088836e-05,
+      "loss": 0.9919,
+      "step": 1459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676323104684736e-05,
+      "loss": 0.9924,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7672841563437897e-05,
+      "loss": 0.9695,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766935775937507e-05,
+      "loss": 0.9795,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766587169352367e-05,
+      "loss": 1.0193,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662383366911785e-05,
+      "loss": 0.9556,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765889278056816e-05,
+      "loss": 0.9758,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655399935522216e-05,
+      "loss": 0.9458,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651904832804038e-05,
+      "loss": 0.9343,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764840747344437e-05,
+      "loss": 0.7095,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764490785847463e-05,
+      "loss": 0.6396,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76414059889269e-05,
+      "loss": 0.9536,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637901865833923e-05,
+      "loss": 0.946,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76343954902291e-05,
+      "loss": 0.9697,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763088686314652e-05,
+      "loss": 0.9641,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627375985620902e-05,
+      "loss": 1.0015,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623862858687665e-05,
+      "loss": 1.0229,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762034748338286e-05,
+      "loss": 0.9966,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761682986074323e-05,
+      "loss": 1.082,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761330999180615e-05,
+      "loss": 0.9629,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609787877609678e-05,
+      "loss": 0.9866,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606263519192535e-05,
+      "loss": 0.9631,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760273691759409e-05,
+      "loss": 0.9888,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599208073854383e-05,
+      "loss": 0.6616,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759567698901412e-05,
+      "loss": 0.6143,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592143664114656e-05,
+      "loss": 0.938,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758860810019801e-05,
+      "loss": 0.9934,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758507029830687e-05,
+      "loss": 1.0144,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581530259484572e-05,
+      "loss": 0.9746,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577987984775125e-05,
+      "loss": 0.969,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574443475223183e-05,
+      "loss": 0.929,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757089673187406e-05,
+      "loss": 1.0068,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567347755773752e-05,
+      "loss": 0.9741,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756379654796888e-05,
+      "loss": 1.0078,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560243109506746e-05,
+      "loss": 0.9507,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556687441435302e-05,
+      "loss": 1.0132,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553129544803158e-05,
+      "loss": 0.9685,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754956942065958e-05,
+      "loss": 1.0139,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546007070054497e-05,
+      "loss": 0.9277,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754244249403849e-05,
+      "loss": 0.9431,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538875693662793e-05,
+      "loss": 0.9946,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535306669979305e-05,
+      "loss": 0.358,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753173542404057e-05,
+      "loss": 0.965,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752816195689979e-05,
+      "loss": 0.9536,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752458626961083e-05,
+      "loss": 0.9741,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521008363228208e-05,
+      "loss": 1.0527,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517428238807085e-05,
+      "loss": 0.9688,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751384589740329e-05,
+      "loss": 0.9998,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510261340073297e-05,
+      "loss": 0.9792,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506674567874234e-05,
+      "loss": 0.7233,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503085581863887e-05,
+      "loss": 1.0068,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499494383100697e-05,
+      "loss": 0.991,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495900972643745e-05,
+      "loss": 0.9585,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492305351552777e-05,
+      "loss": 0.9951,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488707520888183e-05,
+      "loss": 0.9458,
+      "step": 1513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.9861,
+      "step": 1514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481505235082958e-05,
+      "loss": 1.0322,
+      "step": 1515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477900782066365e-05,
+      "loss": 0.9834,
+      "step": 1516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747429412372424e-05,
+      "loss": 0.6582,
+      "step": 1517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747068526112022e-05,
+      "loss": 0.9915,
+      "step": 1518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746707419531861e-05,
+      "loss": 0.9353,
+      "step": 1519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463460927384356e-05,
+      "loss": 1.0286,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745984545838306e-05,
+      "loss": 1.0256,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745622778938096e-05,
+      "loss": 0.9587,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452607921444956e-05,
+      "loss": 0.9211,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744898585564259e-05,
+      "loss": 0.9958,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744536159304206e-05,
+      "loss": 0.9817,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74417351347122e-05,
+      "loss": 0.6416,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438106481722495e-05,
+      "loss": 1.0193,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743447563514309e-05,
+      "loss": 1.0518,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743084259604476e-05,
+      "loss": 1.0056,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742720736549893e-05,
+      "loss": 0.9819,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423569944577677e-05,
+      "loss": 0.6552,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419930334353728e-05,
+      "loss": 0.9675,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741628853590044e-05,
+      "loss": 1.0125,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741264455029183e-05,
+      "loss": 0.9978,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408998378602555e-05,
+      "loss": 0.652,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405350021907912e-05,
+      "loss": 0.9927,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401699481283847e-05,
+      "loss": 0.905,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739804675780695e-05,
+      "loss": 0.8979,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394391852554456e-05,
+      "loss": 1.0149,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390734766604244e-05,
+      "loss": 0.9634,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738707550103483e-05,
+      "loss": 0.9956,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738341405692538e-05,
+      "loss": 0.9446,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379750435355696e-05,
+      "loss": 1.0562,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376084637406222e-05,
+      "loss": 0.9934,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737241666415806e-05,
+      "loss": 0.9929,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368746516692928e-05,
+      "loss": 1.0471,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73650741960932e-05,
+      "loss": 0.9119,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361399703441893e-05,
+      "loss": 0.9783,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735772303982266e-05,
+      "loss": 0.6013,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735404420631979e-05,
+      "loss": 0.9717,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350363204018227e-05,
+      "loss": 0.931,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346680034003528e-05,
+      "loss": 1.0159,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734299469736192e-05,
+      "loss": 0.6837,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339307195180248e-05,
+      "loss": 0.9719,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335617528546e-05,
+      "loss": 0.9961,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733192569854731e-05,
+      "loss": 0.9563,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732823170627294e-05,
+      "loss": 0.9829,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324535552812298e-05,
+      "loss": 0.9414,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732083723925542e-05,
+      "loss": 0.9998,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731713676669299e-05,
+      "loss": 0.9956,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731343413621632e-05,
+      "loss": 0.9512,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309729348917364e-05,
+      "loss": 0.6614,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306022405888705e-05,
+      "loss": 0.9832,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302313308223574e-05,
+      "loss": 0.9822,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729860205701582e-05,
+      "loss": 1.002,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294888653359945e-05,
+      "loss": 1.0225,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291173098351077e-05,
+      "loss": 0.9099,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287455393084972e-05,
+      "loss": 0.9934,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283735538658034e-05,
+      "loss": 0.8889,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728001353616729e-05,
+      "loss": 1.0188,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727628938671041e-05,
+      "loss": 0.9814,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272563091385694e-05,
+      "loss": 0.9194,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268834651292062e-05,
+      "loss": 0.9727,
+      "step": 1573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726510406752908e-05,
+      "loss": 0.969,
+      "step": 1574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726137134119695e-05,
+      "loss": 0.6399,
+      "step": 1575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257636473396496e-05,
+      "loss": 0.9939,
+      "step": 1576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253899465229175e-05,
+      "loss": 0.9036,
+      "step": 1577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725016031779708e-05,
+      "loss": 0.9473,
+      "step": 1578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246419032202926e-05,
+      "loss": 0.6349,
+      "step": 1579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242675609550067e-05,
+      "loss": 0.9634,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723893005094249e-05,
+      "loss": 1.0535,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235182357484802e-05,
+      "loss": 1.0101,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231432530282246e-05,
+      "loss": 0.9985,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722768057044069e-05,
+      "loss": 0.6187,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223926479066636e-05,
+      "loss": 0.9529,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722017025726721e-05,
+      "loss": 0.6838,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721641190615017e-05,
+      "loss": 1.0312,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212651426823895e-05,
+      "loss": 0.927,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208888820397404e-05,
+      "loss": 0.9556,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205124087980334e-05,
+      "loss": 0.9423,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201357230682952e-05,
+      "loss": 0.9255,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197588249616147e-05,
+      "loss": 0.9814,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719381714589144e-05,
+      "loss": 1.0422,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190043920620982e-05,
+      "loss": 0.9729,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186268574917536e-05,
+      "loss": 0.988,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71824911098945e-05,
+      "loss": 0.9941,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178711526665895e-05,
+      "loss": 0.9702,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174929826346373e-05,
+      "loss": 0.6105,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171146010051196e-05,
+      "loss": 0.9849,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716736007889626e-05,
+      "loss": 1.0166,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716357203399809e-05,
+      "loss": 0.9558,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159781876473816e-05,
+      "loss": 1.0063,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715598960744121e-05,
+      "loss": 0.9915,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152195228018662e-05,
+      "loss": 0.7054,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714839873932517e-05,
+      "loss": 0.8691,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714460014248038e-05,
+      "loss": 0.9902,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140799438604533e-05,
+      "loss": 0.9587,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713699662881851e-05,
+      "loss": 0.9651,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133191714243805e-05,
+      "loss": 1.0002,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129384696002533e-05,
+      "loss": 0.9626,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712557557521743e-05,
+      "loss": 0.9951,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712176435301186e-05,
+      "loss": 0.9399,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117951030509795e-05,
+      "loss": 0.9451,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114135608835826e-05,
+      "loss": 0.9712,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110318089115178e-05,
+      "loss": 0.6808,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106498472473674e-05,
+      "loss": 0.9434,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710267676003778e-05,
+      "loss": 0.9331,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098852952934555e-05,
+      "loss": 0.9552,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709502705229169e-05,
+      "loss": 0.6664,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709119905923749e-05,
+      "loss": 0.9473,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087368974900876e-05,
+      "loss": 0.9119,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083536800411392e-05,
+      "loss": 0.9919,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079702536899193e-05,
+      "loss": 0.9888,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075866185495046e-05,
+      "loss": 1.0098,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072027747330348e-05,
+      "loss": 0.9618,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706818722353709e-05,
+      "loss": 0.9958,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70643446152479e-05,
+      "loss": 0.9358,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060499923596002e-05,
+      "loss": 0.6525,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705665314971525e-05,
+      "loss": 0.6471,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70528042947401e-05,
+      "loss": 0.9714,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048953359805637e-05,
+      "loss": 1.0125,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045100346047535e-05,
+      "loss": 0.9612,
+      "step": 1632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041245254602104e-05,
+      "loss": 0.9824,
+      "step": 1633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703738808660626e-05,
+      "loss": 0.8989,
+      "step": 1634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033528843197523e-05,
+      "loss": 0.5979,
+      "step": 1635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702966752551404e-05,
+      "loss": 0.917,
+      "step": 1636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025804134694552e-05,
+      "loss": 1.0002,
+      "step": 1637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021938671878427e-05,
+      "loss": 0.631,
+      "step": 1638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701807113820564e-05,
+      "loss": 0.6364,
+      "step": 1639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7014201534816765e-05,
+      "loss": 0.8943,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010329862853007e-05,
+      "loss": 0.9719,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700645612345616e-05,
+      "loss": 1.0159,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002580317768646e-05,
+      "loss": 0.9502,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699870244693348e-05,
+      "loss": 1.0349,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994822512094306e-05,
+      "loss": 0.9807,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990940514395353e-05,
+      "loss": 0.9624,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987056454981476e-05,
+      "loss": 0.9856,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698317033499813e-05,
+      "loss": 0.9512,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697928215559138e-05,
+      "loss": 1.021,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975391917907903e-05,
+      "loss": 0.3673,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971499623094972e-05,
+      "loss": 1.0129,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967605272300476e-05,
+      "loss": 0.9526,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963708866672907e-05,
+      "loss": 0.9468,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695981040736136e-05,
+      "loss": 1.0073,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955909895515544e-05,
+      "loss": 0.6198,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952007332285766e-05,
+      "loss": 0.9692,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694810271882294e-05,
+      "loss": 0.9495,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944196056278585e-05,
+      "loss": 1.0034,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940287345804828e-05,
+      "loss": 0.9453,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936376588554386e-05,
+      "loss": 0.6678,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69324637856806e-05,
+      "loss": 1.0063,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69285489383374e-05,
+      "loss": 0.9509,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692463204767932e-05,
+      "loss": 1.0146,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920713114861507e-05,
+      "loss": 1.0156,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69167921410397e-05,
+      "loss": 0.9634,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691286912737024e-05,
+      "loss": 0.9734,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908944075010067e-05,
+      "loss": 1.0198,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690501698511674e-05,
+      "loss": 0.9929,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69010878588484e-05,
+      "loss": 0.9624,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897156697363795e-05,
+      "loss": 1.0103,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689322350182227e-05,
+      "loss": 0.9875,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688928827338378e-05,
+      "loss": 0.9893,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885351013208874e-05,
+      "loss": 0.9387,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881411722458688e-05,
+      "loss": 1.0009,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877470402294978e-05,
+      "loss": 1.0095,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873527053880082e-05,
+      "loss": 0.9495,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869581678376944e-05,
+      "loss": 0.9919,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865634276949104e-05,
+      "loss": 0.6465,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68616848507607e-05,
+      "loss": 0.6549,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685773340097647e-05,
+      "loss": 0.6464,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853779928761734e-05,
+      "loss": 0.6383,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849824435282434e-05,
+      "loss": 0.9624,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845866921705086e-05,
+      "loss": 0.9849,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684190738919681e-05,
+      "loss": 0.9243,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683794583892533e-05,
+      "loss": 0.9241,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833982272058942e-05,
+      "loss": 0.9617,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683001668976656e-05,
+      "loss": 0.9424,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826049093217682e-05,
+      "loss": 0.9963,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68220794835824e-05,
+      "loss": 0.9861,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818107862031404e-05,
+      "loss": 0.9905,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681413422973597e-05,
+      "loss": 0.916,
+      "step": 1691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810158587867973e-05,
+      "loss": 0.9478,
+      "step": 1692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806180937599882e-05,
+      "loss": 1.0125,
+      "step": 1693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680220128010475e-05,
+      "loss": 0.9463,
+      "step": 1694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798219616556234e-05,
+      "loss": 0.9695,
+      "step": 1695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794235948128564e-05,
+      "loss": 0.9622,
+      "step": 1696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679025027599659e-05,
+      "loss": 0.9944,
+      "step": 1697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678626260133572e-05,
+      "loss": 1.0198,
+      "step": 1698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678227292532198e-05,
+      "loss": 0.662,
+      "step": 1699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778281249131973e-05,
+      "loss": 0.6519,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774287573942885e-05,
+      "loss": 0.8446,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770291900932513e-05,
+      "loss": 0.9385,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766294231279218e-05,
+      "loss": 0.9688,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762294566161976e-05,
+      "loss": 0.8955,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758292906760323e-05,
+      "loss": 1.0234,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675428925425441e-05,
+      "loss": 0.9492,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750283609824956e-05,
+      "loss": 1.0249,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746275974653278e-05,
+      "loss": 1.0085,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674226634992128e-05,
+      "loss": 1.001,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738254736811446e-05,
+      "loss": 1.0259,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673424113650685e-05,
+      "loss": 1.0098,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673022555019115e-05,
+      "loss": 0.9185,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726207979048604e-05,
+      "loss": 0.8878,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722188424264036e-05,
+      "loss": 0.9717,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718166887022856e-05,
+      "loss": 0.9717,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714143368511076e-05,
+      "loss": 0.9167,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710117869915272e-05,
+      "loss": 0.9985,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670609039242262e-05,
+      "loss": 1.0037,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702060937220873e-05,
+      "loss": 1.0132,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6698029505498364e-05,
+      "loss": 1.0498,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669399609844401e-05,
+      "loss": 0.9216,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668996071724732e-05,
+      "loss": 0.9722,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685923363098374e-05,
+      "loss": 0.926,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681884037187836e-05,
+      "loss": 1.0332,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677842740706958e-05,
+      "loss": 0.6253,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667379947484756e-05,
+      "loss": 0.9692,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666975424080206e-05,
+      "loss": 0.9695,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665707039763445e-05,
+      "loss": 0.9973,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666165787292528e-05,
+      "loss": 1.001,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657606741481723e-05,
+      "loss": 0.9292,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653553646627494e-05,
+      "loss": 0.9829,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649498589557908e-05,
+      "loss": 0.9861,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645441571468847e-05,
+      "loss": 0.9812,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641382593556776e-05,
+      "loss": 0.6356,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637321657018742e-05,
+      "loss": 0.9829,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663325876305236e-05,
+      "loss": 0.9946,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662919391285583e-05,
+      "loss": 0.9963,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625127107627928e-05,
+      "loss": 1.0125,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621058348568008e-05,
+      "loss": 0.9524,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616987636875987e-05,
+      "loss": 0.9265,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661291497375238e-05,
+      "loss": 0.6838,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608840360398257e-05,
+      "loss": 0.978,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660476379801528e-05,
+      "loss": 0.9614,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660068528780567e-05,
+      "loss": 0.8975,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596604830972234e-05,
+      "loss": 0.9824,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592522428718348e-05,
+      "loss": 0.9509,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588438082247966e-05,
+      "loss": 0.907,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584351792765612e-05,
+      "loss": 1.0051,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658026356147638e-05,
+      "loss": 0.989,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576173389585938e-05,
+      "loss": 0.9727,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657208127830053e-05,
+      "loss": 0.9412,
+      "step": 1751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656798722882698e-05,
+      "loss": 0.9343,
+      "step": 1752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656389124237266e-05,
+      "loss": 0.9771,
+      "step": 1753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559793320145536e-05,
+      "loss": 0.9119,
+      "step": 1754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555693463354133e-05,
+      "loss": 0.6458,
+      "step": 1755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655159167320755e-05,
+      "loss": 1.0566,
+      "step": 1756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547487950915458e-05,
+      "loss": 0.959,
+      "step": 1757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654338229768809e-05,
+      "loss": 0.9688,
+      "step": 1758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539274714736255e-05,
+      "loss": 0.9167,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535165203271334e-05,
+      "loss": 0.9822,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653105376450527e-05,
+      "loss": 0.9419,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652694039965058e-05,
+      "loss": 0.9238,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522825109920342e-05,
+      "loss": 0.9504,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651870789652821e-05,
+      "loss": 1.0439,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514588760688397e-05,
+      "loss": 0.7029,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510467703615694e-05,
+      "loss": 1.0088,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506344726525443e-05,
+      "loss": 0.9829,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502219830633567e-05,
+      "loss": 0.9895,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498093017156545e-05,
+      "loss": 1.03,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649396428731143e-05,
+      "loss": 1.0361,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489833642315834e-05,
+      "loss": 0.9556,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648570108338793e-05,
+      "loss": 0.9709,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648156661174647e-05,
+      "loss": 0.9446,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477430228610755e-05,
+      "loss": 0.707,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473291935200657e-05,
+      "loss": 0.979,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646915173273661e-05,
+      "loss": 0.9956,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465009622439607e-05,
+      "loss": 0.9971,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460865605531214e-05,
+      "loss": 0.9116,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645671968323355e-05,
+      "loss": 0.9937,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64525718567693e-05,
+      "loss": 0.9982,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448422127361707e-05,
+      "loss": 0.9736,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444270496234583e-05,
+      "loss": 0.9634,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440116964612287e-05,
+      "loss": 0.9548,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435961533719753e-05,
+      "loss": 0.97,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431804204782468e-05,
+      "loss": 1.0078,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427644979026473e-05,
+      "loss": 0.9683,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642348385767839e-05,
+      "loss": 0.98,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641932084196537e-05,
+      "loss": 0.9385,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641515593311515e-05,
+      "loss": 1.0032,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410989132356005e-05,
+      "loss": 0.6363,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406820440916778e-05,
+      "loss": 1.0029,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402649860026868e-05,
+      "loss": 0.9448,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398477390916235e-05,
+      "loss": 0.9512,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394303034815387e-05,
+      "loss": 0.8796,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639012679295539e-05,
+      "loss": 0.9668,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638594866656788e-05,
+      "loss": 0.8855,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638176865688503e-05,
+      "loss": 0.9507,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637758676513958e-05,
+      "loss": 0.9634,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637340299256482e-05,
+      "loss": 0.9595,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369217340394595e-05,
+      "loss": 0.9697,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365029809863308e-05,
+      "loss": 0.9844,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360840402205914e-05,
+      "loss": 0.99,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635664911865792e-05,
+      "loss": 0.9619,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352455960455385e-05,
+      "loss": 0.9932,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348260928834932e-05,
+      "loss": 0.9001,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634406402503372e-05,
+      "loss": 0.9658,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339865250289468e-05,
+      "loss": 1.0625,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335664605840446e-05,
+      "loss": 0.9578,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331462092925485e-05,
+      "loss": 0.9958,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327257712783945e-05,
+      "loss": 1.0237,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323051466655756e-05,
+      "loss": 0.9397,
+      "step": 1811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318843355781395e-05,
+      "loss": 0.9121,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314633381401882e-05,
+      "loss": 0.9985,
+      "step": 1813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310421544758794e-05,
+      "loss": 0.9539,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630620784709425e-05,
+      "loss": 0.9565,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301992289650925e-05,
+      "loss": 0.9373,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297774873672036e-05,
+      "loss": 0.9194,
+      "step": 1817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629355560040135e-05,
+      "loss": 0.9921,
+      "step": 1818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289334471083194e-05,
+      "loss": 0.9988,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285111486962414e-05,
+      "loss": 1.0088,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280886649284432e-05,
+      "loss": 1.0056,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62766599592952e-05,
+      "loss": 0.9255,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627243141824123e-05,
+      "loss": 0.958,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626820102736956e-05,
+      "loss": 0.9082,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626396878792779e-05,
+      "loss": 1.0461,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259734701164058e-05,
+      "loss": 0.6902,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625549876832705e-05,
+      "loss": 1.0203,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251260990665993e-05,
+      "loss": 0.6913,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624702136943066e-05,
+      "loss": 1.0122,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242779905871375e-05,
+      "loss": 1.0095,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623853660123899e-05,
+      "loss": 0.9153,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234291456784912e-05,
+      "loss": 0.5975,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623004447376108e-05,
+      "loss": 0.6764,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225795653419995e-05,
+      "loss": 0.9475,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221544997014673e-05,
+      "loss": 0.6559,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621729250579869e-05,
+      "loss": 0.988,
+      "step": 1836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213038181026166e-05,
+      "loss": 0.9578,
+      "step": 1837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208782023951744e-05,
+      "loss": 0.9724,
+      "step": 1838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620452403583062e-05,
+      "loss": 0.9832,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620026421791853e-05,
+      "loss": 0.9976,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196002571471745e-05,
+      "loss": 0.9736,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619173909774708e-05,
+      "loss": 0.9727,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618747379800188e-05,
+      "loss": 0.978,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618320667349404e-05,
+      "loss": 0.9653,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178937725481984e-05,
+      "loss": 1.0305,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174666955224686e-05,
+      "loss": 0.98,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617039436398164e-05,
+      "loss": 0.9441,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616611995301289e-05,
+      "loss": 0.9766,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616184372357901e-05,
+      "loss": 0.9458,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157565676941115e-05,
+      "loss": 1.0273,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153285814360856e-05,
+      "loss": 1.0012,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149004137100412e-05,
+      "loss": 1.0024,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614472064642251e-05,
+      "loss": 0.9412,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61404353435904e-05,
+      "loss": 0.644,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613614822986787e-05,
+      "loss": 0.5998,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131859306519243e-05,
+      "loss": 1.0029,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612756857480938e-05,
+      "loss": 0.9727,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123276036003664e-05,
+      "loss": 0.9668,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611898169136802e-05,
+      "loss": 0.938,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114685542168912e-05,
+      "loss": 0.9243,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110387589673315e-05,
+      "loss": 0.989,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610608783514876e-05,
+      "loss": 0.6807,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101786279863285e-05,
+      "loss": 0.8442,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097482925085482e-05,
+      "loss": 0.9597,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609317777208446e-05,
+      "loss": 0.6598,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088870822129862e-05,
+      "loss": 0.9873,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608456207649186e-05,
+      "loss": 0.9854,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608025153644116e-05,
+      "loss": 0.9185,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607593920324899e-05,
+      "loss": 1.0037,
+      "step": 1869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071625078187113e-05,
+      "loss": 0.9763,
+      "step": 1870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067309162527817e-05,
+      "loss": 0.9766,
+      "step": 1871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606299145754392e-05,
+      "loss": 0.9893,
+      "step": 1872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605867196450876e-05,
+      "loss": 0.9304,
+      "step": 1873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054350684696216e-05,
+      "loss": 0.9646,
+      "step": 1874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050027619380687e-05,
+      "loss": 0.9507,
+      "step": 1875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60457027698371e-05,
+      "loss": 0.6884,
+      "step": 1876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041376137340893e-05,
+      "loss": 0.9841,
+      "step": 1877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603704772316806e-05,
+      "loss": 0.9983,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032717528595098e-05,
+      "loss": 0.9822,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602838555489904e-05,
+      "loss": 0.9761,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024051803357425e-05,
+      "loss": 0.9419,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019716275248342e-05,
+      "loss": 0.9685,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601537897185039e-05,
+      "loss": 0.9149,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011039894442685e-05,
+      "loss": 0.9883,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600669904430489e-05,
+      "loss": 0.8955,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002356422717163e-05,
+      "loss": 0.6454,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59980120309602e-05,
+      "loss": 0.9839,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599366587031522e-05,
+      "loss": 0.969,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989317942063955e-05,
+      "loss": 0.9678,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598496824748866e-05,
+      "loss": 0.9099,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980616787872122e-05,
+      "loss": 0.6603,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597626356449764e-05,
+      "loss": 0.9771,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971908578649024e-05,
+      "loss": 0.7192,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967551831610624e-05,
+      "loss": 0.9209,
+      "step": 1894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596319332466729e-05,
+      "loss": 0.9507,
+      "step": 1895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958833059104405e-05,
+      "loss": 0.905,
+      "step": 1896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954471036207865e-05,
+      "loss": 0.9658,
+      "step": 1897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950107257264076e-05,
+      "loss": 0.9153,
+      "step": 1898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945741723559987e-05,
+      "loss": 0.6657,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941374436383035e-05,
+      "loss": 0.6658,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593700539702119e-05,
+      "loss": 0.9697,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932634606762935e-05,
+      "loss": 1.0171,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928262066897277e-05,
+      "loss": 0.9761,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923887778713726e-05,
+      "loss": 0.9434,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591951174350231e-05,
+      "loss": 0.9502,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915133962553584e-05,
+      "loss": 0.9458,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910754437158608e-05,
+      "loss": 0.9519,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906373168608952e-05,
+      "loss": 0.9031,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901990158196712e-05,
+      "loss": 0.9751,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897605407214497e-05,
+      "loss": 0.9983,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589321891695541e-05,
+      "loss": 1.0286,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888830688713096e-05,
+      "loss": 0.9619,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884440723781687e-05,
+      "loss": 1.0017,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880049023455848e-05,
+      "loss": 1.001,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875655589030738e-05,
+      "loss": 0.9795,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871260421802034e-05,
+      "loss": 0.9155,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866863523065933e-05,
+      "loss": 1.0349,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586246489411913e-05,
+      "loss": 0.6425,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858064536258837e-05,
+      "loss": 1.0022,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853662450782772e-05,
+      "loss": 0.9312,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849258638989166e-05,
+      "loss": 0.6235,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844853102176757e-05,
+      "loss": 1.0046,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840445841644793e-05,
+      "loss": 1.022,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836036858693024e-05,
+      "loss": 0.9844,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583162615462172e-05,
+      "loss": 0.9507,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827213730731656e-05,
+      "loss": 0.9604,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822799588324102e-05,
+      "loss": 0.6283,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581838372870085e-05,
+      "loss": 0.9609,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813966153164186e-05,
+      "loss": 0.613,
+      "step": 1929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580954686301691e-05,
+      "loss": 1.0195,
+      "step": 1930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805125859562327e-05,
+      "loss": 1.0562,
+      "step": 1931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800703144104242e-05,
+      "loss": 0.6672,
+      "step": 1932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796278717946972e-05,
+      "loss": 0.9355,
+      "step": 1933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791852582395334e-05,
+      "loss": 0.6428,
+      "step": 1934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787424738754652e-05,
+      "loss": 1.0159,
+      "step": 1935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782995188330746e-05,
+      "loss": 1.0457,
+      "step": 1936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778563932429952e-05,
+      "loss": 0.6667,
+      "step": 1937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57741309723591e-05,
+      "loss": 0.9358,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576969630942552e-05,
+      "loss": 0.9902,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765259944937058e-05,
+      "loss": 0.97,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760821880202047e-05,
+      "loss": 0.6415,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575638211652933e-05,
+      "loss": 0.9651,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751940655228242e-05,
+      "loss": 0.9468,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574749749760863e-05,
+      "loss": 0.917,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743052644980838e-05,
+      "loss": 0.9756,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738606098655708e-05,
+      "loss": 0.9426,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5734157859944574e-05,
+      "loss": 0.9031,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729707930159282e-05,
+      "loss": 0.9375,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572525631061218e-05,
+      "loss": 0.9355,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572080300261609e-05,
+      "loss": 0.969,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716348007484358e-05,
+      "loss": 0.6086,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571189132653081e-05,
+      "loss": 1.021,
+      "step": 1952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707432961069785e-05,
+      "loss": 0.991,
+      "step": 1953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702972912416103e-05,
+      "loss": 0.6946,
+      "step": 1954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569851118188509e-05,
+      "loss": 0.6537,
+      "step": 1955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569404777079257e-05,
+      "loss": 1.0139,
+      "step": 1956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568958268045485e-05,
+      "loss": 0.9338,
+      "step": 1957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685115912188745e-05,
+      "loss": 1.0339,
+      "step": 1958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.9946,
+      "step": 1959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676177347141096e-05,
+      "loss": 1.01,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671705552995642e-05,
+      "loss": 0.9631,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667232086193994e-05,
+      "loss": 0.8984,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566275694805542e-05,
+      "loss": 0.9299,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658280139899706e-05,
+      "loss": 1.0051,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653801663047113e-05,
+      "loss": 0.9585,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56493215188184e-05,
+      "loss": 0.8933,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644839708534817e-05,
+      "loss": 0.9399,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640356233518103e-05,
+      "loss": 1.0037,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635871095090495e-05,
+      "loss": 0.6437,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563138429457471e-05,
+      "loss": 0.9065,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626895833293965e-05,
+      "loss": 1.0078,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562240571257196e-05,
+      "loss": 0.9932,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617913933732892e-05,
+      "loss": 0.9717,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613420498101438e-05,
+      "loss": 0.9336,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608925407002774e-05,
+      "loss": 0.9124,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604428661762548e-05,
+      "loss": 0.9861,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559993026370691e-05,
+      "loss": 0.9587,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595430214162498e-05,
+      "loss": 0.9985,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590928514456428e-05,
+      "loss": 0.8948,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586425165916308e-05,
+      "loss": 0.9092,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5581920169870227e-05,
+      "loss": 0.9788,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557741352764677e-05,
+      "loss": 0.9717,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572905240575004e-05,
+      "loss": 0.9395,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556839530998447e-05,
+      "loss": 0.9675,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563883737205207e-05,
+      "loss": 0.9536,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559370523567734e-05,
+      "loss": 0.9792,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554855670403057e-05,
+      "loss": 0.9907,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555033917904265e-05,
+      "loss": 0.9709,
+      "step": 1988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545821050818494e-05,
+      "loss": 0.9048,
+      "step": 1989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541301287063036e-05,
+      "loss": 0.9524,
+      "step": 1990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536779889109213e-05,
+      "loss": 0.9473,
+      "step": 1991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532256858290442e-05,
+      "loss": 1.0051,
+      "step": 1992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527732195940616e-05,
+      "loss": 1.0437,
+      "step": 1993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523205903394117e-05,
+      "loss": 0.6257,
+      "step": 1994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518677981985802e-05,
+      "loss": 0.6302,
+      "step": 1995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514148433051018e-05,
+      "loss": 0.8579,
+      "step": 1996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550961725792558e-05,
+      "loss": 0.9629,
+      "step": 1997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5505084457945786e-05,
+      "loss": 1.0183,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500550034448415e-05,
+      "loss": 0.9609,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549601398877073e-05,
+      "loss": 0.9646,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549147632225046e-05,
+      "loss": 0.9595,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486937036225824e-05,
+      "loss": 0.9585,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482396132035515e-05,
+      "loss": 0.9192,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477853611018697e-05,
+      "loss": 1.0498,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473309474515014e-05,
+      "loss": 0.6528,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468763723864598e-05,
+      "loss": 0.949,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546421636040804e-05,
+      "loss": 0.6154,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459667385486416e-05,
+      "loss": 1.0088,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545511680044127e-05,
+      "loss": 0.6715,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450564606614635e-05,
+      "loss": 0.9507,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446010805348997e-05,
+      "loss": 0.9695,
+      "step": 2011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441455397987342e-05,
+      "loss": 0.917,
+      "step": 2012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436898385873105e-05,
+      "loss": 1.0269,
+      "step": 2013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432339770350212e-05,
+      "loss": 0.5766,
+      "step": 2014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427779552763053e-05,
+      "loss": 0.915,
+      "step": 2015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423217734456493e-05,
+      "loss": 0.6201,
+      "step": 2016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418654316775868e-05,
+      "loss": 0.6313,
+      "step": 2017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541408930106699e-05,
+      "loss": 0.8933,
+      "step": 2018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409522688676132e-05,
+      "loss": 0.9863,
+      "step": 2019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404954480950047e-05,
+      "loss": 0.9226,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540038467923596e-05,
+      "loss": 0.9624,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539581328488156e-05,
+      "loss": 0.6605,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391240299235e-05,
+      "loss": 0.9729,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386665723644923e-05,
+      "loss": 0.9604,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382089559460423e-05,
+      "loss": 0.9509,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377511808031064e-05,
+      "loss": 0.6497,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372932470706885e-05,
+      "loss": 1.0205,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368351548838386e-05,
+      "loss": 0.9451,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363769043776545e-05,
+      "loss": 0.9729,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359184956872785e-05,
+      "loss": 0.979,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535459928947903e-05,
+      "loss": 0.9478,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350012042947633e-05,
+      "loss": 0.9194,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345423218631438e-05,
+      "loss": 0.6626,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340832817883743e-05,
+      "loss": 0.9775,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533624084205832e-05,
+      "loss": 0.6099,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331647292509394e-05,
+      "loss": 0.9714,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532705217059166e-05,
+      "loss": 1.0217,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322455477660274e-05,
+      "loss": 0.9392,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531785721507086e-05,
+      "loss": 1.0227,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313257384179508e-05,
+      "loss": 0.9353,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308655986342757e-05,
+      "loss": 0.959,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530405302291762e-05,
+      "loss": 0.9744,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529944849526157e-05,
+      "loss": 0.6463,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294842404732535e-05,
+      "loss": 0.99,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529023475268891e-05,
+      "loss": 1.0493,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528562554048955e-05,
+      "loss": 0.9414,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5281014769493766e-05,
+      "loss": 0.9697,
+      "step": 2047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527640244106133e-05,
+      "loss": 0.9436,
+      "step": 2048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271788556552477e-05,
+      "loss": 0.9722,
+      "step": 2049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52671731173279e-05,
+      "loss": 0.9128,
+      "step": 2050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262556124748754e-05,
+      "loss": 0.9663,
+      "step": 2051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257937580176636e-05,
+      "loss": 0.6373,
+      "step": 2052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525331748497362e-05,
+      "loss": 0.6544,
+      "step": 2053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248695840502223e-05,
+      "loss": 1.0222,
+      "step": 2054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244072648125432e-05,
+      "loss": 0.9282,
+      "step": 2055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239447909206679e-05,
+      "loss": 1.071,
+      "step": 2056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234821625109862e-05,
+      "loss": 0.9519,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230193797199324e-05,
+      "loss": 1.0071,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225564426839872e-05,
+      "loss": 0.9587,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522093351539676e-05,
+      "loss": 1.0002,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216301064235706e-05,
+      "loss": 0.9829,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211667074722875e-05,
+      "loss": 0.9194,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520703154822489e-05,
+      "loss": 0.9426,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202394486108823e-05,
+      "loss": 0.6675,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197755889742202e-05,
+      "loss": 0.9614,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193115760493007e-05,
+      "loss": 0.9941,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188474099729669e-05,
+      "loss": 0.9226,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183830908821069e-05,
+      "loss": 0.9807,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179186189136545e-05,
+      "loss": 0.6595,
+      "step": 2069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517453994204588e-05,
+      "loss": 0.9331,
+      "step": 2070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169892168919312e-05,
+      "loss": 0.9338,
+      "step": 2071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165242871127523e-05,
+      "loss": 0.9272,
+      "step": 2072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160592050041654e-05,
+      "loss": 0.9807,
+      "step": 2073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155939707033287e-05,
+      "loss": 1.0161,
+      "step": 2074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151285843474454e-05,
+      "loss": 0.9517,
+      "step": 2075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146630460737638e-05,
+      "loss": 0.9788,
+      "step": 2076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141973560195768e-05,
+      "loss": 0.9331,
+      "step": 2077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137315143222227e-05,
+      "loss": 0.9238,
+      "step": 2078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132655211190832e-05,
+      "loss": 0.9873,
+      "step": 2079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127993765475859e-05,
+      "loss": 0.9836,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512333080745203e-05,
+      "loss": 0.9929,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118666338494501e-05,
+      "loss": 1.0059,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114000359978884e-05,
+      "loss": 0.9578,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109332873281237e-05,
+      "loss": 0.9988,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510466387977806e-05,
+      "loss": 0.9824,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099993380846291e-05,
+      "loss": 1.0278,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509532137786332e-05,
+      "loss": 0.9954,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090647872206984e-05,
+      "loss": 0.9768,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085972865255556e-05,
+      "loss": 0.969,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508129635838775e-05,
+      "loss": 0.6507,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076618352982724e-05,
+      "loss": 0.8967,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071938850420087e-05,
+      "loss": 0.9585,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067257852079876e-05,
+      "loss": 0.9893,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062575359342575e-05,
+      "loss": 0.9741,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057891373589113e-05,
+      "loss": 0.9954,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053205896200856e-05,
+      "loss": 0.9897,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048518928559608e-05,
+      "loss": 1.0295,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504383047204761e-05,
+      "loss": 0.9209,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5039140528047548e-05,
+      "loss": 0.8866,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034449097942549e-05,
+      "loss": 0.8906,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029756183116171e-05,
+      "loss": 0.6432,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025061784952408e-05,
+      "loss": 0.6804,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50203659048357e-05,
+      "loss": 0.9507,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015668544150925e-05,
+      "loss": 0.9922,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010969704283387e-05,
+      "loss": 0.6573,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006269386618833e-05,
+      "loss": 0.9695,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001567592543445e-05,
+      "loss": 0.9253,
+      "step": 2107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499686432344384e-05,
+      "loss": 0.9097,
+      "step": 2108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499215958070707e-05,
+      "loss": 0.9387,
+      "step": 2109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987453365720624e-05,
+      "loss": 0.9807,
+      "step": 2110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982745679872421e-05,
+      "loss": 1.0061,
+      "step": 2111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978036524550816e-05,
+      "loss": 0.9871,
+      "step": 2112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49733259011446e-05,
+      "loss": 0.9768,
+      "step": 2113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496861381104299e-05,
+      "loss": 0.9963,
+      "step": 2114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963900255635644e-05,
+      "loss": 0.6438,
+      "step": 2115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959185236312642e-05,
+      "loss": 0.9097,
+      "step": 2116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954468754464509e-05,
+      "loss": 0.9189,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949750811482187e-05,
+      "loss": 0.6277,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945031408757058e-05,
+      "loss": 0.9819,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940310547680938e-05,
+      "loss": 0.9814,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935588229646057e-05,
+      "loss": 0.9116,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930864456045096e-05,
+      "loss": 0.9468,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926139228271143e-05,
+      "loss": 0.9727,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492141254771774e-05,
+      "loss": 0.9216,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916684415778828e-05,
+      "loss": 0.9448,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911954833848805e-05,
+      "loss": 0.9993,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907223803322474e-05,
+      "loss": 0.8884,
+      "step": 2127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490249132559508e-05,
+      "loss": 0.9641,
+      "step": 2128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897757402062285e-05,
+      "loss": 1.0291,
+      "step": 2129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893022034120186e-05,
+      "loss": 0.9958,
+      "step": 2130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888285223165299e-05,
+      "loss": 0.9773,
+      "step": 2131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883546970594571e-05,
+      "loss": 0.6359,
+      "step": 2132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878807277805366e-05,
+      "loss": 0.9888,
+      "step": 2133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874066146195485e-05,
+      "loss": 0.5812,
+      "step": 2134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869323577163139e-05,
+      "loss": 1.0125,
+      "step": 2135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864579572106977e-05,
+      "loss": 1.0039,
+      "step": 2136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485983413242606e-05,
+      "loss": 0.9822,
+      "step": 2137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855087259519882e-05,
+      "loss": 0.9252,
+      "step": 2138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485033895478835e-05,
+      "loss": 0.9407,
+      "step": 2139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48455892196318e-05,
+      "loss": 1.0,
+      "step": 2140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840838055450986e-05,
+      "loss": 0.946,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836085463647088e-05,
+      "loss": 0.9788,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831331445621699e-05,
+      "loss": 0.9824,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826576002776842e-05,
+      "loss": 0.9641,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821819136514956e-05,
+      "loss": 0.9695,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817060848238902e-05,
+      "loss": 0.9636,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812301139351953e-05,
+      "loss": 0.978,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807540011257806e-05,
+      "loss": 0.9153,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802777465360581e-05,
+      "loss": 0.9756,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798013503064809e-05,
+      "loss": 0.9326,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793248125775442e-05,
+      "loss": 1.002,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478848133489785e-05,
+      "loss": 0.9993,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783713131837813e-05,
+      "loss": 0.988,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778943518001543e-05,
+      "loss": 0.9209,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774172494795651e-05,
+      "loss": 0.7115,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476940006362717e-05,
+      "loss": 0.9609,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764626225903555e-05,
+      "loss": 0.9761,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759850983032666e-05,
+      "loss": 0.9211,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755074336422778e-05,
+      "loss": 0.9585,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475029628748259e-05,
+      "loss": 0.9375,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745516837621207e-05,
+      "loss": 1.0164,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740735988248144e-05,
+      "loss": 0.9451,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735953740773334e-05,
+      "loss": 0.9636,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731170096607121e-05,
+      "loss": 0.9482,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726385057160264e-05,
+      "loss": 0.6619,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721598623843924e-05,
+      "loss": 0.9536,
+      "step": 2166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716810798069683e-05,
+      "loss": 0.9465,
+      "step": 2167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712021581249534e-05,
+      "loss": 0.9529,
+      "step": 2168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707230974795869e-05,
+      "loss": 0.9191,
+      "step": 2169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47024389801215e-05,
+      "loss": 0.9192,
+      "step": 2170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697645598639652e-05,
+      "loss": 1.0051,
+      "step": 2171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692850831763943e-05,
+      "loss": 1.0027,
+      "step": 2172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688054680908414e-05,
+      "loss": 0.9368,
+      "step": 2173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683257147487506e-05,
+      "loss": 0.9414,
+      "step": 2174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678458232916075e-05,
+      "loss": 0.6382,
+      "step": 2175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467365793860938e-05,
+      "loss": 0.7028,
+      "step": 2176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4668856265983085e-05,
+      "loss": 0.9985,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466405321645326e-05,
+      "loss": 0.9387,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659248791436393e-05,
+      "loss": 1.0195,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465444299234936e-05,
+      "loss": 0.6128,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649635820609457e-05,
+      "loss": 0.9832,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644827277634367e-05,
+      "loss": 0.8812,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640017364842205e-05,
+      "loss": 0.9832,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635206083651461e-05,
+      "loss": 0.6608,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630393435481047e-05,
+      "loss": 0.978,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625579421750271e-05,
+      "loss": 0.9141,
+      "step": 2186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620764043878845e-05,
+      "loss": 0.9553,
+      "step": 2187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615947303286888e-05,
+      "loss": 0.9792,
+      "step": 2188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611129201394911e-05,
+      "loss": 1.01,
+      "step": 2189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606309739623834e-05,
+      "loss": 1.0068,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601488919394975e-05,
+      "loss": 1.0442,
+      "step": 2191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596666742130055e-05,
+      "loss": 0.9973,
+      "step": 2192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591843209251191e-05,
+      "loss": 0.9246,
+      "step": 2193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587018322180906e-05,
+      "loss": 0.9419,
+      "step": 2194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458219208234212e-05,
+      "loss": 0.9753,
+      "step": 2195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577364491158147e-05,
+      "loss": 0.6814,
+      "step": 2196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572535550052702e-05,
+      "loss": 0.8884,
+      "step": 2197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567705260449905e-05,
+      "loss": 1.0007,
+      "step": 2198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562873623774267e-05,
+      "loss": 0.9946,
+      "step": 2199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558040641450695e-05,
+      "loss": 0.9509,
+      "step": 2200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553206314904493e-05,
+      "loss": 0.9172,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548370645561372e-05,
+      "loss": 0.9512,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543533634847422e-05,
+      "loss": 0.9712,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538695284189145e-05,
+      "loss": 0.9543,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533855595013424e-05,
+      "loss": 0.9612,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529014568747551e-05,
+      "loss": 0.6353,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524172206819195e-05,
+      "loss": 0.9441,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519328510656435e-05,
+      "loss": 0.9873,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514483481687733e-05,
+      "loss": 0.6021,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450963712134195e-05,
+      "loss": 0.989,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504789431048343e-05,
+      "loss": 0.9683,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499940412236549e-05,
+      "loss": 0.9841,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495090066336604e-05,
+      "loss": 1.0525,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490238394778936e-05,
+      "loss": 1.0005,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485385398994364e-05,
+      "loss": 0.9207,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480531080414096e-05,
+      "loss": 0.6317,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475675440469729e-05,
+      "loss": 0.9595,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470818480593258e-05,
+      "loss": 0.936,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465960202217053e-05,
+      "loss": 0.9446,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461100606773884e-05,
+      "loss": 0.9216,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456239695696903e-05,
+      "loss": 0.9551,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451377470419658e-05,
+      "loss": 0.9734,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446513932376076e-05,
+      "loss": 0.626,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444164908300047e-05,
+      "loss": 0.9863,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436782923727557e-05,
+      "loss": 0.6742,
+      "step": 2225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431915455992416e-05,
+      "loss": 0.9153,
+      "step": 2226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427046681230525e-05,
+      "loss": 0.9825,
+      "step": 2227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422176600877751e-05,
+      "loss": 0.9587,
+      "step": 2228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441730521637034e-05,
+      "loss": 0.6359,
+      "step": 2229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441243252914492e-05,
+      "loss": 0.9773,
+      "step": 2230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407558540638509e-05,
+      "loss": 0.9155,
+      "step": 2231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402683252288508e-05,
+      "loss": 1.0193,
+      "step": 2232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397806665532693e-05,
+      "loss": 0.9312,
+      "step": 2233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392928781809243e-05,
+      "loss": 1.0081,
+      "step": 2234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388049602556692e-05,
+      "loss": 0.9546,
+      "step": 2235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383169129213982e-05,
+      "loss": 1.0132,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378287363220415e-05,
+      "loss": 0.9685,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373404306015695e-05,
+      "loss": 0.6636,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368519959039882e-05,
+      "loss": 0.908,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363634323733442e-05,
+      "loss": 0.9761,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435874740153721e-05,
+      "loss": 0.9622,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353859193892392e-05,
+      "loss": 1.0195,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348969702240583e-05,
+      "loss": 0.5974,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434407892802376e-05,
+      "loss": 0.9561,
+      "step": 2244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339186872684268e-05,
+      "loss": 0.9299,
+      "step": 2245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334293537664836e-05,
+      "loss": 0.9436,
+      "step": 2246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329398924408567e-05,
+      "loss": 0.99,
+      "step": 2247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324503034358946e-05,
+      "loss": 0.979,
+      "step": 2248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319605868959831e-05,
+      "loss": 0.9355,
+      "step": 2249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314707429655456e-05,
+      "loss": 0.9534,
+      "step": 2250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309807717890432e-05,
+      "loss": 0.917,
+      "step": 2251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304906735109746e-05,
+      "loss": 0.8992,
+      "step": 2252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300004482758755e-05,
+      "loss": 0.9692,
+      "step": 2253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295100962283197e-05,
+      "loss": 0.6704,
+      "step": 2254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290196175129181e-05,
+      "loss": 1.0054,
+      "step": 2255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285290122743186e-05,
+      "loss": 0.9807,
+      "step": 2256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280382806572071e-05,
+      "loss": 0.9617,
+      "step": 2257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427547422806306e-05,
+      "loss": 0.9526,
+      "step": 2258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270564388663761e-05,
+      "loss": 0.9717,
+      "step": 2259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265653289822141e-05,
+      "loss": 0.9097,
+      "step": 2260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260740932986546e-05,
+      "loss": 0.9922,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255827319605686e-05,
+      "loss": 1.0066,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250912451128654e-05,
+      "loss": 1.011,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245996329004901e-05,
+      "loss": 0.6292,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241078954684253e-05,
+      "loss": 0.8696,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236160329616903e-05,
+      "loss": 0.946,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231240455253419e-05,
+      "loss": 0.9834,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226319333044729e-05,
+      "loss": 0.981,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221396964442135e-05,
+      "loss": 0.9531,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216473350897301e-05,
+      "loss": 0.991,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211548493862266e-05,
+      "loss": 0.6888,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206622394789432e-05,
+      "loss": 0.9475,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201695055131556e-05,
+      "loss": 0.651,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196766476341783e-05,
+      "loss": 1.0193,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191836659873613e-05,
+      "loss": 0.6891,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186905607180901e-05,
+      "loss": 0.9199,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181973319717882e-05,
+      "loss": 0.9392,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177039798939143e-05,
+      "loss": 0.9246,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172105046299647e-05,
+      "loss": 0.9739,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167169063254708e-05,
+      "loss": 0.9539,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416223185126001e-05,
+      "loss": 0.9346,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157293411771603e-05,
+      "loss": 0.9744,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152353746245888e-05,
+      "loss": 0.9863,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147412856139638e-05,
+      "loss": 1.0029,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142470742909976e-05,
+      "loss": 0.9133,
+      "step": 2285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137527408014397e-05,
+      "loss": 0.9189,
+      "step": 2286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132582852910753e-05,
+      "loss": 0.9922,
+      "step": 2287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412763707905725e-05,
+      "loss": 0.6556,
+      "step": 2288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122690087912462e-05,
+      "loss": 0.9775,
+      "step": 2289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117741880935318e-05,
+      "loss": 0.6779,
+      "step": 2290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4112792459585105e-05,
+      "loss": 1.0332,
+      "step": 2291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107841825321465e-05,
+      "loss": 0.9851,
+      "step": 2292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4102889979604403e-05,
+      "loss": 0.9299,
+      "step": 2293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097936923894282e-05,
+      "loss": 1.0381,
+      "step": 2294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092982659651822e-05,
+      "loss": 0.9739,
+      "step": 2295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088027188338088e-05,
+      "loss": 0.9451,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083070511414514e-05,
+      "loss": 0.9417,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078112630342891e-05,
+      "loss": 0.9536,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073153546585352e-05,
+      "loss": 0.9292,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068193261604397e-05,
+      "loss": 0.9612,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063231776862872e-05,
+      "loss": 0.9949,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058269093823982e-05,
+      "loss": 0.9668,
+      "step": 2302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053305213951283e-05,
+      "loss": 0.6611,
+      "step": 2303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048340138708684e-05,
+      "loss": 0.9099,
+      "step": 2304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404337386956045e-05,
+      "loss": 0.6092,
+      "step": 2305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038406407971192e-05,
+      "loss": 1.0273,
+      "step": 2306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033437755405878e-05,
+      "loss": 0.9429,
+      "step": 2307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028467913329826e-05,
+      "loss": 1.0098,
+      "step": 2308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023496883208702e-05,
+      "loss": 0.9128,
+      "step": 2309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018524666508526e-05,
+      "loss": 0.6334,
+      "step": 2310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013551264695663e-05,
+      "loss": 0.9792,
+      "step": 2311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008576679236834e-05,
+      "loss": 0.6472,
+      "step": 2312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003600911599105e-05,
+      "loss": 0.9141,
+      "step": 2313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998623963249894e-05,
+      "loss": 0.9756,
+      "step": 2314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993645835656955e-05,
+      "loss": 1.0576,
+      "step": 2315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988666530288413e-05,
+      "loss": 1.0688,
+      "step": 2316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983686048612716e-05,
+      "loss": 0.9871,
+      "step": 2317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978704392098678e-05,
+      "loss": 0.9753,
+      "step": 2318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973721562215443e-05,
+      "loss": 1.0654,
+      "step": 2319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396873756043252e-05,
+      "loss": 0.9753,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396375238821974e-05,
+      "loss": 0.929,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958766047047298e-05,
+      "loss": 0.9314,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953778538385732e-05,
+      "loss": 0.968,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948789863705914e-05,
+      "loss": 0.9375,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943800024479068e-05,
+      "loss": 0.9636,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393880902217676e-05,
+      "loss": 0.937,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933816858270898e-05,
+      "loss": 0.6506,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928823534233733e-05,
+      "loss": 0.9294,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923829051537858e-05,
+      "loss": 0.921,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918833411656205e-05,
+      "loss": 0.9202,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913836616062056e-05,
+      "loss": 0.9075,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908838666229023e-05,
+      "loss": 0.9954,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903839563631065e-05,
+      "loss": 0.9534,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389883930974248e-05,
+      "loss": 0.3439,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893837906037903e-05,
+      "loss": 0.9294,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888835353992312e-05,
+      "loss": 0.9355,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388383165508102e-05,
+      "loss": 0.9131,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878826810779677e-05,
+      "loss": 0.9319,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873820822564282e-05,
+      "loss": 0.9575,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868813691911154e-05,
+      "loss": 1.0232,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863805420296965e-05,
+      "loss": 0.9084,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858796009198708e-05,
+      "loss": 1.0193,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385378546009373e-05,
+      "loss": 0.9807,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848773774459696e-05,
+      "loss": 1.0334,
+      "step": 2344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843760953774616e-05,
+      "loss": 1.0232,
+      "step": 2345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838746999516836e-05,
+      "loss": 0.6958,
+      "step": 2346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833731913165032e-05,
+      "loss": 0.9761,
+      "step": 2347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828715696198214e-05,
+      "loss": 0.9268,
+      "step": 2348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823698350095726e-05,
+      "loss": 0.9541,
+      "step": 2349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381867987633725e-05,
+      "loss": 0.9897,
+      "step": 2350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813660276402787e-05,
+      "loss": 0.925,
+      "step": 2351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808639551772688e-05,
+      "loss": 0.9141,
+      "step": 2352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803617703927621e-05,
+      "loss": 0.9258,
+      "step": 2353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798594734348596e-05,
+      "loss": 0.9758,
+      "step": 2354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793570644516947e-05,
+      "loss": 0.9143,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788545435914337e-05,
+      "loss": 0.588,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783519110022768e-05,
+      "loss": 0.6462,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778491668324561e-05,
+      "loss": 0.9985,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773463112302374e-05,
+      "loss": 0.96,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768433443439189e-05,
+      "loss": 0.97,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763402663218317e-05,
+      "loss": 0.929,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758370773123397e-05,
+      "loss": 0.96,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753337774638397e-05,
+      "loss": 0.9692,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3748303669247611e-05,
+      "loss": 0.937,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374326845843566e-05,
+      "loss": 0.9719,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738232143687489e-05,
+      "loss": 0.9585,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733194726488366e-05,
+      "loss": 0.9976,
+      "step": 2367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728156208323892e-05,
+      "loss": 0.9221,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723116590679986e-05,
+      "loss": 0.9934,
+      "step": 2369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718075875042895e-05,
+      "loss": 0.9585,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713034062899191e-05,
+      "loss": 0.9636,
+      "step": 2371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707991155735761e-05,
+      "loss": 0.9502,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702947155039826e-05,
+      "loss": 0.947,
+      "step": 2373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369790206229892e-05,
+      "loss": 0.9683,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692855879000908e-05,
+      "loss": 0.9353,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687808606633965e-05,
+      "loss": 0.9509,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682760246686604e-05,
+      "loss": 0.9509,
+      "step": 2377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367771080064764e-05,
+      "loss": 0.9231,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367266027000622e-05,
+      "loss": 0.9399,
+      "step": 2379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667608656251808e-05,
+      "loss": 0.9778,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366255596087419e-05,
+      "loss": 0.6171,
+      "step": 2381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657502185363467e-05,
+      "loss": 0.9475,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652447331210057e-05,
+      "loss": 0.9736,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647391399904701e-05,
+      "loss": 0.9868,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642334392938458e-05,
+      "loss": 0.969,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637276311802701e-05,
+      "loss": 0.9724,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632217157989118e-05,
+      "loss": 0.9717,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627156932989719e-05,
+      "loss": 0.8862,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622095638296827e-05,
+      "loss": 0.926,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617033275403078e-05,
+      "loss": 1.02,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611969845801427e-05,
+      "loss": 0.9722,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606905350985143e-05,
+      "loss": 0.9263,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601839792447809e-05,
+      "loss": 0.9941,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596773171683317e-05,
+      "loss": 0.9272,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591705490185875e-05,
+      "loss": 0.9561,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586636749450015e-05,
+      "loss": 0.9028,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581566950970562e-05,
+      "loss": 0.9587,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576496096242665e-05,
+      "loss": 1.0303,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571424186761784e-05,
+      "loss": 0.9844,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566351224023684e-05,
+      "loss": 0.9175,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561277209524449e-05,
+      "loss": 0.9126,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556202144760461e-05,
+      "loss": 0.8804,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355112603122843e-05,
+      "loss": 0.9646,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.9778,
+      "step": 2404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540970663848565e-05,
+      "loss": 0.6556,
+      "step": 2405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535891412995675e-05,
+      "loss": 0.9565,
+      "step": 2406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530811119364623e-05,
+      "loss": 0.3342,
+      "step": 2407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525729784453648e-05,
+      "loss": 0.947,
+      "step": 2408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520647409761305e-05,
+      "loss": 0.9531,
+      "step": 2409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515563996786443e-05,
+      "loss": 0.947,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510479547028226e-05,
+      "loss": 0.9553,
+      "step": 2411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505394061986122e-05,
+      "loss": 0.9866,
+      "step": 2412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500307543159902e-05,
+      "loss": 0.9402,
+      "step": 2413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495219992049644e-05,
+      "loss": 0.9871,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349013141015573e-05,
+      "loss": 0.9226,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485041798978846e-05,
+      "loss": 0.9634,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347995116001998e-05,
+      "loss": 0.96,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3474859494780423e-05,
+      "loss": 0.8914,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469766804761774e-05,
+      "loss": 0.9312,
+      "step": 2419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464673091465929e-05,
+      "loss": 0.9902,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459578356395088e-05,
+      "loss": 1.0012,
+      "step": 2421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345448260105175e-05,
+      "loss": 1.0229,
+      "step": 2422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344938582693872e-05,
+      "loss": 0.8953,
+      "step": 2423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444288035559092e-05,
+      "loss": 0.9114,
+      "step": 2424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439189228416271e-05,
+      "loss": 0.9299,
+      "step": 2425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434089407013963e-05,
+      "loss": 0.9668,
+      "step": 2426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428988572856163e-05,
+      "loss": 0.9749,
+      "step": 2427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423886727447176e-05,
+      "loss": 0.9355,
+      "step": 2428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418783872291595e-05,
+      "loss": 0.8931,
+      "step": 2429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413680008894312e-05,
+      "loss": 0.9111,
+      "step": 2430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408575138760526e-05,
+      "loss": 0.9819,
+      "step": 2431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403469263395725e-05,
+      "loss": 0.9675,
+      "step": 2432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339836238430569e-05,
+      "loss": 0.9043,
+      "step": 2433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393254502996511e-05,
+      "loss": 0.9172,
+      "step": 2434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388145620974558e-05,
+      "loss": 0.9512,
+      "step": 2435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383035739746507e-05,
+      "loss": 0.9534,
+      "step": 2436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377924860819325e-05,
+      "loss": 0.9448,
+      "step": 2437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372812985700274e-05,
+      "loss": 0.9236,
+      "step": 2438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367700115896908e-05,
+      "loss": 0.9397,
+      "step": 2439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362586252917079e-05,
+      "loss": 1.0159,
+      "step": 2440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335747139826892e-05,
+      "loss": 0.6431,
+      "step": 2441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352355553460873e-05,
+      "loss": 0.9719,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347238720001663e-05,
+      "loss": 0.9585,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342120899400305e-05,
+      "loss": 0.9736,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333700209316611e-05,
+      "loss": 0.9021,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331882302808679e-05,
+      "loss": 0.9673,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326761529837895e-05,
+      "loss": 0.9795,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321639775763942e-05,
+      "loss": 0.967,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316517042097285e-05,
+      "loss": 0.9832,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311393330348689e-05,
+      "loss": 0.9822,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306268642029194e-05,
+      "loss": 0.9539,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301142978650133e-05,
+      "loss": 1.0269,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296016341723135e-05,
+      "loss": 0.9275,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32908887327601e-05,
+      "loss": 0.9192,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328576015327323e-05,
+      "loss": 0.9758,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280630604775002e-05,
+      "loss": 0.9133,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275500088778186e-05,
+      "loss": 0.9749,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270368606795833e-05,
+      "loss": 0.9456,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265236160341282e-05,
+      "loss": 0.9873,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260102750928154e-05,
+      "loss": 0.9817,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254968380070356e-05,
+      "loss": 0.9375,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324983304928208e-05,
+      "loss": 0.9666,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244696760077793e-05,
+      "loss": 0.9585,
+      "step": 2463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239559513972251e-05,
+      "loss": 0.9919,
+      "step": 2464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234421312480494e-05,
+      "loss": 0.9731,
+      "step": 2465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229282157117847e-05,
+      "loss": 0.9277,
+      "step": 2466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224142049399896e-05,
+      "loss": 0.949,
+      "step": 2467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219000990842531e-05,
+      "loss": 0.9602,
+      "step": 2468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321385898296192e-05,
+      "loss": 0.691,
+      "step": 2469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208716027274493e-05,
+      "loss": 0.9761,
+      "step": 2470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320357212529698e-05,
+      "loss": 0.9646,
+      "step": 2471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198427278546376e-05,
+      "loss": 0.9485,
+      "step": 2472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193281488539963e-05,
+      "loss": 0.913,
+      "step": 2473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188134756795296e-05,
+      "loss": 0.9783,
+      "step": 2474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182987084830213e-05,
+      "loss": 0.9456,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317783847416282e-05,
+      "loss": 0.957,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172688926311515e-05,
+      "loss": 0.9641,
+      "step": 2477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316753844279496e-05,
+      "loss": 0.9905,
+      "step": 2478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162387025132088e-05,
+      "loss": 0.9395,
+      "step": 2479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315723467484213e-05,
+      "loss": 0.9563,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152081393444568e-05,
+      "loss": 0.9353,
+      "step": 2481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146927182459172e-05,
+      "loss": 0.9495,
+      "step": 2482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141772043405983e-05,
+      "loss": 0.9448,
+      "step": 2483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136615977805312e-05,
+      "loss": 0.9098,
+      "step": 2484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131458987177752e-05,
+      "loss": 0.9229,
+      "step": 2485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126301073044158e-05,
+      "loss": 0.9604,
+      "step": 2486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121142236925666e-05,
+      "loss": 0.9368,
+      "step": 2487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115982480343679e-05,
+      "loss": 0.8923,
+      "step": 2488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110821804819877e-05,
+      "loss": 0.9246,
+      "step": 2489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31056602118762e-05,
+      "loss": 0.9307,
+      "step": 2490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100497703034874e-05,
+      "loss": 0.9739,
+      "step": 2491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095334279818377e-05,
+      "loss": 0.9526,
+      "step": 2492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.9568,
+      "step": 2493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085004696351192e-05,
+      "loss": 0.8645,
+      "step": 2494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079838539146824e-05,
+      "loss": 0.9871,
+      "step": 2495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074671473659932e-05,
+      "loss": 0.655,
+      "step": 2496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069503501414349e-05,
+      "loss": 0.8497,
+      "step": 2497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064334623934174e-05,
+      "loss": 0.9766,
+      "step": 2498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059164842743774e-05,
+      "loss": 0.9536,
+      "step": 2499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305399415936778e-05,
+      "loss": 0.9565,
+      "step": 2500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048822575331093e-05,
+      "loss": 0.9714,
+      "step": 2501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043650092158875e-05,
+      "loss": 0.9456,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038476711376558e-05,
+      "loss": 0.6464,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033302434509831e-05,
+      "loss": 0.979,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028127263084656e-05,
+      "loss": 1.0366,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022951198627254e-05,
+      "loss": 0.8811,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017774242664112e-05,
+      "loss": 1.0142,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012596396721978e-05,
+      "loss": 0.9211,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007417662327857e-05,
+      "loss": 0.907,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3002238041009032e-05,
+      "loss": 0.9646,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997057534293032e-05,
+      "loss": 0.9509,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991876143707653e-05,
+      "loss": 0.9707,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986693870780949e-05,
+      "loss": 0.8667,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981510717041244e-05,
+      "loss": 0.9526,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976326684017104e-05,
+      "loss": 0.9463,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297114177323737e-05,
+      "loss": 0.6437,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965955986231138e-05,
+      "loss": 0.8716,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960769324527758e-05,
+      "loss": 0.6388,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955581789656844e-05,
+      "loss": 0.978,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950393383148257e-05,
+      "loss": 0.9729,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294520410653213e-05,
+      "loss": 0.9795,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940013961338843e-05,
+      "loss": 0.9507,
+      "step": 2522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934822949099032e-05,
+      "loss": 0.926,
+      "step": 2523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929631071343588e-05,
+      "loss": 0.9419,
+      "step": 2524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924438329603668e-05,
+      "loss": 0.3563,
+      "step": 2525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919244725410673e-05,
+      "loss": 0.9863,
+      "step": 2526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914050260296258e-05,
+      "loss": 0.5724,
+      "step": 2527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908854935792339e-05,
+      "loss": 0.9519,
+      "step": 2528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903658753431077e-05,
+      "loss": 0.9409,
+      "step": 2529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898461714744897e-05,
+      "loss": 0.9238,
+      "step": 2530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893263821266462e-05,
+      "loss": 0.9707,
+      "step": 2531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28880650745287e-05,
+      "loss": 0.947,
+      "step": 2532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882865476064787e-05,
+      "loss": 0.915,
+      "step": 2533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877665027408146e-05,
+      "loss": 0.9622,
+      "step": 2534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872463730092457e-05,
+      "loss": 0.9733,
+      "step": 2535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286726158565164e-05,
+      "loss": 0.6126,
+      "step": 2536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862058595619882e-05,
+      "loss": 0.9321,
+      "step": 2537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856854761531601e-05,
+      "loss": 0.9034,
+      "step": 2538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851650084921474e-05,
+      "loss": 0.9316,
+      "step": 2539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846444567324428e-05,
+      "loss": 0.9734,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284123821027563e-05,
+      "loss": 0.9253,
+      "step": 2541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836031015310504e-05,
+      "loss": 0.9631,
+      "step": 2542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283082298396471e-05,
+      "loss": 0.9629,
+      "step": 2543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825614117774166e-05,
+      "loss": 0.9897,
+      "step": 2544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282040441827503e-05,
+      "loss": 0.9524,
+      "step": 2545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815193887003707e-05,
+      "loss": 1.0132,
+      "step": 2546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809982525496845e-05,
+      "loss": 0.9258,
+      "step": 2547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804770335291344e-05,
+      "loss": 0.9751,
+      "step": 2548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799557317924336e-05,
+      "loss": 0.9429,
+      "step": 2549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2794343474933211e-05,
+      "loss": 0.9253,
+      "step": 2550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789128807855592e-05,
+      "loss": 0.651,
+      "step": 2551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278391331822935e-05,
+      "loss": 1.0061,
+      "step": 2552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778697007592602e-05,
+      "loss": 0.9532,
+      "step": 2553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773479877483695e-05,
+      "loss": 0.9519,
+      "step": 2554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768261929441228e-05,
+      "loss": 0.918,
+      "step": 2555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763043165004038e-05,
+      "loss": 1.0112,
+      "step": 2556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757823585711209e-05,
+      "loss": 0.6737,
+      "step": 2557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275260319310205e-05,
+      "loss": 0.9629,
+      "step": 2558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747381988716127e-05,
+      "loss": 0.8649,
+      "step": 2559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742159974093235e-05,
+      "loss": 0.8938,
+      "step": 2560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736937150773413e-05,
+      "loss": 0.9058,
+      "step": 2561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2731713520296931e-05,
+      "loss": 1.0002,
+      "step": 2562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2726489084204308e-05,
+      "loss": 0.947,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272126384403629e-05,
+      "loss": 0.6183,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716037801333871e-05,
+      "loss": 0.8992,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710810957638267e-05,
+      "loss": 0.9146,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705583314490947e-05,
+      "loss": 0.9321,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700354873433608e-05,
+      "loss": 0.9795,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695125636008176e-05,
+      "loss": 0.9392,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268989560375682e-05,
+      "loss": 0.9473,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684664778221943e-05,
+      "loss": 0.979,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679433160946177e-05,
+      "loss": 0.9526,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674200753472395e-05,
+      "loss": 0.9612,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668967557343695e-05,
+      "loss": 0.9746,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663733574103415e-05,
+      "loss": 0.9419,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658498805295118e-05,
+      "loss": 0.9231,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653263252462607e-05,
+      "loss": 0.9153,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648026917149902e-05,
+      "loss": 0.9753,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642789800901273e-05,
+      "loss": 0.9253,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637551905261209e-05,
+      "loss": 0.9778,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263231323177442e-05,
+      "loss": 0.6194,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262707378198587e-05,
+      "loss": 0.905,
+      "step": 2582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262183355744073e-05,
+      "loss": 0.9604,
+      "step": 2583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616592559684408e-05,
+      "loss": 0.9517,
+      "step": 2584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261135079026254e-05,
+      "loss": 0.9534,
+      "step": 2585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606108250720985e-05,
+      "loss": 0.8569,
+      "step": 2586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260086494260584e-05,
+      "loss": 0.9561,
+      "step": 2587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595620867463416e-05,
+      "loss": 0.9338,
+      "step": 2588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590376026840251e-05,
+      "loss": 0.936,
+      "step": 2589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585130422283129e-05,
+      "loss": 0.9556,
+      "step": 2590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579884055339032e-05,
+      "loss": 0.947,
+      "step": 2591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574636927555179e-05,
+      "loss": 0.9045,
+      "step": 2592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569389040479012e-05,
+      "loss": 0.9705,
+      "step": 2593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564140395658207e-05,
+      "loss": 0.9287,
+      "step": 2594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558890994640644e-05,
+      "loss": 0.9602,
+      "step": 2595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553640838974439e-05,
+      "loss": 0.9277,
+      "step": 2596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548389930207932e-05,
+      "loss": 0.9221,
+      "step": 2597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543138269889679e-05,
+      "loss": 0.8997,
+      "step": 2598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253788585956846e-05,
+      "loss": 1.0117,
+      "step": 2599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253263270079327e-05,
+      "loss": 0.8953,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527378795113339e-05,
+      "loss": 0.9102,
+      "step": 2601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522124144078103e-05,
+      "loss": 0.9788,
+      "step": 2602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251686874923723e-05,
+      "loss": 0.9507,
+      "step": 2603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251161261214059e-05,
+      "loss": 0.9802,
+      "step": 2604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506355734338293e-05,
+      "loss": 0.948,
+      "step": 2605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250109811738066e-05,
+      "loss": 0.9124,
+      "step": 2606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495839762818216e-05,
+      "loss": 0.8984,
+      "step": 2607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249058067220172e-05,
+      "loss": 0.9983,
+      "step": 2608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485320847082143e-05,
+      "loss": 0.9473,
+      "step": 2609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480060289010677e-05,
+      "loss": 0.9417,
+      "step": 2610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247479899953872e-05,
+      "loss": 0.6838,
+      "step": 2611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246953698021789e-05,
+      "loss": 0.9409,
+      "step": 2612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464274232600029e-05,
+      "loss": 0.9702,
+      "step": 2613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459010758237178e-05,
+      "loss": 0.9983,
+      "step": 2614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453746558681607e-05,
+      "loss": 0.9585,
+      "step": 2615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448481635485792e-05,
+      "loss": 0.905,
+      "step": 2616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443215990202424e-05,
+      "loss": 0.9197,
+      "step": 2617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437949624384405e-05,
+      "loss": 0.9438,
+      "step": 2618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432682539584856e-05,
+      "loss": 0.6799,
+      "step": 2619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427414737357097e-05,
+      "loss": 1.0183,
+      "step": 2620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242214621925468e-05,
+      "loss": 0.8735,
+      "step": 2621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416876986831348e-05,
+      "loss": 0.9927,
+      "step": 2622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411607041641062e-05,
+      "loss": 0.6476,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406336385238e-05,
+      "loss": 0.968,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240106501917654e-05,
+      "loss": 0.9482,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395792945011273e-05,
+      "loss": 0.9385,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390520164297e-05,
+      "loss": 0.659,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385246678588732e-05,
+      "loss": 0.9773,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379972489441678e-05,
+      "loss": 0.9099,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237469759841127e-05,
+      "loss": 0.9329,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236942200705313e-05,
+      "loss": 0.9468,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364145716923103e-05,
+      "loss": 0.9329,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358868729577232e-05,
+      "loss": 1.012,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235359104657176e-05,
+      "loss": 0.9663,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348312669463145e-05,
+      "loss": 0.9333,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343033599808044e-05,
+      "loss": 0.9133,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337753839163324e-05,
+      "loss": 0.9395,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332473389086043e-05,
+      "loss": 0.6469,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327192251133482e-05,
+      "loss": 0.8892,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321910426863108e-05,
+      "loss": 0.9319,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316627917832598e-05,
+      "loss": 0.9287,
+      "step": 2641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311344725599828e-05,
+      "loss": 0.9314,
+      "step": 2642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306060851722877e-05,
+      "loss": 1.0293,
+      "step": 2643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300776297760026e-05,
+      "loss": 0.9641,
+      "step": 2644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295491065269753e-05,
+      "loss": 0.9961,
+      "step": 2645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229020515581074e-05,
+      "loss": 0.9709,
+      "step": 2646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284918570941871e-05,
+      "loss": 0.929,
+      "step": 2647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279631312222218e-05,
+      "loss": 0.8992,
+      "step": 2648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274343381211067e-05,
+      "loss": 0.9749,
+      "step": 2649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269054779467884e-05,
+      "loss": 0.9934,
+      "step": 2650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263765508552354e-05,
+      "loss": 0.9395,
+      "step": 2651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258475570024343e-05,
+      "loss": 0.9529,
+      "step": 2652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253184965443918e-05,
+      "loss": 0.9617,
+      "step": 2653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2247893696371347e-05,
+      "loss": 0.8838,
+      "step": 2654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224260176436709e-05,
+      "loss": 0.8999,
+      "step": 2655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237309170991805e-05,
+      "loss": 0.8921,
+      "step": 2656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223201591780634e-05,
+      "loss": 0.9368,
+      "step": 2657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222672200637174e-05,
+      "loss": 0.6362,
+      "step": 2658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222142743824925e-05,
+      "loss": 0.9553,
+      "step": 2659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22161322150003e-05,
+      "loss": 0.9668,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210836338186515e-05,
+      "loss": 0.9023,
+      "step": 2661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205539809369719e-05,
+      "loss": 0.8821,
+      "step": 2662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200242630111926e-05,
+      "loss": 0.9768,
+      "step": 2663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194944801975334e-05,
+      "loss": 0.9214,
+      "step": 2664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218964632652234e-05,
+      "loss": 0.6469,
+      "step": 2665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218434720531553e-05,
+      "loss": 0.9238,
+      "step": 2666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179047439917681e-05,
+      "loss": 0.9832,
+      "step": 2667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217374703189176e-05,
+      "loss": 0.938,
+      "step": 2668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168445982800923e-05,
+      "loss": 0.97,
+      "step": 2669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163144294208511e-05,
+      "loss": 0.9792,
+      "step": 2670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157841967678064e-05,
+      "loss": 0.9839,
+      "step": 2671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152539004773298e-05,
+      "loss": 0.9712,
+      "step": 2672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147235407058124e-05,
+      "loss": 0.9709,
+      "step": 2673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141931176096643e-05,
+      "loss": 0.6601,
+      "step": 2674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136626313453136e-05,
+      "loss": 0.9971,
+      "step": 2675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131320820692067e-05,
+      "loss": 0.9741,
+      "step": 2676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21260146993781e-05,
+      "loss": 1.002,
+      "step": 2677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120707951076069e-05,
+      "loss": 0.8597,
+      "step": 2678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115400577351009e-05,
+      "loss": 0.991,
+      "step": 2679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110092579768117e-05,
+      "loss": 0.8933,
+      "step": 2680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104783959892798e-05,
+      "loss": 0.9614,
+      "step": 2681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099474719290625e-05,
+      "loss": 0.9521,
+      "step": 2682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094164859527363e-05,
+      "loss": 0.9617,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088854382168946e-05,
+      "loss": 0.9214,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083543288781512e-05,
+      "loss": 0.9353,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078231580931359e-05,
+      "loss": 0.9385,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072919260184977e-05,
+      "loss": 0.8888,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067606328109038e-05,
+      "loss": 0.957,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062292786270391e-05,
+      "loss": 0.5912,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056978636236065e-05,
+      "loss": 1.01,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051663879573265e-05,
+      "loss": 1.0186,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046348517849384e-05,
+      "loss": 0.9724,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041032552631984e-05,
+      "loss": 0.9736,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035715985488814e-05,
+      "loss": 0.554,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030398817987785e-05,
+      "loss": 0.9578,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025081051697007e-05,
+      "loss": 0.9771,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019762688184753e-05,
+      "loss": 0.9993,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201444372901947e-05,
+      "loss": 0.9553,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009124175769784e-05,
+      "loss": 0.9531,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003804030004502e-05,
+      "loss": 0.9033,
+      "step": 2700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998483293292602e-05,
+      "loss": 0.991,
+      "step": 2701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993161967203228e-05,
+      "loss": 0.9714,
+      "step": 2702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987840053305706e-05,
+      "loss": 0.8926,
+      "step": 2703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982517553169544e-05,
+      "loss": 0.897,
+      "step": 2704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977194468364403e-05,
+      "loss": 0.9629,
+      "step": 2705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971870800460132e-05,
+      "loss": 0.9448,
+      "step": 2706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196654655102674e-05,
+      "loss": 1.0217,
+      "step": 2707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961221721634425e-05,
+      "loss": 0.9089,
+      "step": 2708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955896313853535e-05,
+      "loss": 0.9207,
+      "step": 2709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950570329254604e-05,
+      "loss": 0.9351,
+      "step": 2710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945243769408329e-05,
+      "loss": 0.937,
+      "step": 2711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939916635885575e-05,
+      "loss": 0.9329,
+      "step": 2712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193458893025739e-05,
+      "loss": 0.9561,
+      "step": 2713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192926065409497e-05,
+      "loss": 0.9153,
+      "step": 2714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923931808969693e-05,
+      "loss": 0.9607,
+      "step": 2715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918602396453102e-05,
+      "loss": 0.9048,
+      "step": 2716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191327241811691e-05,
+      "loss": 1.019,
+      "step": 2717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907941875532987e-05,
+      "loss": 0.9021,
+      "step": 2718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902610770273379e-05,
+      "loss": 0.9204,
+      "step": 2719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897279103910297e-05,
+      "loss": 0.6592,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891946878016114e-05,
+      "loss": 0.9434,
+      "step": 2721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886614094163368e-05,
+      "loss": 0.9004,
+      "step": 2722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881280753924763e-05,
+      "loss": 0.9763,
+      "step": 2723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187594685887317e-05,
+      "loss": 0.9749,
+      "step": 2724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870612410581617e-05,
+      "loss": 0.9802,
+      "step": 2725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18652774106233e-05,
+      "loss": 0.9111,
+      "step": 2726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185994186057158e-05,
+      "loss": 0.9888,
+      "step": 2727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185460576199997e-05,
+      "loss": 0.9255,
+      "step": 2728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849269116482162e-05,
+      "loss": 0.8928,
+      "step": 2729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843931925591987e-05,
+      "loss": 0.979,
+      "step": 2730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838594190903456e-05,
+      "loss": 0.9309,
+      "step": 2731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833255913990732e-05,
+      "loss": 0.9443,
+      "step": 2732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827917096428137e-05,
+      "loss": 0.9224,
+      "step": 2733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822577739790153e-05,
+      "loss": 0.9563,
+      "step": 2734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817237845651429e-05,
+      "loss": 0.9753,
+      "step": 2735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811897415586761e-05,
+      "loss": 0.968,
+      "step": 2736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806556451171108e-05,
+      "loss": 0.6618,
+      "step": 2737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801214953979584e-05,
+      "loss": 0.9575,
+      "step": 2738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179587292558747e-05,
+      "loss": 0.9355,
+      "step": 2739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790530367570194e-05,
+      "loss": 0.9836,
+      "step": 2740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785187281503336e-05,
+      "loss": 0.9395,
+      "step": 2741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779843668962647e-05,
+      "loss": 0.9861,
+      "step": 2742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774499531524018e-05,
+      "loss": 0.8667,
+      "step": 2743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769154870763504e-05,
+      "loss": 0.8774,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763809688257313e-05,
+      "loss": 0.9683,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758463985581803e-05,
+      "loss": 0.9888,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753117764313491e-05,
+      "loss": 0.96,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747771026029038e-05,
+      "loss": 0.9392,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742423772305269e-05,
+      "loss": 0.7147,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737076004719155e-05,
+      "loss": 0.9399,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731727724847816e-05,
+      "loss": 0.9346,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726378934268525e-05,
+      "loss": 0.9312,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172102963455871e-05,
+      "loss": 0.6418,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715679827295947e-05,
+      "loss": 0.9424,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710329514057956e-05,
+      "loss": 0.915,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704978696422614e-05,
+      "loss": 0.619,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169962737596794e-05,
+      "loss": 0.8435,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169427555427211e-05,
+      "loss": 0.9749,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688923232913436e-05,
+      "loss": 0.8601,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683570413470384e-05,
+      "loss": 0.9231,
+      "step": 2760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678217097521576e-05,
+      "loss": 0.9661,
+      "step": 2761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672863286645763e-05,
+      "loss": 0.9626,
+      "step": 2762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667508982421853e-05,
+      "loss": 0.8892,
+      "step": 2763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662154186428891e-05,
+      "loss": 0.612,
+      "step": 2764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656798900246082e-05,
+      "loss": 0.9939,
+      "step": 2765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165144312545276e-05,
+      "loss": 1.0232,
+      "step": 2766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164608686362841e-05,
+      "loss": 0.8899,
+      "step": 2767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640730116352661e-05,
+      "loss": 0.9565,
+      "step": 2768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635372885205285e-05,
+      "loss": 0.9729,
+      "step": 2769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630015171766196e-05,
+      "loss": 0.9011,
+      "step": 2770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624656977615445e-05,
+      "loss": 0.9724,
+      "step": 2771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619298304333236e-05,
+      "loss": 0.6656,
+      "step": 2772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1613939153499903e-05,
+      "loss": 0.9441,
+      "step": 2773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160857952669593e-05,
+      "loss": 0.948,
+      "step": 2774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603219425501935e-05,
+      "loss": 0.9338,
+      "step": 2775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159785885149868e-05,
+      "loss": 0.9312,
+      "step": 2776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592497806267064e-05,
+      "loss": 0.998,
+      "step": 2777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587136291388126e-05,
+      "loss": 0.9348,
+      "step": 2778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581774308443042e-05,
+      "loss": 0.9761,
+      "step": 2779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576411859013125e-05,
+      "loss": 0.6643,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571048944679835e-05,
+      "loss": 0.9319,
+      "step": 2781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565685567024754e-05,
+      "loss": 0.9712,
+      "step": 2782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560321727629614e-05,
+      "loss": 0.9077,
+      "step": 2783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554957428076272e-05,
+      "loss": 0.9192,
+      "step": 2784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549592669946732e-05,
+      "loss": 0.9021,
+      "step": 2785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544227454823124e-05,
+      "loss": 0.9688,
+      "step": 2786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538861784287718e-05,
+      "loss": 0.9648,
+      "step": 2787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533495659922917e-05,
+      "loss": 0.958,
+      "step": 2788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528129083311257e-05,
+      "loss": 0.9634,
+      "step": 2789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522762056035407e-05,
+      "loss": 0.8948,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517394579678171e-05,
+      "loss": 0.6808,
+      "step": 2791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512026655822483e-05,
+      "loss": 0.97,
+      "step": 2792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506658286051415e-05,
+      "loss": 0.8945,
+      "step": 2793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501289471948158e-05,
+      "loss": 0.8931,
+      "step": 2794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495920215096046e-05,
+      "loss": 0.8987,
+      "step": 2795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490550517078538e-05,
+      "loss": 0.9216,
+      "step": 2796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485180379479229e-05,
+      "loss": 0.9368,
+      "step": 2797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479809803881829e-05,
+      "loss": 0.978,
+      "step": 2798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474438791870201e-05,
+      "loss": 0.9163,
+      "step": 2799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146906734502831e-05,
+      "loss": 0.9468,
+      "step": 2800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146369546494027e-05,
+      "loss": 0.9651,
+      "step": 2801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458323153190314e-05,
+      "loss": 0.9065,
+      "step": 2802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14529504113628e-05,
+      "loss": 0.9607,
+      "step": 2803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144757724104222e-05,
+      "loss": 0.9573,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442203643813184e-05,
+      "loss": 0.915,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436829621260436e-05,
+      "loss": 0.6392,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431455174968839e-05,
+      "loss": 0.9656,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426080306523388e-05,
+      "loss": 0.9434,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420705017509197e-05,
+      "loss": 0.916,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14153293095115e-05,
+      "loss": 0.9336,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1409953184115668e-05,
+      "loss": 0.9866,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404576642907179e-05,
+      "loss": 0.9827,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399199687471647e-05,
+      "loss": 0.9792,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13938223193948e-05,
+      "loss": 0.9292,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388444540262496e-05,
+      "loss": 0.927,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383066351660698e-05,
+      "loss": 0.908,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377687755175511e-05,
+      "loss": 0.9436,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372308752393144e-05,
+      "loss": 0.9368,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366929344899931e-05,
+      "loss": 0.9792,
+      "step": 2819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361549534282332e-05,
+      "loss": 0.9062,
+      "step": 2820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356169322126911e-05,
+      "loss": 0.978,
+      "step": 2821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135078871002037e-05,
+      "loss": 0.9624,
+      "step": 2822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345407699549504e-05,
+      "loss": 1.0195,
+      "step": 2823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340026292301252e-05,
+      "loss": 0.9236,
+      "step": 2824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334644489862652e-05,
+      "loss": 0.9309,
+      "step": 2825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329262293820863e-05,
+      "loss": 0.9321,
+      "step": 2826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132387970576317e-05,
+      "loss": 0.9763,
+      "step": 2827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318496727276953e-05,
+      "loss": 0.9232,
+      "step": 2828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313113359949726e-05,
+      "loss": 0.9985,
+      "step": 2829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307729605369109e-05,
+      "loss": 0.9856,
+      "step": 2830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302345465122839e-05,
+      "loss": 0.9626,
+      "step": 2831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129696094079876e-05,
+      "loss": 0.917,
+      "step": 2832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1291576033984844e-05,
+      "loss": 0.9539,
+      "step": 2833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286190746269164e-05,
+      "loss": 1.011,
+      "step": 2834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280805079239903e-05,
+      "loss": 0.9192,
+      "step": 2835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275419034485365e-05,
+      "loss": 0.9263,
+      "step": 2836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127003261359396e-05,
+      "loss": 0.8838,
+      "step": 2837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264645818154211e-05,
+      "loss": 0.9546,
+      "step": 2838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125925864975475e-05,
+      "loss": 0.925,
+      "step": 2839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125387110998432e-05,
+      "loss": 0.9202,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248483200431774e-05,
+      "loss": 0.9336,
+      "step": 2841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124309492268607e-05,
+      "loss": 0.9332,
+      "step": 2842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237706278336284e-05,
+      "loss": 0.9626,
+      "step": 2843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232317268971586e-05,
+      "loss": 0.9639,
+      "step": 2844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226927896181271e-05,
+      "loss": 0.9902,
+      "step": 2845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221538161554724e-05,
+      "loss": 0.9192,
+      "step": 2846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1216148066681448e-05,
+      "loss": 0.9893,
+      "step": 2847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1210757613151044e-05,
+      "loss": 0.9993,
+      "step": 2848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.9199,
+      "step": 2849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199975636477822e-05,
+      "loss": 0.9648,
+      "step": 2850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119458411651474e-05,
+      "loss": 0.9744,
+      "step": 2851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189192244254006e-05,
+      "loss": 0.9939,
+      "step": 2852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183800021285756e-05,
+      "loss": 0.8992,
+      "step": 2853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178407449200222e-05,
+      "loss": 0.9634,
+      "step": 2854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173014529587734e-05,
+      "loss": 0.9932,
+      "step": 2855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167621264038742e-05,
+      "loss": 0.6819,
+      "step": 2856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162227654143777e-05,
+      "loss": 0.921,
+      "step": 2857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156833701493486e-05,
+      "loss": 0.9365,
+      "step": 2858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151439407678612e-05,
+      "loss": 0.9734,
+      "step": 2859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146044774289997e-05,
+      "loss": 0.9778,
+      "step": 2860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140649802918591e-05,
+      "loss": 0.897,
+      "step": 2861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113525449515543e-05,
+      "loss": 0.9246,
+      "step": 2862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129858852591662e-05,
+      "loss": 0.9407,
+      "step": 2863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124462876818526e-05,
+      "loss": 1.0327,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119066569427366e-05,
+      "loss": 0.9094,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113669932009619e-05,
+      "loss": 0.6311,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108272966156812e-05,
+      "loss": 0.6304,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102875673460592e-05,
+      "loss": 0.9236,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097478055512674e-05,
+      "loss": 0.9177,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092080113904886e-05,
+      "loss": 0.9558,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086681850229153e-05,
+      "loss": 0.9602,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081283266077488e-05,
+      "loss": 0.8838,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075884363041994e-05,
+      "loss": 0.9548,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107048514271488e-05,
+      "loss": 0.9163,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065085606688442e-05,
+      "loss": 0.6459,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059685756555069e-05,
+      "loss": 0.97,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054285593907248e-05,
+      "loss": 0.8997,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048885120337546e-05,
+      "loss": 0.9568,
+      "step": 2878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043484337438637e-05,
+      "loss": 0.6478,
+      "step": 2879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038083246803275e-05,
+      "loss": 0.9658,
+      "step": 2880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032681850024307e-05,
+      "loss": 0.917,
+      "step": 2881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027280148694676e-05,
+      "loss": 0.9377,
+      "step": 2882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021878144407408e-05,
+      "loss": 0.9758,
+      "step": 2883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016475838755623e-05,
+      "loss": 0.6048,
+      "step": 2884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011073233332525e-05,
+      "loss": 0.9592,
+      "step": 2885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100567032973141e-05,
+      "loss": 0.9207,
+      "step": 2886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000267129545661e-05,
+      "loss": 0.9169,
+      "step": 2887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994863634368754e-05,
+      "loss": 0.9346,
+      "step": 2888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989459845794234e-05,
+      "loss": 0.9585,
+      "step": 2889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984055765415757e-05,
+      "loss": 0.8542,
+      "step": 2890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978651394827054e-05,
+      "loss": 0.9658,
+      "step": 2891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973246735621932e-05,
+      "loss": 0.9019,
+      "step": 2892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967841789394296e-05,
+      "loss": 0.896,
+      "step": 2893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962436557738131e-05,
+      "loss": 0.9255,
+      "step": 2894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095703104224751e-05,
+      "loss": 0.96,
+      "step": 2895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951625244516584e-05,
+      "loss": 0.9768,
+      "step": 2896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946219166139587e-05,
+      "loss": 0.9028,
+      "step": 2897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094081280871084e-05,
+      "loss": 0.9038,
+      "step": 2898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935406173824749e-05,
+      "loss": 0.9507,
+      "step": 2899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929999263075795e-05,
+      "loss": 0.5726,
+      "step": 2900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924592078058538e-05,
+      "loss": 0.9841,
+      "step": 2901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919184620367634e-05,
+      "loss": 0.8994,
+      "step": 2902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913776891597798e-05,
+      "loss": 0.9858,
+      "step": 2903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908368893343843e-05,
+      "loss": 0.9907,
+      "step": 2904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902960627200655e-05,
+      "loss": 0.853,
+      "step": 2905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897552094763196e-05,
+      "loss": 0.9551,
+      "step": 2906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892143297626511e-05,
+      "loss": 0.6328,
+      "step": 2907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886734237385718e-05,
+      "loss": 0.8938,
+      "step": 2908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088132491563602e-05,
+      "loss": 0.9509,
+      "step": 2909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875915333972689e-05,
+      "loss": 0.8728,
+      "step": 2910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870505493991084e-05,
+      "loss": 0.9441,
+      "step": 2911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865095397286623e-05,
+      "loss": 0.9863,
+      "step": 2912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859685045454821e-05,
+      "loss": 0.9934,
+      "step": 2913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854274440091254e-05,
+      "loss": 0.9573,
+      "step": 2914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848863582791576e-05,
+      "loss": 0.6196,
+      "step": 2915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843452475151515e-05,
+      "loss": 0.9272,
+      "step": 2916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838041118766876e-05,
+      "loss": 0.9939,
+      "step": 2917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832629515233534e-05,
+      "loss": 0.9753,
+      "step": 2918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827217666147435e-05,
+      "loss": 0.9229,
+      "step": 2919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821805573104607e-05,
+      "loss": 0.9302,
+      "step": 2920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816393237701134e-05,
+      "loss": 0.96,
+      "step": 2921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081098066153319e-05,
+      "loss": 0.9448,
+      "step": 2922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805567846197007e-05,
+      "loss": 0.9517,
+      "step": 2923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800154793288888e-05,
+      "loss": 0.6694,
+      "step": 2924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794741504405215e-05,
+      "loss": 0.9712,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078932798114243e-05,
+      "loss": 0.9187,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078391422509705e-05,
+      "loss": 0.906,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778500237865656e-05,
+      "loss": 0.9369,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773086021044905e-05,
+      "loss": 0.9355,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767671576231512e-05,
+      "loss": 0.9141,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762256905022266e-05,
+      "loss": 0.9358,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756842009014016e-05,
+      "loss": 0.9485,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751426889803689e-05,
+      "loss": 0.6411,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746011548988269e-05,
+      "loss": 0.998,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07405959881648e-05,
+      "loss": 0.9272,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073518020893041e-05,
+      "loss": 0.9521,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729764212882268e-05,
+      "loss": 0.9062,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724348001617626e-05,
+      "loss": 0.9458,
+      "step": 2938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718931576733787e-05,
+      "loss": 0.9636,
+      "step": 2939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713514939828122e-05,
+      "loss": 0.9956,
+      "step": 2940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708098092498071e-05,
+      "loss": 1.0144,
+      "step": 2941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070268103634112e-05,
+      "loss": 0.8579,
+      "step": 2942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697263772954831e-05,
+      "loss": 0.9392,
+      "step": 2943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691846303936822e-05,
+      "loss": 0.9893,
+      "step": 2944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686428630884774e-05,
+      "loss": 0.9429,
+      "step": 2945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681010755396417e-05,
+      "loss": 0.9541,
+      "step": 2946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675592679069558e-05,
+      "loss": 0.9502,
+      "step": 2947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670174403502051e-05,
+      "loss": 0.9771,
+      "step": 2948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664755930291817e-05,
+      "loss": 0.9897,
+      "step": 2949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659337261036824e-05,
+      "loss": 0.9443,
+      "step": 2950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065391839733511e-05,
+      "loss": 0.9576,
+      "step": 2951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648499340784765e-05,
+      "loss": 0.887,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643080092983933e-05,
+      "loss": 0.9541,
+      "step": 2953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637660655530814e-05,
+      "loss": 0.9294,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632241030023678e-05,
+      "loss": 1.0,
+      "step": 2955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062682121806083e-05,
+      "loss": 0.9971,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621401221240646e-05,
+      "loss": 0.9316,
+      "step": 2957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061598104116154e-05,
+      "loss": 0.6036,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610560679422007e-05,
+      "loss": 0.9495,
+      "step": 2959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605140137620562e-05,
+      "loss": 0.9214,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599719417355801e-05,
+      "loss": 0.9424,
+      "step": 2961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594298520226354e-05,
+      "loss": 0.9055,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588877447830917e-05,
+      "loss": 0.9424,
+      "step": 2963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058345620176823e-05,
+      "loss": 1.0083,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057803478363708e-05,
+      "loss": 0.7058,
+      "step": 2965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572613195036319e-05,
+      "loss": 0.9875,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567191437564832e-05,
+      "loss": 0.9934,
+      "step": 2967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561769512821571e-05,
+      "loss": 1.0325,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055634742240552e-05,
+      "loss": 0.9351,
+      "step": 2969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055092516791573e-05,
+      "loss": 0.9927,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545502750951282e-05,
+      "loss": 0.9421,
+      "step": 2971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540080173111318e-05,
+      "loss": 0.6427,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534657435995024e-05,
+      "loss": 0.8928,
+      "step": 2973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529234541201631e-05,
+      "loss": 0.9089,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523811490330422e-05,
+      "loss": 0.9316,
+      "step": 2975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518388284980714e-05,
+      "loss": 0.938,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512964926751882e-05,
+      "loss": 0.9644,
+      "step": 2977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507541417243344e-05,
+      "loss": 0.8562,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502117758054556e-05,
+      "loss": 0.8933,
+      "step": 2979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496693950785023e-05,
+      "loss": 0.9692,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0491269997034292e-05,
+      "loss": 0.6101,
+      "step": 2981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485845898401959e-05,
+      "loss": 0.9827,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048042165648765e-05,
+      "loss": 0.9668,
+      "step": 2983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047499727289105e-05,
+      "loss": 0.9026,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469572749211869e-05,
+      "loss": 0.9595,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464148087049872e-05,
+      "loss": 0.938,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458723288004858e-05,
+      "loss": 0.936,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453298353676662e-05,
+      "loss": 0.9092,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447873285665168e-05,
+      "loss": 0.5923,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04424480855703e-05,
+      "loss": 0.9348,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437022754992011e-05,
+      "loss": 0.9165,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431597295530299e-05,
+      "loss": 0.9338,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426171708785203e-05,
+      "loss": 0.9551,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420745996356792e-05,
+      "loss": 0.9536,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415320159845175e-05,
+      "loss": 0.9388,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04098942008505e-05,
+      "loss": 0.9155,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040446812097295e-05,
+      "loss": 0.9819,
+      "step": 2997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399041921812745e-05,
+      "loss": 0.9749,
+      "step": 2998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393615604970132e-05,
+      "loss": 0.988,
+      "step": 2999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388189172045407e-05,
+      "loss": 0.6744,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382762624638885e-05,
+      "loss": 0.6223,
+      "step": 3001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377335964350928e-05,
+      "loss": 0.9617,
+      "step": 3002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371909192781922e-05,
+      "loss": 0.9414,
+      "step": 3003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366482311532286e-05,
+      "loss": 0.9604,
+      "step": 3004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361055322202482e-05,
+      "loss": 0.9663,
+      "step": 3005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355628226392994e-05,
+      "loss": 0.9487,
+      "step": 3006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350201025704334e-05,
+      "loss": 1.0046,
+      "step": 3007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0344773721737058e-05,
+      "loss": 0.969,
+      "step": 3008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339346316091742e-05,
+      "loss": 0.9426,
+      "step": 3009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333918810368994e-05,
+      "loss": 0.9448,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328491206169456e-05,
+      "loss": 1.0166,
+      "step": 3011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0323063505093791e-05,
+      "loss": 0.875,
+      "step": 3012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.03176357087427e-05,
+      "loss": 0.9468,
+      "step": 3013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312207818716905e-05,
+      "loss": 0.9668,
+      "step": 3014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306779836617156e-05,
+      "loss": 0.9827,
+      "step": 3015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301351764044239e-05,
+      "loss": 0.6676,
+      "step": 3016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295923602598951e-05,
+      "loss": 0.6279,
+      "step": 3017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290495353882131e-05,
+      "loss": 0.9702,
+      "step": 3018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285067019494632e-05,
+      "loss": 0.9495,
+      "step": 3019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279638601037338e-05,
+      "loss": 0.9121,
+      "step": 3020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027421010011116e-05,
+      "loss": 0.9856,
+      "step": 3021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026878151831703e-05,
+      "loss": 0.9277,
+      "step": 3022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263352857255899e-05,
+      "loss": 0.9023,
+      "step": 3023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257924118528749e-05,
+      "loss": 0.6312,
+      "step": 3024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252495303736584e-05,
+      "loss": 0.9678,
+      "step": 3025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247066414480424e-05,
+      "loss": 0.9224,
+      "step": 3026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241637452361323e-05,
+      "loss": 0.9871,
+      "step": 3027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236208418980345e-05,
+      "loss": 0.9177,
+      "step": 3028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023077931593858e-05,
+      "loss": 0.9058,
+      "step": 3029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225350144837134e-05,
+      "loss": 0.9597,
+      "step": 3030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021992090727714e-05,
+      "loss": 0.9093,
+      "step": 3031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021449160485975e-05,
+      "loss": 0.9456,
+      "step": 3032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209062239186127e-05,
+      "loss": 0.9648,
+      "step": 3033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203632811857462e-05,
+      "loss": 0.8689,
+      "step": 3034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019820332447496e-05,
+      "loss": 0.9453,
+      "step": 3035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192773778639842e-05,
+      "loss": 0.9619,
+      "step": 3036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187344175953351e-05,
+      "loss": 0.8624,
+      "step": 3037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018191451801674e-05,
+      "loss": 0.918,
+      "step": 3038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176484806431288e-05,
+      "loss": 0.9042,
+      "step": 3039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017105504279828e-05,
+      "loss": 0.937,
+      "step": 3040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165625228719021e-05,
+      "loss": 0.9011,
+      "step": 3041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160195365794832e-05,
+      "loss": 0.9502,
+      "step": 3042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154765455627048e-05,
+      "loss": 0.9341,
+      "step": 3043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014933549981701e-05,
+      "loss": 0.9424,
+      "step": 3044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143905499966086e-05,
+      "loss": 0.8613,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013847545767565e-05,
+      "loss": 0.9158,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133045374547084e-05,
+      "loss": 0.9083,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127615252181792e-05,
+      "loss": 0.6215,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122185092181178e-05,
+      "loss": 0.9316,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116754896146672e-05,
+      "loss": 0.959,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01113246656797e-05,
+      "loss": 0.9343,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105894402381703e-05,
+      "loss": 0.967,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100464107854137e-05,
+      "loss": 0.854,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095033783698462e-05,
+      "loss": 0.9409,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008960343151615e-05,
+      "loss": 0.9246,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084173052908675e-05,
+      "loss": 0.9384,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078742649477522e-05,
+      "loss": 0.9424,
+      "step": 3057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073312222824187e-05,
+      "loss": 0.9146,
+      "step": 3058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067881774550175e-05,
+      "loss": 0.9275,
+      "step": 3059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062451306256981e-05,
+      "loss": 0.8525,
+      "step": 3060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057020819546122e-05,
+      "loss": 0.8689,
+      "step": 3061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051590316019122e-05,
+      "loss": 0.9165,
+      "step": 3062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046159797277497e-05,
+      "loss": 0.8975,
+      "step": 3063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040729264922778e-05,
+      "loss": 0.9849,
+      "step": 3064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035298720556493e-05,
+      "loss": 0.9207,
+      "step": 3065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029868165780178e-05,
+      "loss": 0.9312,
+      "step": 3066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024437602195371e-05,
+      "loss": 1.0242,
+      "step": 3067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019007031403614e-05,
+      "loss": 0.958,
+      "step": 3068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013576455006446e-05,
+      "loss": 0.96,
+      "step": 3069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008145874605414e-05,
+      "loss": 0.8799,
+      "step": 3070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002715291802065e-05,
+      "loss": 0.9536,
+      "step": 3071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997284708197938e-06,
+      "loss": 0.9265,
+      "step": 3072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991854125394587e-06,
+      "loss": 0.6029,
+      "step": 3073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986423544993559e-06,
+      "loss": 0.9165,
+      "step": 3074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98099296859639e-06,
+      "loss": 0.9824,
+      "step": 3075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975562397804632e-06,
+      "loss": 0.9446,
+      "step": 3076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970131834219823e-06,
+      "loss": 0.6635,
+      "step": 3077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964701279443509e-06,
+      "loss": 0.9238,
+      "step": 3078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959270735077226e-06,
+      "loss": 0.9766,
+      "step": 3079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953840202722505e-06,
+      "loss": 0.9412,
+      "step": 3080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94840968398088e-06,
+      "loss": 0.9695,
+      "step": 3081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942979180453878e-06,
+      "loss": 0.9402,
+      "step": 3082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937548693743024e-06,
+      "loss": 0.9751,
+      "step": 3083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932118225449832e-06,
+      "loss": 0.8549,
+      "step": 3084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926687777175815e-06,
+      "loss": 0.9756,
+      "step": 3085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921257350522481e-06,
+      "loss": 0.9419,
+      "step": 3086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915826947091328e-06,
+      "loss": 0.9292,
+      "step": 3087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910396568483855e-06,
+      "loss": 0.9441,
+      "step": 3088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90496621630154e-06,
+      "loss": 0.939,
+      "step": 3089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899535892145864e-06,
+      "loss": 0.9766,
+      "step": 3090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894105597618297e-06,
+      "loss": 0.9451,
+      "step": 3091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888675334320302e-06,
+      "loss": 0.9878,
+      "step": 3092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883245103853333e-06,
+      "loss": 0.9341,
+      "step": 3093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877814907818824e-06,
+      "loss": 0.9199,
+      "step": 3094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87238474781821e-06,
+      "loss": 0.6174,
+      "step": 3095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866954625452918e-06,
+      "loss": 0.9163,
+      "step": 3096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861524542324355e-06,
+      "loss": 0.9675,
+      "step": 3097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856094500033916e-06,
+      "loss": 0.9023,
+      "step": 3098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850664500182991e-06,
+      "loss": 0.6615,
+      "step": 3099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845234544372956e-06,
+      "loss": 0.9795,
+      "step": 3100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839804634205168e-06,
+      "loss": 0.9153,
+      "step": 3101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834374771280982e-06,
+      "loss": 0.9512,
+      "step": 3102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828944957201724e-06,
+      "loss": 0.9922,
+      "step": 3103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823515193568715e-06,
+      "loss": 0.9351,
+      "step": 3104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818085481983262e-06,
+      "loss": 0.8455,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812655824046654e-06,
+      "loss": 0.9065,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807226221360163e-06,
+      "loss": 0.9036,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801796675525044e-06,
+      "loss": 0.946,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796367188142541e-06,
+      "loss": 0.9934,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790937760813874e-06,
+      "loss": 0.9316,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785508395140255e-06,
+      "loss": 0.9746,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780079092722863e-06,
+      "loss": 0.9038,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77464985516287e-06,
+      "loss": 0.6255,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769220684061422e-06,
+      "loss": 0.894,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76379158101966e-06,
+      "loss": 0.9795,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75836254763868e-06,
+      "loss": 0.9863,
+      "step": 3116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752933585519578e-06,
+      "loss": 0.9215,
+      "step": 3117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747504696263417e-06,
+      "loss": 0.9089,
+      "step": 3118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742075881471251e-06,
+      "loss": 0.8765,
+      "step": 3119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736647142744106e-06,
+      "loss": 0.9612,
+      "step": 3120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731218481682973e-06,
+      "loss": 0.9651,
+      "step": 3121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72578989988884e-06,
+      "loss": 0.9255,
+      "step": 3122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720361398962661e-06,
+      "loss": 0.9783,
+      "step": 3123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714932980505372e-06,
+      "loss": 0.9546,
+      "step": 3124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709504646117874e-06,
+      "loss": 0.9399,
+      "step": 3125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70407639740105e-06,
+      "loss": 0.9268,
+      "step": 3126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698648235955764e-06,
+      "loss": 0.9614,
+      "step": 3127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693220163382844e-06,
+      "loss": 0.9973,
+      "step": 3128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687792181283098e-06,
+      "loss": 0.9131,
+      "step": 3129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682364291257304e-06,
+      "loss": 0.9958,
+      "step": 3130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67693649490621e-06,
+      "loss": 0.9646,
+      "step": 3131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671508793830548e-06,
+      "loss": 0.9307,
+      "step": 3132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666081189631007e-06,
+      "loss": 0.9385,
+      "step": 3133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660653683908263e-06,
+      "loss": 0.9785,
+      "step": 3134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655226278262945e-06,
+      "loss": 0.8752,
+      "step": 3135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649798974295668e-06,
+      "loss": 0.9573,
+      "step": 3136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644371773607008e-06,
+      "loss": 0.9053,
+      "step": 3137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638944677797522e-06,
+      "loss": 0.8708,
+      "step": 3138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633517688467717e-06,
+      "loss": 0.948,
+      "step": 3139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628090807218082e-06,
+      "loss": 0.9155,
+      "step": 3140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622664035649074e-06,
+      "loss": 0.9235,
+      "step": 3141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617237375361115e-06,
+      "loss": 0.9656,
+      "step": 3142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6118108279546e-06,
+      "loss": 0.9365,
+      "step": 3143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60638439502987e-06,
+      "loss": 0.8984,
+      "step": 3144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600958078187258e-06,
+      "loss": 0.9368,
+      "step": 3145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595531879027052e-06,
+      "loss": 0.8906,
+      "step": 3146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590105799149502e-06,
+      "loss": 0.9114,
+      "step": 3147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584679840154829e-06,
+      "loss": 0.9988,
+      "step": 3148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579254003643213e-06,
+      "loss": 0.9753,
+      "step": 3149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573828291214799e-06,
+      "loss": 0.9028,
+      "step": 3150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568402704469701e-06,
+      "loss": 0.9609,
+      "step": 3151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562977245007994e-06,
+      "loss": 0.9731,
+      "step": 3152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557551914429703e-06,
+      "loss": 0.6273,
+      "step": 3153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552126714334834e-06,
+      "loss": 0.9253,
+      "step": 3154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546701646323341e-06,
+      "loss": 0.8826,
+      "step": 3155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541276711995149e-06,
+      "loss": 0.9233,
+      "step": 3156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535851912950133e-06,
+      "loss": 0.9377,
+      "step": 3157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530427250788134e-06,
+      "loss": 0.6364,
+      "step": 3158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525002727108954e-06,
+      "loss": 0.9153,
+      "step": 3159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51957834351235e-06,
+      "loss": 0.9521,
+      "step": 3160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514154101598046e-06,
+      "loss": 0.907,
+      "step": 3161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.508730002965711e-06,
+      "loss": 0.9404,
+      "step": 3162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50330604921498e-06,
+      "loss": 0.9014,
+      "step": 3163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497882241945446e-06,
+      "loss": 0.9382,
+      "step": 3164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492458582756658e-06,
+      "loss": 0.8875,
+      "step": 3165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487035073248123e-06,
+      "loss": 0.5473,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481611715019287e-06,
+      "loss": 0.9272,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476188509669582e-06,
+      "loss": 0.9316,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470765458798369e-06,
+      "loss": 0.9192,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46534256400498e-06,
+      "loss": 0.9863,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459919826888685e-06,
+      "loss": 0.9592,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454497249048723e-06,
+      "loss": 0.9382,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449074832084274e-06,
+      "loss": 0.8877,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443652577594482e-06,
+      "loss": 0.665,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438230487178434e-06,
+      "loss": 0.9326,
+      "step": 3175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43280856243517e-06,
+      "loss": 0.6071,
+      "step": 3176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427386804963684e-06,
+      "loss": 0.9312,
+      "step": 3177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.421965216362921e-06,
+      "loss": 0.9446,
+      "step": 3178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416543798231777e-06,
+      "loss": 0.9766,
+      "step": 3179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411122552169086e-06,
+      "loss": 0.9431,
+      "step": 3180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405701479773649e-06,
+      "loss": 0.9966,
+      "step": 3181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400280582644204e-06,
+      "loss": 0.8796,
+      "step": 3182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394859862379438e-06,
+      "loss": 0.9365,
+      "step": 3183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389439320578e-06,
+      "loss": 0.9189,
+      "step": 3184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384018958838462e-06,
+      "loss": 0.9526,
+      "step": 3185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378598778759357e-06,
+      "loss": 0.9326,
+      "step": 3186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373178781939172e-06,
+      "loss": 0.9551,
+      "step": 3187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367758969976327e-06,
+      "loss": 0.9039,
+      "step": 3188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362339344469189e-06,
+      "loss": 0.9043,
+      "step": 3189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35691990701607e-06,
+      "loss": 0.9744,
+      "step": 3190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.351500659215238e-06,
+      "loss": 0.8823,
+      "step": 3191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34608160266489e-06,
+      "loss": 0.9133,
+      "step": 3192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34066273896318e-06,
+      "loss": 0.9666,
+      "step": 3193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335244069708188e-06,
+      "loss": 0.905,
+      "step": 3194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32982559649795e-06,
+      "loss": 0.9187,
+      "step": 3195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324407320930444e-06,
+      "loss": 0.9373,
+      "step": 3196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318989244603586e-06,
+      "loss": 0.9673,
+      "step": 3197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313571369115231e-06,
+      "loss": 0.6075,
+      "step": 3198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30815369606318e-06,
+      "loss": 1.022,
+      "step": 3199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30273622704517e-06,
+      "loss": 0.6654,
+      "step": 3200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297318963658882e-06,
+      "loss": 1.0076,
+      "step": 3201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291901907501934e-06,
+      "loss": 0.9543,
+      "step": 3202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286485060171881e-06,
+      "loss": 0.9534,
+      "step": 3203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281068423266217e-06,
+      "loss": 0.9661,
+      "step": 3204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275651998382377e-06,
+      "loss": 0.9744,
+      "step": 3205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270235787117732e-06,
+      "loss": 0.845,
+      "step": 3206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264819791069596e-06,
+      "loss": 0.8882,
+      "step": 3207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259404011835203e-06,
+      "loss": 0.9248,
+      "step": 3208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253988451011734e-06,
+      "loss": 0.8813,
+      "step": 3209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24857311019631e-06,
+      "loss": 1.0073,
+      "step": 3210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243157990985989e-06,
+      "loss": 0.896,
+      "step": 3211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237743094977737e-06,
+      "loss": 0.9397,
+      "step": 3212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232328423768491e-06,
+      "loss": 0.9626,
+      "step": 3213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226913978955097e-06,
+      "loss": 0.9033,
+      "step": 3214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221499762134344e-06,
+      "loss": 0.9675,
+      "step": 3215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216085774902952e-06,
+      "loss": 1.0234,
+      "step": 3216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210672018857572e-06,
+      "loss": 0.9583,
+      "step": 3217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205258495594786e-06,
+      "loss": 0.9102,
+      "step": 3218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199845206711112e-06,
+      "loss": 0.9268,
+      "step": 3219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194432153802996e-06,
+      "loss": 0.8579,
+      "step": 3220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189019338466812e-06,
+      "loss": 0.8796,
+      "step": 3221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183606762298867e-06,
+      "loss": 0.9375,
+      "step": 3222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178194426895396e-06,
+      "loss": 0.9041,
+      "step": 3223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172782333852566e-06,
+      "loss": 0.9258,
+      "step": 3224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16737048476647e-06,
+      "loss": 0.916,
+      "step": 3225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161958881233129e-06,
+      "loss": 0.9382,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156547524848489e-06,
+      "loss": 0.9631,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151136417208424e-06,
+      "loss": 0.9058,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14572555990875e-06,
+      "loss": 0.9224,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140314954545184e-06,
+      "loss": 0.9248,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13490460271338e-06,
+      "loss": 0.99,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12949450600892e-06,
+      "loss": 0.6075,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124084666027311e-06,
+      "loss": 0.9565,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118675084363986e-06,
+      "loss": 0.9117,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113265762614285e-06,
+      "loss": 0.9299,
+      "step": 3235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107856702373492e-06,
+      "loss": 0.9612,
+      "step": 3236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102447905236806e-06,
+      "loss": 0.8977,
+      "step": 3237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097039372799345e-06,
+      "loss": 0.9695,
+      "step": 3238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091631106656159e-06,
+      "loss": 0.8921,
+      "step": 3239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086223108402203e-06,
+      "loss": 0.8779,
+      "step": 3240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08081537963237e-06,
+      "loss": 0.9443,
+      "step": 3241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075407921941463e-06,
+      "loss": 0.969,
+      "step": 3242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07000073692421e-06,
+      "loss": 0.9382,
+      "step": 3243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064593826175254e-06,
+      "loss": 0.9758,
+      "step": 3244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059187191289161e-06,
+      "loss": 0.9363,
+      "step": 3245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053780833860416e-06,
+      "loss": 0.9626,
+      "step": 3246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04837475548342e-06,
+      "loss": 0.9702,
+      "step": 3247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042968957752493e-06,
+      "loss": 0.9128,
+      "step": 3248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037563442261872e-06,
+      "loss": 0.5904,
+      "step": 3249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032158210605708e-06,
+      "loss": 0.9324,
+      "step": 3250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02675326437807e-06,
+      "loss": 0.9851,
+      "step": 3251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021348605172951e-06,
+      "loss": 0.9104,
+      "step": 3252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015944234584245e-06,
+      "loss": 0.9485,
+      "step": 3253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010540154205767e-06,
+      "loss": 0.9583,
+      "step": 3254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00513636563125e-06,
+      "loss": 0.9961,
+      "step": 3255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999732870454339e-06,
+      "loss": 0.9204,
+      "step": 3256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994329670268595e-06,
+      "loss": 0.9053,
+      "step": 3257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988926766667478e-06,
+      "loss": 0.9021,
+      "step": 3258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983524161244379e-06,
+      "loss": 0.928,
+      "step": 3259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978121855592593e-06,
+      "loss": 0.9148,
+      "step": 3260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972719851305327e-06,
+      "loss": 0.9583,
+      "step": 3261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967318149975695e-06,
+      "loss": 0.9783,
+      "step": 3262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961916753196728e-06,
+      "loss": 0.9336,
+      "step": 3263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956515662561364e-06,
+      "loss": 0.6266,
+      "step": 3264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951114879662455e-06,
+      "loss": 0.8602,
+      "step": 3265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945714406092757e-06,
+      "loss": 0.95,
+      "step": 3266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940314243444933e-06,
+      "loss": 0.9463,
+      "step": 3267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93491439331156e-06,
+      "loss": 0.9568,
+      "step": 3268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929514857285122e-06,
+      "loss": 0.5928,
+      "step": 3269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924115636958011e-06,
+      "loss": 0.6685,
+      "step": 3270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918716733922519e-06,
+      "loss": 0.9202,
+      "step": 3271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91331814977085e-06,
+      "loss": 0.6238,
+      "step": 3272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907919886095115e-06,
+      "loss": 0.9199,
+      "step": 3273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902521944487328e-06,
+      "loss": 0.9285,
+      "step": 3274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897124326539415e-06,
+      "loss": 0.6644,
+      "step": 3275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89172703384319e-06,
+      "loss": 0.9219,
+      "step": 3276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886330067990385e-06,
+      "loss": 0.9783,
+      "step": 3277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880933430572634e-06,
+      "loss": 0.9307,
+      "step": 3278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875537123181473e-06,
+      "loss": 0.9812,
+      "step": 3279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870141147408343e-06,
+      "loss": 1.0457,
+      "step": 3280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864745504844572e-06,
+      "loss": 1.0188,
+      "step": 3281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859350197081412e-06,
+      "loss": 0.9277,
+      "step": 3282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853955225710003e-06,
+      "loss": 0.9268,
+      "step": 3283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848560592321391e-06,
+      "loss": 0.9441,
+      "step": 3284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843166298506517e-06,
+      "loss": 0.967,
+      "step": 3285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837772345856226e-06,
+      "loss": 0.9294,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832378735961263e-06,
+      "loss": 0.8918,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826985470412267e-06,
+      "loss": 0.8978,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821592550799784e-06,
+      "loss": 0.9487,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816199978714249e-06,
+      "loss": 0.8975,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810807755745997e-06,
+      "loss": 0.9368,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805415883485264e-06,
+      "loss": 0.8363,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800024363522181e-06,
+      "loss": 0.8528,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 1.0237,
+      "step": 3294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789242386848958e-06,
+      "loss": 0.9092,
+      "step": 3295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783851933318555e-06,
+      "loss": 0.9045,
+      "step": 3296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778461838445277e-06,
+      "loss": 0.9529,
+      "step": 3297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773072103818734e-06,
+      "loss": 0.9138,
+      "step": 3298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767682731028415e-06,
+      "loss": 0.8804,
+      "step": 3299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762293721663718e-06,
+      "loss": 0.9736,
+      "step": 3300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75690507731393e-06,
+      "loss": 0.9294,
+      "step": 3301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75151679956823e-06,
+      "loss": 0.9658,
+      "step": 3302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746128890015684e-06,
+      "loss": 0.9712,
+      "step": 3303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740741350245252e-06,
+      "loss": 0.9465,
+      "step": 3304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73535418184579e-06,
+      "loss": 1.0017,
+      "step": 3305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729967386406042e-06,
+      "loss": 0.9192,
+      "step": 3306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724580965514638e-06,
+      "loss": 0.9265,
+      "step": 3307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7191949207601e-06,
+      "loss": 0.8892,
+      "step": 3308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71380925373084e-06,
+      "loss": 0.9302,
+      "step": 3309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.708423966015156e-06,
+      "loss": 0.9338,
+      "step": 3310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70303905920124e-06,
+      "loss": 0.9495,
+      "step": 3311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697654534877166e-06,
+      "loss": 0.8857,
+      "step": 3312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692270394630894e-06,
+      "loss": 0.9202,
+      "step": 3313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686886640050277e-06,
+      "loss": 0.9895,
+      "step": 3314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68150327272305e-06,
+      "loss": 0.8914,
+      "step": 3315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676120294236836e-06,
+      "loss": 0.9441,
+      "step": 3316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670737706179138e-06,
+      "loss": 0.8657,
+      "step": 3317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665355510137351e-06,
+      "loss": 0.927,
+      "step": 3318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659973707698751e-06,
+      "loss": 0.9319,
+      "step": 3319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654592300450496e-06,
+      "loss": 0.9377,
+      "step": 3320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649211289979638e-06,
+      "loss": 0.9204,
+      "step": 3321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643830677873092e-06,
+      "loss": 0.9089,
+      "step": 3322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638450465717671e-06,
+      "loss": 0.931,
+      "step": 3323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633070655100069e-06,
+      "loss": 0.9314,
+      "step": 3324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627691247606862e-06,
+      "loss": 0.9521,
+      "step": 3325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622312244824492e-06,
+      "loss": 0.6583,
+      "step": 3326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616933648339303e-06,
+      "loss": 0.9102,
+      "step": 3327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61155545973751e-06,
+      "loss": 0.918,
+      "step": 3328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6061776806052e-06,
+      "loss": 0.9434,
+      "step": 3329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600800312528356e-06,
+      "loss": 0.9846,
+      "step": 3330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595423357092823e-06,
+      "loss": 0.9294,
+      "step": 3331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590046815884337e-06,
+      "loss": 0.9402,
+      "step": 3332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584670690488501e-06,
+      "loss": 0.9902,
+      "step": 3333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579294982490808e-06,
+      "loss": 0.9731,
+      "step": 3334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573919693476614e-06,
+      "loss": 0.9395,
+      "step": 3335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568544825031163e-06,
+      "loss": 0.938,
+      "step": 3336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563170378739567e-06,
+      "loss": 0.9434,
+      "step": 3337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557796356186818e-06,
+      "loss": 0.9719,
+      "step": 3338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552422758957785e-06,
+      "loss": 0.9651,
+      "step": 3339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547049588637204e-06,
+      "loss": 0.897,
+      "step": 3340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54167684680969e-06,
+      "loss": 0.9385,
+      "step": 3341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536304535059731e-06,
+      "loss": 0.8708,
+      "step": 3342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530932654971693e-06,
+      "loss": 0.9504,
+      "step": 3343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525561208129805e-06,
+      "loss": 0.9124,
+      "step": 3344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520190196118173e-06,
+      "loss": 0.8909,
+      "step": 3345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514819620520773e-06,
+      "loss": 0.8975,
+      "step": 3346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509449482921462e-06,
+      "loss": 0.9106,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50407978490396e-06,
+      "loss": 0.9094,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498710528051845e-06,
+      "loss": 0.8848,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493341713948588e-06,
+      "loss": 0.9495,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487973344177517e-06,
+      "loss": 0.9302,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482605420321829e-06,
+      "loss": 0.8708,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477237943964596e-06,
+      "loss": 0.6467,
+      "step": 3353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471870916688745e-06,
+      "loss": 0.9436,
+      "step": 3354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466504340077084e-06,
+      "loss": 0.9006,
+      "step": 3355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461138215712283e-06,
+      "loss": 0.8958,
+      "step": 3356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45577254517688e-06,
+      "loss": 0.9565,
+      "step": 3357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450407330053271e-06,
+      "loss": 0.897,
+      "step": 3358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44504257192373e-06,
+      "loss": 0.9541,
+      "step": 3359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439678272370391e-06,
+      "loss": 0.8215,
+      "step": 3360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434314432975248e-06,
+      "loss": 1.0066,
+      "step": 3361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428951055320168e-06,
+      "loss": 0.9431,
+      "step": 3362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423588140986876e-06,
+      "loss": 0.877,
+      "step": 3363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418225691556962e-06,
+      "loss": 0.8875,
+      "step": 3364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412863708611874e-06,
+      "loss": 0.9575,
+      "step": 3365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40750219373294e-06,
+      "loss": 0.9431,
+      "step": 3366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402141148501323e-06,
+      "loss": 0.9153,
+      "step": 3367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396780574498066e-06,
+      "loss": 0.9126,
+      "step": 3368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391420473304071e-06,
+      "loss": 0.6702,
+      "step": 3369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386060846500097e-06,
+      "loss": 0.9153,
+      "step": 3370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38070169566677e-06,
+      "loss": 0.9106,
+      "step": 3371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375343022384559e-06,
+      "loss": 0.969,
+      "step": 3372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369984828233807e-06,
+      "loss": 0.8992,
+      "step": 3373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364627114794716e-06,
+      "loss": 0.9006,
+      "step": 3374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35926988364734e-06,
+      "loss": 0.9512,
+      "step": 3375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353913136371592e-06,
+      "loss": 0.9221,
+      "step": 3376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348556874547242e-06,
+      "loss": 0.8821,
+      "step": 3377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343201099753921e-06,
+      "loss": 0.9172,
+      "step": 3378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337845813571109e-06,
+      "loss": 0.8994,
+      "step": 3379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332491017578152e-06,
+      "loss": 0.9226,
+      "step": 3380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32713671335424e-06,
+      "loss": 0.9048,
+      "step": 3381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321782902478426e-06,
+      "loss": 0.908,
+      "step": 3382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316429586529616e-06,
+      "loss": 0.8796,
+      "step": 3383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311076767086566e-06,
+      "loss": 0.9241,
+      "step": 3384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305724445727896e-06,
+      "loss": 0.8962,
+      "step": 3385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300372624032062e-06,
+      "loss": 0.8955,
+      "step": 3386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29502130357739e-06,
+      "loss": 0.9058,
+      "step": 3387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289670485942044e-06,
+      "loss": 0.9446,
+      "step": 3388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284320172704058e-06,
+      "loss": 0.9052,
+      "step": 3389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278970365441292e-06,
+      "loss": 0.916,
+      "step": 3390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273621065731479e-06,
+      "loss": 0.8733,
+      "step": 3391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268272275152187e-06,
+      "loss": 0.6635,
+      "step": 3392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262923995280847e-06,
+      "loss": 1.0034,
+      "step": 3393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257576227694736e-06,
+      "loss": 0.9656,
+      "step": 3394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252228973970964e-06,
+      "loss": 0.6165,
+      "step": 3395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246882235686512e-06,
+      "loss": 0.9561,
+      "step": 3396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241536014418199e-06,
+      "loss": 0.927,
+      "step": 3397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23619031174269e-06,
+      "loss": 0.6261,
+      "step": 3398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230845129236498e-06,
+      "loss": 0.9089,
+      "step": 3399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225500468475984e-06,
+      "loss": 0.948,
+      "step": 3400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220156331037355e-06,
+      "loss": 0.9365,
+      "step": 3401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214812718496666e-06,
+      "loss": 0.8735,
+      "step": 3402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209469632429811e-06,
+      "loss": 1.0186,
+      "step": 3403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204127074412531e-06,
+      "loss": 0.9609,
+      "step": 3404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198785046020417e-06,
+      "loss": 0.6421,
+      "step": 3405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193443548828896e-06,
+      "loss": 0.939,
+      "step": 3406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188102584413244e-06,
+      "loss": 0.8887,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182762154348575e-06,
+      "loss": 0.9431,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177422260209849e-06,
+      "loss": 0.9272,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172082903571868e-06,
+      "loss": 0.9343,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16674408600927e-06,
+      "loss": 0.9192,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161405809096547e-06,
+      "loss": 0.8644,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156068074408016e-06,
+      "loss": 0.8977,
+      "step": 3413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150730883517841e-06,
+      "loss": 0.9478,
+      "step": 3414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145394238000028e-06,
+      "loss": 1.0098,
+      "step": 3415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140058139428425e-06,
+      "loss": 0.9475,
+      "step": 3416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134722589376704e-06,
+      "loss": 0.9255,
+      "step": 3417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129387589418385e-06,
+      "loss": 0.8896,
+      "step": 3418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124053141126833e-06,
+      "loss": 0.8997,
+      "step": 3419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118719246075239e-06,
+      "loss": 0.9353,
+      "step": 3420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113385905836636e-06,
+      "loss": 0.928,
+      "step": 3421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10805312198389e-06,
+      "loss": 0.9255,
+      "step": 3422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102720896089705e-06,
+      "loss": 0.9128,
+      "step": 3423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097389229726623e-06,
+      "loss": 0.9258,
+      "step": 3424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092058124467014e-06,
+      "loss": 0.9231,
+      "step": 3425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086727581883094e-06,
+      "loss": 0.9343,
+      "step": 3426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.0813976035469e-06,
+      "loss": 0.925,
+      "step": 3427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076068191030309e-06,
+      "loss": 0.9551,
+      "step": 3428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070739345905032e-06,
+      "loss": 0.9199,
+      "step": 3429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065411069742615e-06,
+      "loss": 0.9546,
+      "step": 3430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060083364114427e-06,
+      "loss": 0.9785,
+      "step": 3431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054756230591676e-06,
+      "loss": 0.6172,
+      "step": 3432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0494296707454e-06,
+      "loss": 0.9421,
+      "step": 3433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044103686146465e-06,
+      "loss": 0.959,
+      "step": 3434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03877827836558e-06,
+      "loss": 0.9431,
+      "step": 3435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033453448973261e-06,
+      "loss": 0.9971,
+      "step": 3436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02812919953987e-06,
+      "loss": 0.8694,
+      "step": 3437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022805531635599e-06,
+      "loss": 0.9431,
+      "step": 3438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01748244683046e-06,
+      "loss": 1.0039,
+      "step": 3439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012159946694295e-06,
+      "loss": 0.95,
+      "step": 3440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006838032796775e-06,
+      "loss": 0.5908,
+      "step": 3441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001516706707401e-06,
+      "loss": 0.9297,
+      "step": 3442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996195969995498e-06,
+      "loss": 0.623,
+      "step": 3443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990875824230217e-06,
+      "loss": 0.9536,
+      "step": 3444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985556270980534e-06,
+      "loss": 0.8899,
+      "step": 3445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98023731181525e-06,
+      "loss": 0.6545,
+      "step": 3446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974918948302993e-06,
+      "loss": 0.917,
+      "step": 3447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969601182012217e-06,
+      "loss": 1.0027,
+      "step": 3448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964284014511193e-06,
+      "loss": 0.8923,
+      "step": 3449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958967447368018e-06,
+      "loss": 0.9211,
+      "step": 3450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953651482150619e-06,
+      "loss": 0.9,
+      "step": 3451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948336120426736e-06,
+      "loss": 0.8767,
+      "step": 3452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943021363763941e-06,
+      "loss": 0.9529,
+      "step": 3453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937707213729612e-06,
+      "loss": 0.95,
+      "step": 3454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932393671890965e-06,
+      "loss": 0.9233,
+      "step": 3455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927080739815025e-06,
+      "loss": 0.9238,
+      "step": 3456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921768419068643e-06,
+      "loss": 0.6776,
+      "step": 3457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916456711218494e-06,
+      "loss": 0.9325,
+      "step": 3458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911145617831057e-06,
+      "loss": 0.9363,
+      "step": 3459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90583514047264e-06,
+      "loss": 0.9128,
+      "step": 3460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900525280709375e-06,
+      "loss": 0.9111,
+      "step": 3461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895216040107207e-06,
+      "loss": 0.9375,
+      "step": 3462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889907420231884e-06,
+      "loss": 0.9226,
+      "step": 3463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884599422648995e-06,
+      "loss": 0.929,
+      "step": 3464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87929204892393e-06,
+      "loss": 0.9797,
+      "step": 3465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873985300621902e-06,
+      "loss": 0.9062,
+      "step": 3466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868679179307934e-06,
+      "loss": 0.979,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863373686546868e-06,
+      "loss": 0.9771,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858068823903358e-06,
+      "loss": 0.946,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852764592941876e-06,
+      "loss": 0.9348,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847460995226706e-06,
+      "loss": 0.9312,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84215803232194e-06,
+      "loss": 0.6351,
+      "step": 3472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836855705791492e-06,
+      "loss": 0.908,
+      "step": 3473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83155401719908e-06,
+      "loss": 0.9668,
+      "step": 3474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826252968108242e-06,
+      "loss": 0.9319,
+      "step": 3475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820952560082322e-06,
+      "loss": 0.9163,
+      "step": 3476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815652794684472e-06,
+      "loss": 0.6235,
+      "step": 3477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810353673477664e-06,
+      "loss": 0.9531,
+      "step": 3478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805055198024668e-06,
+      "loss": 0.9619,
+      "step": 3479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799757369888079e-06,
+      "loss": 0.9636,
+      "step": 3480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794460190630283e-06,
+      "loss": 0.9536,
+      "step": 3481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789163661813486e-06,
+      "loss": 0.6605,
+      "step": 3482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783867784999701e-06,
+      "loss": 1.0356,
+      "step": 3483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778572561750752e-06,
+      "loss": 0.683,
+      "step": 3484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773277993628265e-06,
+      "loss": 0.6013,
+      "step": 3485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767984082193662e-06,
+      "loss": 0.8926,
+      "step": 3486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762690829008197e-06,
+      "loss": 0.9102,
+      "step": 3487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75739823563291e-06,
+      "loss": 1.0042,
+      "step": 3488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752106303628656e-06,
+      "loss": 0.6257,
+      "step": 3489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746815034556084e-06,
+      "loss": 0.3094,
+      "step": 3490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74152442997566e-06,
+      "loss": 0.9253,
+      "step": 3491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736234491447649e-06,
+      "loss": 0.9636,
+      "step": 3492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730945220532116e-06,
+      "loss": 0.9111,
+      "step": 3493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725656618788938e-06,
+      "loss": 0.9583,
+      "step": 3494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720368687777784e-06,
+      "loss": 0.8322,
+      "step": 3495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715081429058132e-06,
+      "loss": 0.9353,
+      "step": 3496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709794844189262e-06,
+      "loss": 0.9082,
+      "step": 3497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70450893473025e-06,
+      "loss": 0.9409,
+      "step": 3498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69922370223998e-06,
+      "loss": 0.9021,
+      "step": 3499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693939148277128e-06,
+      "loss": 0.8411,
+      "step": 3500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688655274400176e-06,
+      "loss": 0.8838,
+      "step": 3501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683372082167404e-06,
+      "loss": 0.9924,
+      "step": 3502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678089573136897e-06,
+      "loss": 0.8997,
+      "step": 3503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672807748866523e-06,
+      "loss": 0.855,
+      "step": 3504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667526610913958e-06,
+      "loss": 0.947,
+      "step": 3505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66224616083668e-06,
+      "loss": 0.6497,
+      "step": 3506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656966400191956e-06,
+      "loss": 0.916,
+      "step": 3507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651687330536862e-06,
+      "loss": 0.9062,
+      "step": 3508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646408953428243e-06,
+      "loss": 0.9241,
+      "step": 3509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641131270422772e-06,
+      "loss": 0.9619,
+      "step": 3510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635854283076897e-06,
+      "loss": 0.969,
+      "step": 3511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630577992946871e-06,
+      "loss": 0.9666,
+      "step": 3512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625302401588735e-06,
+      "loss": 0.876,
+      "step": 3513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620027510558324e-06,
+      "loss": 1.0034,
+      "step": 3514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6147533214112725e-06,
+      "loss": 0.9414,
+      "step": 3515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609479835703002e-06,
+      "loss": 0.9282,
+      "step": 3516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60420705498873e-06,
+      "loss": 0.9177,
+      "step": 3517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598934980823465e-06,
+      "loss": 0.9153,
+      "step": 3518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593663614762004e-06,
+      "loss": 0.9556,
+      "step": 3519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58839295835894e-06,
+      "loss": 0.9575,
+      "step": 3520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5831230131686585e-06,
+      "loss": 0.9077,
+      "step": 3521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577853780745324e-06,
+      "loss": 0.9617,
+      "step": 3522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572585262642904e-06,
+      "loss": 0.853,
+      "step": 3523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567317460415149e-06,
+      "loss": 0.9233,
+      "step": 3524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562050375615595e-06,
+      "loss": 0.9382,
+      "step": 3525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556784009797581e-06,
+      "loss": 0.8838,
+      "step": 3526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551518364514212e-06,
+      "loss": 0.9541,
+      "step": 3527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546253441318396e-06,
+      "loss": 0.9956,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5409892417628215e-06,
+      "loss": 0.9165,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535725767399973e-06,
+      "loss": 0.9326,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530463019782114e-06,
+      "loss": 0.9163,
+      "step": 3531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525201000461284e-06,
+      "loss": 0.9873,
+      "step": 3532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519939710989326e-06,
+      "loss": 0.9106,
+      "step": 3533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514679152917857e-06,
+      "loss": 0.9109,
+      "step": 3534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509419327798283e-06,
+      "loss": 0.9285,
+      "step": 3535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504160237181788e-06,
+      "loss": 0.9233,
+      "step": 3536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498901882619345e-06,
+      "loss": 0.9175,
+      "step": 3537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493644265661706e-06,
+      "loss": 0.957,
+      "step": 3538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48838738785941e-06,
+      "loss": 0.8674,
+      "step": 3539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483131250762776e-06,
+      "loss": 0.9736,
+      "step": 3540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477875855921899e-06,
+      "loss": 0.8374,
+      "step": 3541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472621204886665e-06,
+      "loss": 0.6584,
+      "step": 3542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467367299206733e-06,
+      "loss": 0.6002,
+      "step": 3543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4621141404315465e-06,
+      "loss": 0.6385,
+      "step": 3544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456861730110326e-06,
+      "loss": 0.9292,
+      "step": 3545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45161006979207e-06,
+      "loss": 0.9517,
+      "step": 3546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4463591610255626e-06,
+      "loss": 0.9224,
+      "step": 3547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441109005359357e-06,
+      "loss": 0.9827,
+      "step": 3548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4358596043418e-06,
+      "loss": 0.5754,
+      "step": 3549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430610959520991e-06,
+      "loss": 0.8879,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425363072444824e-06,
+      "loss": 0.9192,
+      "step": 3551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420115944660971e-06,
+      "loss": 1.0044,
+      "step": 3552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414869577716877e-06,
+      "loss": 0.9653,
+      "step": 3553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409623973159749e-06,
+      "loss": 0.9409,
+      "step": 3554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404379132536588e-06,
+      "loss": 0.9746,
+      "step": 3555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399135057394162e-06,
+      "loss": 0.9131,
+      "step": 3556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393891749279015e-06,
+      "loss": 0.9429,
+      "step": 3557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388649209737464e-06,
+      "loss": 0.9534,
+      "step": 3558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383407440315595e-06,
+      "loss": 0.9558,
+      "step": 3559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378166442559271e-06,
+      "loss": 0.938,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372926218014131e-06,
+      "loss": 0.9119,
+      "step": 3561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367686768225582e-06,
+      "loss": 0.9197,
+      "step": 3562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362448094738797e-06,
+      "loss": 0.83,
+      "step": 3563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35721019909873e-06,
+      "loss": 0.7063,
+      "step": 3564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351973082850099e-06,
+      "loss": 0.9114,
+      "step": 3565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346736747537397e-06,
+      "loss": 0.8535,
+      "step": 3566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341501194704885e-06,
+      "loss": 0.9434,
+      "step": 3567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336266425896589e-06,
+      "loss": 0.894,
+      "step": 3568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331032442656307e-06,
+      "loss": 0.8225,
+      "step": 3569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325799246527609e-06,
+      "loss": 0.9548,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320566839053824e-06,
+      "loss": 0.8867,
+      "step": 3571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315335221778064e-06,
+      "loss": 0.8884,
+      "step": 3572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310104396243185e-06,
+      "loss": 0.8369,
+      "step": 3573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304874363991828e-06,
+      "loss": 0.9543,
+      "step": 3574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299645126566395e-06,
+      "loss": 0.9497,
+      "step": 3575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294416685509056e-06,
+      "loss": 0.9187,
+      "step": 3576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289189042361736e-06,
+      "loss": 0.9573,
+      "step": 3577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2839621986661325e-06,
+      "loss": 0.9246,
+      "step": 3578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27873615596371e-06,
+      "loss": 0.8955,
+      "step": 3579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273510915795694e-06,
+      "loss": 0.9172,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2682864797030705e-06,
+      "loss": 0.9636,
+      "step": 3581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26306284922659e-06,
+      "loss": 0.9783,
+      "step": 3582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257840025906766e-06,
+      "loss": 0.9399,
+      "step": 3583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.252618011283874e-06,
+      "loss": 0.9199,
+      "step": 3584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247396806897953e-06,
+      "loss": 0.9219,
+      "step": 3585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242176414288795e-06,
+      "loss": 0.9678,
+      "step": 3586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236956834995963e-06,
+      "loss": 0.9753,
+      "step": 3587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231738070558775e-06,
+      "loss": 0.9017,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226520122516308e-06,
+      "loss": 0.9121,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221302992407404e-06,
+      "loss": 0.9407,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216086681770652e-06,
+      "loss": 0.9626,
+      "step": 3591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210871192144411e-06,
+      "loss": 0.9219,
+      "step": 3592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205656525066793e-06,
+      "loss": 0.9744,
+      "step": 3593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200442682075668e-06,
+      "loss": 0.9102,
+      "step": 3594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195229664708663e-06,
+      "loss": 0.9021,
+      "step": 3595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190017474503159e-06,
+      "loss": 0.8965,
+      "step": 3596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184806112996294e-06,
+      "loss": 0.9458,
+      "step": 3597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179595581724971e-06,
+      "loss": 0.9231,
+      "step": 3598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174385882225838e-06,
+      "loss": 0.9617,
+      "step": 3599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169177016035293e-06,
+      "loss": 0.9478,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1639689846895e-06,
+      "loss": 0.9331,
+      "step": 3601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158761789724371e-06,
+      "loss": 0.9243,
+      "step": 3602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153555432675574e-06,
+      "loss": 0.9292,
+      "step": 3603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148349915078528e-06,
+      "loss": 0.9346,
+      "step": 3604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143145238468403e-06,
+      "loss": 0.9204,
+      "step": 3605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137941404380122e-06,
+      "loss": 0.9749,
+      "step": 3606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132738414348361e-06,
+      "loss": 0.6605,
+      "step": 3607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127536269907549e-06,
+      "loss": 0.9744,
+      "step": 3608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122334972591858e-06,
+      "loss": 0.8604,
+      "step": 3609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117134523935217e-06,
+      "loss": 0.8955,
+      "step": 3610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111934925471302e-06,
+      "loss": 0.9282,
+      "step": 3611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10673617873354e-06,
+      "loss": 0.9431,
+      "step": 3612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101538285255108e-06,
+      "loss": 0.9028,
+      "step": 3613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096341246568925e-06,
+      "loss": 0.9524,
+      "step": 3614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091145064207665e-06,
+      "loss": 0.9675,
+      "step": 3615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0859497397037416e-06,
+      "loss": 0.863,
+      "step": 3616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080755274589332e-06,
+      "loss": 0.927,
+      "step": 3617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075561670396334e-06,
+      "loss": 0.9116,
+      "step": 3618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070368928656413e-06,
+      "loss": 0.8955,
+      "step": 3619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065177050900971e-06,
+      "loss": 0.9363,
+      "step": 3620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05998603866116e-06,
+      "loss": 0.9736,
+      "step": 3621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054795893467875e-06,
+      "loss": 0.918,
+      "step": 3622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0496066168517455e-06,
+      "loss": 0.9255,
+      "step": 3623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044418210343161e-06,
+      "loss": 0.9199,
+      "step": 3624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039230675472244e-06,
+      "loss": 0.9396,
+      "step": 3625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034044013768868e-06,
+      "loss": 0.9172,
+      "step": 3626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028858226762632e-06,
+      "loss": 0.9878,
+      "step": 3627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0236733159829e-06,
+      "loss": 0.6341,
+      "step": 3628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0184892829587605e-06,
+      "loss": 0.8914,
+      "step": 3629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013306129219052e-06,
+      "loss": 0.9226,
+      "step": 3630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008123856292352e-06,
+      "loss": 0.8928,
+      "step": 3631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002942465706972e-06,
+      "loss": 0.877,
+      "step": 3632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9977619589909705e-06,
+      "loss": 0.9185,
+      "step": 3633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992582337672145e-06,
+      "loss": 0.988,
+      "step": 3634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987403603278029e-06,
+      "loss": 0.9749,
+      "step": 3635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982225757335892e-06,
+      "loss": 0.9011,
+      "step": 3636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97704880137275e-06,
+      "loss": 0.9246,
+      "step": 3637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971872736915348e-06,
+      "loss": 0.9531,
+      "step": 3638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9666975654901704e-06,
+      "loss": 0.3273,
+      "step": 3639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9615232886234475e-06,
+      "loss": 0.6348,
+      "step": 3640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956349907841128e-06,
+      "loss": 0.9299,
+      "step": 3641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951177424668909e-06,
+      "loss": 0.6187,
+      "step": 3642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9460058406322195e-06,
+      "loss": 0.8771,
+      "step": 3643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940835157256226e-06,
+      "loss": 0.8923,
+      "step": 3644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93566537606583e-06,
+      "loss": 0.9099,
+      "step": 3645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930496498585654e-06,
+      "loss": 0.9209,
+      "step": 3646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925328526340072e-06,
+      "loss": 0.9163,
+      "step": 3647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920161460853179e-06,
+      "loss": 0.9673,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914995303648811e-06,
+      "loss": 0.6943,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.6161,
+      "step": 3650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9046657201816245e-06,
+      "loss": 0.8667,
+      "step": 3651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89950229696513e-06,
+      "loss": 0.9155,
+      "step": 3652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894339788123801e-06,
+      "loss": 0.8879,
+      "step": 3653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889178195180128e-06,
+      "loss": 0.9746,
+      "step": 3654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884017519656323e-06,
+      "loss": 0.9255,
+      "step": 3655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878857763074336e-06,
+      "loss": 0.9004,
+      "step": 3656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873698926955844e-06,
+      "loss": 0.9287,
+      "step": 3657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868541012822253e-06,
+      "loss": 0.9275,
+      "step": 3658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86338402219469e-06,
+      "loss": 0.9133,
+      "step": 3659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858227956594022e-06,
+      "loss": 0.6641,
+      "step": 3660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853072817540831e-06,
+      "loss": 0.6592,
+      "step": 3661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847918606555433e-06,
+      "loss": 1.009,
+      "step": 3662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842765325157874e-06,
+      "loss": 0.8687,
+      "step": 3663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837612974867913e-06,
+      "loss": 0.9917,
+      "step": 3664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832461557205043e-06,
+      "loss": 0.642,
+      "step": 3665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827311073688485e-06,
+      "loss": 0.9268,
+      "step": 3666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822161525837183e-06,
+      "loss": 0.9316,
+      "step": 3667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8170129151697925e-06,
+      "loss": 0.9304,
+      "step": 3668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811865243204706e-06,
+      "loss": 0.8665,
+      "step": 3669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80671851146004e-06,
+      "loss": 0.8619,
+      "step": 3670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801572721453625e-06,
+      "loss": 0.9136,
+      "step": 3671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796427874703025e-06,
+      "loss": 0.895,
+      "step": 3672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791283972725509e-06,
+      "loss": 0.9712,
+      "step": 3673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786141017038084e-06,
+      "loss": 0.9814,
+      "step": 3674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780999009157469e-06,
+      "loss": 0.8994,
+      "step": 3675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775857950600107e-06,
+      "loss": 0.8684,
+      "step": 3676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77071784288216e-06,
+      "loss": 0.8899,
+      "step": 3677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765578687519508e-06,
+      "loss": 0.6033,
+      "step": 3678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760440486027751e-06,
+      "loss": 0.9312,
+      "step": 3679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7553032399222104e-06,
+      "loss": 0.9768,
+      "step": 3680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750166950717925e-06,
+      "loss": 0.9214,
+      "step": 3681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745031619929646e-06,
+      "loss": 0.9211,
+      "step": 3682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739897249071847e-06,
+      "loss": 0.8835,
+      "step": 3683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73476383965872e-06,
+      "loss": 0.9829,
+      "step": 3684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729631393204167e-06,
+      "loss": 0.8602,
+      "step": 3685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7244999112218175e-06,
+      "loss": 0.9272,
+      "step": 3686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719369395225002e-06,
+      "loss": 0.926,
+      "step": 3687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714239846726771e-06,
+      "loss": 0.8547,
+      "step": 3688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7091112672399e-06,
+      "loss": 0.9121,
+      "step": 3689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70398365827687e-06,
+      "loss": 0.8945,
+      "step": 3690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69885702134987e-06,
+      "loss": 0.9019,
+      "step": 3691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6937313579708074e-06,
+      "loss": 0.9286,
+      "step": 3692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688606669651312e-06,
+      "loss": 0.9016,
+      "step": 3693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683482957902714e-06,
+      "loss": 0.9553,
+      "step": 3694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678360224236062e-06,
+      "loss": 0.8965,
+      "step": 3695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673238470162109e-06,
+      "loss": 0.9038,
+      "step": 3696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.668117697191325e-06,
+      "loss": 0.8938,
+      "step": 3697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66299790683389e-06,
+      "loss": 0.9829,
+      "step": 3698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657879100599697e-06,
+      "loss": 0.9951,
+      "step": 3699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652761279998338e-06,
+      "loss": 0.6649,
+      "step": 3700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647644446539129e-06,
+      "loss": 0.9385,
+      "step": 3701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642528601731082e-06,
+      "loss": 0.9636,
+      "step": 3702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6374137470829256e-06,
+      "loss": 0.913,
+      "step": 3703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632299884103096e-06,
+      "loss": 0.8826,
+      "step": 3704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62718701429973e-06,
+      "loss": 0.8955,
+      "step": 3705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622075139180678e-06,
+      "loss": 0.9121,
+      "step": 3706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616964260253496e-06,
+      "loss": 0.9387,
+      "step": 3707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6118543790254465e-06,
+      "loss": 0.9353,
+      "step": 3708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606745497003495e-06,
+      "loss": 0.8567,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601637615694313e-06,
+      "loss": 0.9189,
+      "step": 3710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596530736604278e-06,
+      "loss": 0.9783,
+      "step": 3711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591424861239475e-06,
+      "loss": 0.9607,
+      "step": 3712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586319991105692e-06,
+      "loss": 0.637,
+      "step": 3713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58121612770841e-06,
+      "loss": 0.9116,
+      "step": 3714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5761132725528265e-06,
+      "loss": 0.937,
+      "step": 3715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571011427143837e-06,
+      "loss": 0.8679,
+      "step": 3716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565910592986038e-06,
+      "loss": 0.9221,
+      "step": 3717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5608107715837295e-06,
+      "loss": 0.9363,
+      "step": 3718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5557119644409115e-06,
+      "loss": 0.9033,
+      "step": 3719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550614173061285e-06,
+      "loss": 0.9441,
+      "step": 3720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54551739894825e-06,
+      "loss": 0.9407,
+      "step": 3721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540421643604914e-06,
+      "loss": 0.9536,
+      "step": 3722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535326908534072e-06,
+      "loss": 0.9885,
+      "step": 3723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530233195238227e-06,
+      "loss": 0.8748,
+      "step": 3724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525140505219578e-06,
+      "loss": 0.9026,
+      "step": 3725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520048839980023e-06,
+      "loss": 0.9045,
+      "step": 3726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5149582010211595e-06,
+      "loss": 0.9365,
+      "step": 3727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509868589844274e-06,
+      "loss": 0.9231,
+      "step": 3728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50478000795036e-06,
+      "loss": 0.9294,
+      "step": 3729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499692456840101e-06,
+      "loss": 0.9382,
+      "step": 3730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494605938013883e-06,
+      "loss": 0.948,
+      "step": 3731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489520452971778e-06,
+      "loss": 0.8687,
+      "step": 3732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484436003213561e-06,
+      "loss": 0.8928,
+      "step": 3733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479352590238697e-06,
+      "loss": 0.9023,
+      "step": 3734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474270215546352e-06,
+      "loss": 0.8816,
+      "step": 3735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469188880635384e-06,
+      "loss": 0.9067,
+      "step": 3736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46410858700433e-06,
+      "loss": 0.886,
+      "step": 3737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4590293361514385e-06,
+      "loss": 0.9316,
+      "step": 3738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.9092,
+      "step": 3739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448873968771575e-06,
+      "loss": 0.9404,
+      "step": 3740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44379785523954e-06,
+      "loss": 0.9119,
+      "step": 3741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438722790475555e-06,
+      "loss": 0.9094,
+      "step": 3742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433648775976317e-06,
+      "loss": 0.9309,
+      "step": 3743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428575813238218e-06,
+      "loss": 0.6521,
+      "step": 3744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423503903757338e-06,
+      "loss": 0.8965,
+      "step": 3745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418433049029441e-06,
+      "loss": 0.8862,
+      "step": 3746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4133632505499876e-06,
+      "loss": 0.9658,
+      "step": 3747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4082945098141244e-06,
+      "loss": 0.9045,
+      "step": 3748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403226828316686e-06,
+      "loss": 0.8799,
+      "step": 3749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398160207552196e-06,
+      "loss": 0.9355,
+      "step": 3750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39309464901486e-06,
+      "loss": 0.9651,
+      "step": 3751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3880301541985746e-06,
+      "loss": 0.9014,
+      "step": 3752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382966724596922e-06,
+      "loss": 0.9067,
+      "step": 3753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3779043617031775e-06,
+      "loss": 0.9038,
+      "step": 3754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372843067010284e-06,
+      "loss": 1.0166,
+      "step": 3755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367782842010885e-06,
+      "loss": 0.6136,
+      "step": 3756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3627236881973e-06,
+      "loss": 0.8896,
+      "step": 3757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357665607061542e-06,
+      "loss": 0.9719,
+      "step": 3758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352608600095302e-06,
+      "loss": 0.9495,
+      "step": 3759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3475526687899445e-06,
+      "loss": 0.6155,
+      "step": 3760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342497814636536e-06,
+      "loss": 0.9189,
+      "step": 3761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3374440391258105e-06,
+      "loss": 0.9392,
+      "step": 3762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332391343748196e-06,
+      "loss": 0.9231,
+      "step": 3763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327339729993783e-06,
+      "loss": 0.9429,
+      "step": 3764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322289199352364e-06,
+      "loss": 0.9404,
+      "step": 3765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3172397533134e-06,
+      "loss": 0.9302,
+      "step": 3766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312191393366036e-06,
+      "loss": 0.9309,
+      "step": 3767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307144120999097e-06,
+      "loss": 0.939,
+      "step": 3768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302097937701083e-06,
+      "loss": 0.8469,
+      "step": 3769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297052844960178e-06,
+      "loss": 0.925,
+      "step": 3770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292008844264242e-06,
+      "loss": 0.9978,
+      "step": 3771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286965937100813e-06,
+      "loss": 0.9187,
+      "step": 3772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281924124957107e-06,
+      "loss": 0.9973,
+      "step": 3773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276883409320016e-06,
+      "loss": 0.9543,
+      "step": 3774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271843791676111e-06,
+      "loss": 0.6245,
+      "step": 3775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266805273511635e-06,
+      "loss": 0.9092,
+      "step": 3776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261767856312517e-06,
+      "loss": 0.8796,
+      "step": 3777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256731541564344e-06,
+      "loss": 0.9744,
+      "step": 3778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25169633075239e-06,
+      "loss": 0.8452,
+      "step": 3779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246662225361603e-06,
+      "loss": 0.8989,
+      "step": 3780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2416292268766075e-06,
+      "loss": 0.8872,
+      "step": 3781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236597336781689e-06,
+      "loss": 0.9204,
+      "step": 3782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231566556560814e-06,
+      "loss": 0.9109,
+      "step": 3783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226536887697628e-06,
+      "loss": 0.9585,
+      "step": 3784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221508331675441e-06,
+      "loss": 0.9253,
+      "step": 3785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216480889977236e-06,
+      "loss": 0.8506,
+      "step": 3786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.211454564085665e-06,
+      "loss": 0.9321,
+      "step": 3787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206429355483056e-06,
+      "loss": 0.9521,
+      "step": 3788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201405265651406e-06,
+      "loss": 0.95,
+      "step": 3789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1963822960723805e-06,
+      "loss": 0.9058,
+      "step": 3790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191360448227317e-06,
+      "loss": 0.9375,
+      "step": 3791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186339723597216e-06,
+      "loss": 0.9683,
+      "step": 3792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181320123662755e-06,
+      "loss": 0.8542,
+      "step": 3793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176301649904276e-06,
+      "loss": 0.9453,
+      "step": 3794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17128430380179e-06,
+      "loss": 0.9263,
+      "step": 3795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166268086834972e-06,
+      "loss": 0.9226,
+      "step": 3796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1612530004831675e-06,
+      "loss": 0.9514,
+      "step": 3797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156239046225386e-06,
+      "loss": 0.9092,
+      "step": 3798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151226225540306e-06,
+      "loss": 0.9636,
+      "step": 3799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146214539906277e-06,
+      "loss": 0.9077,
+      "step": 3800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141203990801296e-06,
+      "loss": 0.929,
+      "step": 3801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136194579703039e-06,
+      "loss": 0.9248,
+      "step": 3802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131186308088846e-06,
+      "loss": 1.0173,
+      "step": 3803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126179177435722e-06,
+      "loss": 0.9141,
+      "step": 3804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121173189220325e-06,
+      "loss": 0.9231,
+      "step": 3805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116168344918982e-06,
+      "loss": 0.9714,
+      "step": 3806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11116464600769e-06,
+      "loss": 0.8982,
+      "step": 3807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106162093962098e-06,
+      "loss": 0.9705,
+      "step": 3808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101160690257523e-06,
+      "loss": 0.8899,
+      "step": 3809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0961604363689364e-06,
+      "loss": 0.8921,
+      "step": 3810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091161333770978e-06,
+      "loss": 0.9111,
+      "step": 3811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086163383937946e-06,
+      "loss": 0.8638,
+      "step": 3812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081166588343798e-06,
+      "loss": 0.946,
+      "step": 3813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076170948462146e-06,
+      "loss": 0.8696,
+      "step": 3814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07117646576627e-06,
+      "loss": 0.8833,
+      "step": 3815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066183141729104e-06,
+      "loss": 0.894,
+      "step": 3816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061190977823242e-06,
+      "loss": 0.6212,
+      "step": 3817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056199975520936e-06,
+      "loss": 0.939,
+      "step": 3818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051210136294089e-06,
+      "loss": 0.9016,
+      "step": 3819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046221461614271e-06,
+      "loss": 0.8965,
+      "step": 3820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041233952952703e-06,
+      "loss": 0.9595,
+      "step": 3821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036247611780265e-06,
+      "loss": 0.6459,
+      "step": 3822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031262439567487e-06,
+      "loss": 0.908,
+      "step": 3823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026278437784558e-06,
+      "loss": 0.926,
+      "step": 3824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021295607901323e-06,
+      "loss": 0.6719,
+      "step": 3825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016313951387283e-06,
+      "loss": 0.9329,
+      "step": 3826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011333469711592e-06,
+      "loss": 0.8613,
+      "step": 3827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006354164343047e-06,
+      "loss": 0.9253,
+      "step": 3828
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001376036750111e-06,
+      "loss": 0.8923,
+      "step": 3829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996399088400895e-06,
+      "loss": 0.8813,
+      "step": 3830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9914233207631655e-06,
+      "loss": 1.0225,
+      "step": 3831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986448735304339e-06,
+      "loss": 0.9244,
+      "step": 3832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981475333491477e-06,
+      "loss": 0.8826,
+      "step": 3833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976503116791299e-06,
+      "loss": 0.8984,
+      "step": 3834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971532086670174e-06,
+      "loss": 0.9824,
+      "step": 3835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966562244594123e-06,
+      "loss": 0.9167,
+      "step": 3836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9615935920288096e-06,
+      "loss": 0.6074,
+      "step": 3837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956626130439552e-06,
+      "loss": 0.6006,
+      "step": 3838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951659861291317e-06,
+      "loss": 0.6201,
+      "step": 3839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946694786048719e-06,
+      "loss": 0.8733,
+      "step": 3840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941730906176022e-06,
+      "loss": 0.9888,
+      "step": 3841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936768223137133e-06,
+      "loss": 0.9468,
+      "step": 3842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931806738395607e-06,
+      "loss": 0.9597,
+      "step": 3843
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926846453414649e-06,
+      "loss": 0.8931,
+      "step": 3844
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921887369657113e-06,
+      "loss": 0.9426,
+      "step": 3845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916929488585488e-06,
+      "loss": 0.9277,
+      "step": 3846
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9119728116619165e-06,
+      "loss": 0.967,
+      "step": 3847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907017340348182e-06,
+      "loss": 1.019,
+      "step": 3848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9020630761057174e-06,
+      "loss": 0.9309,
+      "step": 3849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897110020395601e-06,
+      "loss": 0.885,
+      "step": 3850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892158174678538e-06,
+      "loss": 0.9377,
+      "step": 3851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887207540414899e-06,
+      "loss": 0.9109,
+      "step": 3852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8822581190646835e-06,
+      "loss": 0.8936,
+      "step": 3853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877309912087542e-06,
+      "loss": 0.9229,
+      "step": 3854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872362920942751e-06,
+      "loss": 0.9727,
+      "step": 3855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8674171470892506e-06,
+      "loss": 0.9453,
+      "step": 3856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862472591985605e-06,
+      "loss": 0.9177,
+      "step": 3857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857529257090027e-06,
+      "loss": 0.8569,
+      "step": 3858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852587143860368e-06,
+      "loss": 0.8536,
+      "step": 3859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847646253754115e-06,
+      "loss": 0.6185,
+      "step": 3860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8427065882284e-06,
+      "loss": 0.9429,
+      "step": 3861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837768148739989e-06,
+      "loss": 0.9329,
+      "step": 3862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832830936745294e-06,
+      "loss": 0.897,
+      "step": 3863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827894953700357e-06,
+      "loss": 0.8855,
+      "step": 3864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822960201060858e-06,
+      "loss": 0.8507,
+      "step": 3865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818026680282123e-06,
+      "loss": 0.9656,
+      "step": 3866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8130943928191e-06,
+      "loss": 0.9253,
+      "step": 3867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80816334012639e-06,
+      "loss": 0.9182,
+      "step": 3868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803233523658219e-06,
+      "loss": 0.7062,
+      "step": 3869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7983049448684444e-06,
+      "loss": 0.8999,
+      "step": 3870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793377605210575e-06,
+      "loss": 0.9377,
+      "step": 3871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788451506137736e-06,
+      "loss": 0.8994,
+      "step": 3872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783526649102701e-06,
+      "loss": 0.8616,
+      "step": 3873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7786030355578704e-06,
+      "loss": 0.9189,
+      "step": 3874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773680666955272e-06,
+      "loss": 0.9558,
+      "step": 3875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768759544746585e-06,
+      "loss": 0.9004,
+      "step": 3876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7638396703831e-06,
+      "loss": 0.9399,
+      "step": 3877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758921045315751e-06,
+      "loss": 0.8989,
+      "step": 3878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754003670995104e-06,
+      "loss": 0.9021,
+      "step": 3879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749087548871349e-06,
+      "loss": 0.9448,
+      "step": 3880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744172680394313e-06,
+      "loss": 0.8689,
+      "step": 3881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73925906701346e-06,
+      "loss": 0.9062,
+      "step": 3882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734346710177863e-06,
+      "loss": 0.8933,
+      "step": 3883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729435611336239e-06,
+      "loss": 0.9414,
+      "step": 3884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724525771936941e-06,
+      "loss": 0.8945,
+      "step": 3885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719617193427933e-06,
+      "loss": 0.9348,
+      "step": 3886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714709877256817e-06,
+      "loss": 0.9302,
+      "step": 3887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709803824870823e-06,
+      "loss": 0.9451,
+      "step": 3888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704899037716803e-06,
+      "loss": 0.6451,
+      "step": 3889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699995517241245e-06,
+      "loss": 0.9138,
+      "step": 3890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695093264890258e-06,
+      "loss": 0.9099,
+      "step": 3891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690192282109568e-06,
+      "loss": 0.6223,
+      "step": 3892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685292570344547e-06,
+      "loss": 0.9507,
+      "step": 3893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68039413104017e-06,
+      "loss": 0.9504,
+      "step": 3894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675496965641057e-06,
+      "loss": 0.9563,
+      "step": 3895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6706010755914375e-06,
+      "loss": 0.6168,
+      "step": 3896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665706462335167e-06,
+      "loss": 0.925,
+      "step": 3897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6608131273157366e-06,
+      "loss": 0.9465,
+      "step": 3898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655921071976242e-06,
+      "loss": 0.9556,
+      "step": 3899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651030297759419e-06,
+      "loss": 0.9746,
+      "step": 3900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646140806107614e-06,
+      "loss": 0.8967,
+      "step": 3901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641252598462794e-06,
+      "loss": 0.968,
+      "step": 3902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636365676266556e-06,
+      "loss": 0.9209,
+      "step": 3903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6314800409601176e-06,
+      "loss": 0.9338,
+      "step": 3904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6265956939843116e-06,
+      "loss": 0.9146,
+      "step": 3905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621712636779589e-06,
+      "loss": 0.9541,
+      "step": 3906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616830870786022e-06,
+      "loss": 0.9751,
+      "step": 3907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611950397443307e-06,
+      "loss": 0.8347,
+      "step": 3908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607071218190764e-06,
+      "loss": 0.949,
+      "step": 3909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602193334467307e-06,
+      "loss": 0.8911,
+      "step": 3910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597316747711495e-06,
+      "loss": 0.8894,
+      "step": 3911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592441459361493e-06,
+      "loss": 0.9805,
+      "step": 3912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587567470855079e-06,
+      "loss": 0.8982,
+      "step": 3913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582694783629663e-06,
+      "loss": 0.958,
+      "step": 3914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577823399122251e-06,
+      "loss": 0.8436,
+      "step": 3915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572953318769475e-06,
+      "loss": 0.8546,
+      "step": 3916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5680845440075885e-06,
+      "loss": 0.9006,
+      "step": 3917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563217076272449e-06,
+      "loss": 0.9341,
+      "step": 3918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55835091699953e-06,
+      "loss": 0.9106,
+      "step": 3919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55348606762393e-06,
+      "loss": 0.5874,
+      "step": 3920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548622529580343e-06,
+      "loss": 0.8638,
+      "step": 3921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543760304303098e-06,
+      "loss": 0.9238,
+      "step": 3922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538899393226122e-06,
+      "loss": 0.8904,
+      "step": 3923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534039797782948e-06,
+      "loss": 0.9612,
+      "step": 3924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529181519406745e-06,
+      "loss": 0.9238,
+      "step": 3925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524324559530269e-06,
+      "loss": 0.6134,
+      "step": 3926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519468919585908e-06,
+      "loss": 0.8638,
+      "step": 3927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514614601005641e-06,
+      "loss": 0.8472,
+      "step": 3928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509761605221067e-06,
+      "loss": 0.9319,
+      "step": 3929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504909933663398e-06,
+      "loss": 0.6211,
+      "step": 3930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5000595877634555e-06,
+      "loss": 0.9119,
+      "step": 3931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495210568951661e-06,
+      "loss": 0.6636,
+      "step": 3932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490362878658049e-06,
+      "loss": 0.8987,
+      "step": 3933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485516518312269e-06,
+      "loss": 0.9429,
+      "step": 3934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480671489343568e-06,
+      "loss": 0.8928,
+      "step": 3935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475827793180808e-06,
+      "loss": 0.9319,
+      "step": 3936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470985431252455e-06,
+      "loss": 0.8806,
+      "step": 3937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4661444049865755e-06,
+      "loss": 0.6482,
+      "step": 3938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461304715810858e-06,
+      "loss": 0.9631,
+      "step": 3939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456466365152577e-06,
+      "loss": 0.9224,
+      "step": 3940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451629354438631e-06,
+      "loss": 0.8684,
+      "step": 3941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4467936850955105e-06,
+      "loss": 0.8911,
+      "step": 3942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441959358549309e-06,
+      "loss": 0.9204,
+      "step": 3943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437126376225738e-06,
+      "loss": 0.9119,
+      "step": 3944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4322947395500955e-06,
+      "loss": 0.9478,
+      "step": 3945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4274644499473e-06,
+      "loss": 0.9138,
+      "step": 3946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422635508841859e-06,
+      "loss": 0.9456,
+      "step": 3947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4178079176578836e-06,
+      "loss": 0.8429,
+      "step": 3948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412981677819094e-06,
+      "loss": 0.9326,
+      "step": 3949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408156790748813e-06,
+      "loss": 0.8878,
+      "step": 3950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40333325786995e-06,
+      "loss": 0.9006,
+      "step": 3951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398511080605027e-06,
+      "loss": 0.9451,
+      "step": 3952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39369026037617e-06,
+      "loss": 0.6268,
+      "step": 3953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388870798605089e-06,
+      "loss": 0.9675,
+      "step": 3954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384052696713118e-06,
+      "loss": 0.9312,
+      "step": 3955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379235956121156e-06,
+      "loss": 0.5817,
+      "step": 3956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37442057824973e-06,
+      "loss": 0.9143,
+      "step": 3957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369606564518957e-06,
+      "loss": 0.9116,
+      "step": 3958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364793916348544e-06,
+      "loss": 0.9927,
+      "step": 3959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359982635157799e-06,
+      "loss": 0.9229,
+      "step": 3960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355172722365634e-06,
+      "loss": 0.9297,
+      "step": 3961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3503641793905485e-06,
+      "loss": 0.6482,
+      "step": 3962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345557007650642e-06,
+      "loss": 0.6104,
+      "step": 3963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340751208563611e-06,
+      "loss": 0.9294,
+      "step": 3964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3359467835467394e-06,
+      "loss": 0.8792,
+      "step": 3965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33114373401692e-06,
+      "loss": 0.8562,
+      "step": 3966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326342061390621e-06,
+      "loss": 0.9285,
+      "step": 3967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321541767083927e-06,
+      "loss": 0.8884,
+      "step": 3968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3167428525124976e-06,
+      "loss": 0.8572,
+      "step": 3969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311945319091589e-06,
+      "loss": 0.8962,
+      "step": 3970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307149168236061e-06,
+      "loss": 0.9236,
+      "step": 3971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30235440136035e-06,
+      "loss": 0.9307,
+      "step": 3972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297561019878501e-06,
+      "loss": 0.9382,
+      "step": 3973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292769025204137e-06,
+      "loss": 0.9343,
+      "step": 3974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28797841875047e-06,
+      "loss": 0.8844,
+      "step": 3975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283189201930316e-06,
+      "loss": 0.9082,
+      "step": 3976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2784013761560794e-06,
+      "loss": 0.9214,
+      "step": 3977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273614942839742e-06,
+      "loss": 0.9231,
+      "step": 3978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26882990339288e-06,
+      "loss": 0.8774,
+      "step": 3979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2640462592266695e-06,
+      "loss": 0.929,
+      "step": 3980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259264011751857e-06,
+      "loss": 0.9209,
+      "step": 3981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254483162378795e-06,
+      "loss": 0.9282,
+      "step": 3982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249703712517411e-06,
+      "loss": 0.8982,
+      "step": 3983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244925663577221e-06,
+      "loss": 0.9268,
+      "step": 3984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240149016967338e-06,
+      "loss": 0.9177,
+      "step": 3985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235373774096446e-06,
+      "loss": 0.9561,
+      "step": 3986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230599936372831e-06,
+      "loss": 0.9736,
+      "step": 3987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225827505204355e-06,
+      "loss": 0.9155,
+      "step": 3988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221056481998459e-06,
+      "loss": 0.8734,
+      "step": 3989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21628686816219e-06,
+      "loss": 0.8835,
+      "step": 3990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2115186651021575e-06,
+      "loss": 0.9333,
+      "step": 3991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2067518742245615e-06,
+      "loss": 0.6146,
+      "step": 3992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201986496935196e-06,
+      "loss": 0.6089,
+      "step": 3993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1972225346394216e-06,
+      "loss": 0.8958,
+      "step": 3994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192459988742193e-06,
+      "loss": 0.6161,
+      "step": 3995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187698860648054e-06,
+      "loss": 0.8884,
+      "step": 3996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182939151761102e-06,
+      "loss": 0.969,
+      "step": 3997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178180863485043e-06,
+      "loss": 0.9033,
+      "step": 3998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173423997223159e-06,
+      "loss": 0.9497,
+      "step": 3999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168668554378305e-06,
+      "loss": 0.9187,
+      "step": 4000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163914536352919e-06,
+      "loss": 0.9177,
+      "step": 4001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159161944549017e-06,
+      "loss": 0.9534,
+      "step": 4002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154410780368202e-06,
+      "loss": 0.9465,
+      "step": 4003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1496610452116516e-06,
+      "loss": 0.9409,
+      "step": 4004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144912740480122e-06,
+      "loss": 0.8955,
+      "step": 4005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14016586757394e-06,
+      "loss": 0.894,
+      "step": 4006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135420427893026e-06,
+      "loss": 0.9568,
+      "step": 4007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13067642283686e-06,
+      "loss": 0.947,
+      "step": 4008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125933853804519e-06,
+      "loss": 0.938,
+      "step": 4009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121192722194637e-06,
+      "loss": 0.9114,
+      "step": 4010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116453029405431e-06,
+      "loss": 0.9324,
+      "step": 4011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111714776834704e-06,
+      "loss": 0.9109,
+      "step": 4012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106977965879816e-06,
+      "loss": 0.8718,
+      "step": 4013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1022425979377174e-06,
+      "loss": 0.856,
+      "step": 4014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0975086744049265e-06,
+      "loss": 0.9265,
+      "step": 4015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092776196677529e-06,
+      "loss": 0.9741,
+      "step": 4016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088045166151201e-06,
+      "loss": 0.9553,
+      "step": 4017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083315584221174e-06,
+      "loss": 0.9204,
+      "step": 4018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078587452282267e-06,
+      "loss": 0.8936,
+      "step": 4019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07386077172886e-06,
+      "loss": 0.9064,
+      "step": 4020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069135543954909e-06,
+      "loss": 0.8936,
+      "step": 4021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064411770353941e-06,
+      "loss": 0.927,
+      "step": 4022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059689452319068e-06,
+      "loss": 0.9099,
+      "step": 4023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0549685912429434e-06,
+      "loss": 0.9202,
+      "step": 4024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050249188517814e-06,
+      "loss": 0.8882,
+      "step": 4025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.045531245535495e-06,
+      "loss": 0.9302,
+      "step": 4026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040814763687358e-06,
+      "loss": 0.918,
+      "step": 4027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03609974436436e-06,
+      "loss": 0.8843,
+      "step": 4028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031386188957013e-06,
+      "loss": 0.908,
+      "step": 4029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026674098855402e-06,
+      "loss": 0.9258,
+      "step": 4030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021963475449185e-06,
+      "loss": 0.9253,
+      "step": 4031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017254320127584e-06,
+      "loss": 0.9031,
+      "step": 4032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012546634279378e-06,
+      "loss": 0.9661,
+      "step": 4033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007840419292933e-06,
+      "loss": 0.9614,
+      "step": 4034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0031356765561625e-06,
+      "loss": 0.939,
+      "step": 4035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998432407456559e-06,
+      "loss": 0.8911,
+      "step": 4036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993730613381173e-06,
+      "loss": 0.957,
+      "step": 4037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989030295716615e-06,
+      "loss": 0.8914,
+      "step": 4038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984331455849077e-06,
+      "loss": 0.9241,
+      "step": 4039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979634095164298e-06,
+      "loss": 1.0222,
+      "step": 4040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974938215047595e-06,
+      "loss": 0.9402,
+      "step": 4041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970243816883835e-06,
+      "loss": 0.938,
+      "step": 4042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965550902057453e-06,
+      "loss": 0.8877,
+      "step": 4043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960859471952451e-06,
+      "loss": 0.6579,
+      "step": 4044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956169527952392e-06,
+      "loss": 0.8811,
+      "step": 4045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951481071440398e-06,
+      "loss": 0.8926,
+      "step": 4046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946794103799146e-06,
+      "loss": 0.9053,
+      "step": 4047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94210862641089e-06,
+      "loss": 0.9739,
+      "step": 4048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9374246406574254e-06,
+      "loss": 0.9236,
+      "step": 4049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932742147920128e-06,
+      "loss": 0.9377,
+      "step": 4050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928061149579919e-06,
+      "loss": 0.6218,
+      "step": 4051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923381647017278e-06,
+      "loss": 0.9001,
+      "step": 4052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918703641612255e-06,
+      "loss": 0.9468,
+      "step": 4053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914027134744446e-06,
+      "loss": 0.9077,
+      "step": 4054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909352127793016e-06,
+      "loss": 0.9211,
+      "step": 4055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904678622136682e-06,
+      "loss": 0.9275,
+      "step": 4056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900006619153711e-06,
+      "loss": 0.8899,
+      "step": 4057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895336120221945e-06,
+      "loss": 0.6285,
+      "step": 4058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890667126718763e-06,
+      "loss": 0.9175,
+      "step": 4059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885999640021117e-06,
+      "loss": 0.9722,
+      "step": 4060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881333661505504e-06,
+      "loss": 0.8865,
+      "step": 4061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876669192547972e-06,
+      "loss": 0.967,
+      "step": 4062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872006234524138e-06,
+      "loss": 0.8872,
+      "step": 4063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867344788809172e-06,
+      "loss": 0.9502,
+      "step": 4064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862684856777776e-06,
+      "loss": 0.979,
+      "step": 4065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8580264398042355e-06,
+      "loss": 0.9539,
+      "step": 4066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853369539262365e-06,
+      "loss": 0.9114,
+      "step": 4067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848714156525547e-06,
+      "loss": 0.8867,
+      "step": 4068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844060292966719e-06,
+      "loss": 0.8347,
+      "step": 4069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839407949958349e-06,
+      "loss": 0.8398,
+      "step": 4070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834757128872477e-06,
+      "loss": 0.9077,
+      "step": 4071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830107831080692e-06,
+      "loss": 0.9182,
+      "step": 4072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825460057954124e-06,
+      "loss": 0.9343,
+      "step": 4073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820813810863457e-06,
+      "loss": 0.863,
+      "step": 4074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816169091178935e-06,
+      "loss": 0.9075,
+      "step": 4075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811525900270334e-06,
+      "loss": 0.6407,
+      "step": 4076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806884239506996e-06,
+      "loss": 0.9248,
+      "step": 4077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802244110257802e-06,
+      "loss": 0.8879,
+      "step": 4078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797605513891179e-06,
+      "loss": 0.8516,
+      "step": 4079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792968451775113e-06,
+      "loss": 0.8743,
+      "step": 4080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788332925277126e-06,
+      "loss": 0.9041,
+      "step": 4081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783698935764297e-06,
+      "loss": 0.9341,
+      "step": 4082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7790664846032455e-06,
+      "loss": 0.9275,
+      "step": 4083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774435573160132e-06,
+      "loss": 0.9631,
+      "step": 4084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769806202800681e-06,
+      "loss": 0.5933,
+      "step": 4085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76517837489014e-06,
+      "loss": 0.9143,
+      "step": 4086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760552090793322e-06,
+      "loss": 0.9263,
+      "step": 4087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755927351874572e-06,
+      "loss": 0.8848,
+      "step": 4088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7513041594977785e-06,
+      "loss": 0.8987,
+      "step": 4089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746682515026382e-06,
+      "loss": 0.9043,
+      "step": 4090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742062419823367e-06,
+      "loss": 0.9666,
+      "step": 4091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737443875251251e-06,
+      "loss": 0.9265,
+      "step": 4092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732826882672099e-06,
+      "loss": 0.8865,
+      "step": 4093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728211443447525e-06,
+      "loss": 0.925,
+      "step": 4094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7235975589386715e-06,
+      "loss": 0.9658,
+      "step": 4095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718985230506239e-06,
+      "loss": 0.8762,
+      "step": 4096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714374459510456e-06,
+      "loss": 0.9558,
+      "step": 4097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709765247311092e-06,
+      "loss": 0.8647,
+      "step": 4098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7051575952674675e-06,
+      "loss": 0.926,
+      "step": 4099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700551504738431e-06,
+      "loss": 0.929,
+      "step": 4100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695946977082381e-06,
+      "loss": 0.9724,
+      "step": 4101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691344013657245e-06,
+      "loss": 0.9583,
+      "step": 4102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686742615820493e-06,
+      "loss": 0.9402,
+      "step": 4103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682142784929142e-06,
+      "loss": 0.8665,
+      "step": 4104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6775445223397306e-06,
+      "loss": 0.8821,
+      "step": 4105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672947829408346e-06,
+      "loss": 0.8425,
+      "step": 4106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668352707490611e-06,
+      "loss": 0.9197,
+      "step": 4107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663759157941682e-06,
+      "loss": 0.9487,
+      "step": 4108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6591671821162535e-06,
+      "loss": 0.9456,
+      "step": 4109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654576781368566e-06,
+      "loss": 0.9175,
+      "step": 4110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6499879570523684e-06,
+      "loss": 0.8682,
+      "step": 4111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645400710520972e-06,
+      "loss": 0.8779,
+      "step": 4112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6408150431272135e-06,
+      "loss": 0.9622,
+      "step": 4113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636230956223462e-06,
+      "loss": 0.8954,
+      "step": 4114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6316484511616185e-06,
+      "loss": 0.9448,
+      "step": 4115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627067529293119e-06,
+      "loss": 0.9768,
+      "step": 4116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622488191968938e-06,
+      "loss": 0.905,
+      "step": 4117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61791044053958e-06,
+      "loss": 0.9636,
+      "step": 4118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613334276355079e-06,
+      "loss": 0.9043,
+      "step": 4119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608759700764998e-06,
+      "loss": 0.6307,
+      "step": 4120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604186715118445e-06,
+      "loss": 0.958,
+      "step": 4121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599615320764041e-06,
+      "loss": 0.5754,
+      "step": 4122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595045519049954e-06,
+      "loss": 0.2825,
+      "step": 4123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590477311323872e-06,
+      "loss": 0.9021,
+      "step": 4124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585910698933013e-06,
+      "loss": 0.8733,
+      "step": 4125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581345683224133e-06,
+      "loss": 0.9082,
+      "step": 4126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576782265543507e-06,
+      "loss": 0.887,
+      "step": 4127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572220447236948e-06,
+      "loss": 0.614,
+      "step": 4128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567660229649791e-06,
+      "loss": 0.8713,
+      "step": 4129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563101614126895e-06,
+      "loss": 0.874,
+      "step": 4130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5585446020126634e-06,
+      "loss": 0.8668,
+      "step": 4131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553989194651003e-06,
+      "loss": 0.9021,
+      "step": 4132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549435393385371e-06,
+      "loss": 0.9163,
+      "step": 4133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544883199558735e-06,
+      "loss": 0.9067,
+      "step": 4134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540332614513589e-06,
+      "loss": 0.5795,
+      "step": 4135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535783639591961e-06,
+      "loss": 0.9265,
+      "step": 4136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5312362761354065e-06,
+      "loss": 0.8699,
+      "step": 4137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526690525484987e-06,
+      "loss": 0.9556,
+      "step": 4138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522146388981306e-06,
+      "loss": 0.8699,
+      "step": 4139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51760386796449e-06,
+      "loss": 0.9556,
+      "step": 4140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513062963774176e-06,
+      "loss": 0.8992,
+      "step": 4141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508523677749542e-06,
+      "loss": 0.8838,
+      "step": 4142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5039860112292755e-06,
+      "loss": 0.9194,
+      "step": 4143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4994499655515865e-06,
+      "loss": 0.6472,
+      "step": 4144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494915542054219e-06,
+      "loss": 0.8323,
+      "step": 4145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490382742074428e-06,
+      "loss": 0.9368,
+      "step": 4146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485851566948985e-06,
+      "loss": 0.9475,
+      "step": 4147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4813220180141995e-06,
+      "loss": 0.8801,
+      "step": 4148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476794096605884e-06,
+      "loss": 0.9834,
+      "step": 4149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472267804059387e-06,
+      "loss": 0.9202,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467743141709563e-06,
+      "loss": 0.9077,
+      "step": 4151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463220110890787e-06,
+      "loss": 0.8928,
+      "step": 4152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458698712936966e-06,
+      "loss": 0.8801,
+      "step": 4153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454178949181507e-06,
+      "loss": 0.9258,
+      "step": 4154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4496608209573485e-06,
+      "loss": 0.8801,
+      "step": 4155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445144329596951e-06,
+      "loss": 0.6039,
+      "step": 4156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440629476432268e-06,
+      "loss": 0.9333,
+      "step": 4157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436116262794793e-06,
+      "loss": 0.9006,
+      "step": 4158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431604690015532e-06,
+      "loss": 0.9424,
+      "step": 4159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427094759425e-06,
+      "loss": 0.9094,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422586472353229e-06,
+      "loss": 0.927,
+      "step": 4161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418079830129773e-06,
+      "loss": 0.8923,
+      "step": 4162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413574834083694e-06,
+      "loss": 0.9734,
+      "step": 4163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4090714855435746e-06,
+      "loss": 0.6283,
+      "step": 4164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404569785837505e-06,
+      "loss": 0.9395,
+      "step": 4165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40006973629309e-06,
+      "loss": 0.9341,
+      "step": 4166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395571338237455e-06,
+      "loss": 0.949,
+      "step": 4167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391074592997228e-06,
+      "loss": 0.9236,
+      "step": 4168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386579501898563e-06,
+      "loss": 0.9285,
+      "step": 4169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38208606626711e-06,
+      "loss": 0.9167,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37759428742804e-06,
+      "loss": 0.5825,
+      "step": 4171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373104166706039e-06,
+      "loss": 0.9683,
+      "step": 4172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368615705425292e-06,
+      "loss": 0.6125,
+      "step": 4173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364128904909509e-06,
+      "loss": 0.937,
+      "step": 4174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359643766481901e-06,
+      "loss": 0.8611,
+      "step": 4175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355160291465186e-06,
+      "loss": 0.941,
+      "step": 4176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3506784811815995e-06,
+      "loss": 0.9192,
+      "step": 4177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346198336952892e-06,
+      "loss": 0.9678,
+      "step": 4178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341719860100296e-06,
+      "loss": 0.9189,
+      "step": 4179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337243051944583e-06,
+      "loss": 0.9436,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3327679138060105e-06,
+      "loss": 0.9966,
+      "step": 4181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3282944470043576e-06,
+      "loss": 0.9558,
+      "step": 4182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323822652858911e-06,
+      "loss": 0.9177,
+      "step": 4183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.8933,
+      "step": 4184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314884087811256e-06,
+      "loss": 0.9646,
+      "step": 4185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310417319545154e-06,
+      "loss": 0.8154,
+      "step": 4186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305952229207436e-06,
+      "loss": 0.8711,
+      "step": 4187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30148881811491e-06,
+      "loss": 0.8914,
+      "step": 4188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297027087583901e-06,
+      "loss": 0.8894,
+      "step": 4189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292567038930218e-06,
+      "loss": 0.9258,
+      "step": 4190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288108673469192e-06,
+      "loss": 0.8691,
+      "step": 4191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283651992515649e-06,
+      "loss": 0.876,
+      "step": 4192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279196997383912e-06,
+      "loss": 0.9592,
+      "step": 4193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274743689387826e-06,
+      "loss": 0.6193,
+      "step": 4194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270292069840716e-06,
+      "loss": 0.938,
+      "step": 4195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265842140055428e-06,
+      "loss": 0.9087,
+      "step": 4196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261393901344299e-06,
+      "loss": 0.9575,
+      "step": 4197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256947355019163e-06,
+      "loss": 0.8726,
+      "step": 4198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252502502391373e-06,
+      "loss": 0.8848,
+      "step": 4199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24805934477176e-06,
+      "loss": 0.8645,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243617883470675e-06,
+      "loss": 0.6642,
+      "step": 4201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2391781197979575e-06,
+      "loss": 0.821,
+      "step": 4202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2347400550629435e-06,
+      "loss": 0.8547,
+      "step": 4203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230303690574478e-06,
+      "loss": 0.967,
+      "step": 4204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225869027640902e-06,
+      "loss": 0.9355,
+      "step": 4205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22143606757005e-06,
+      "loss": 0.899,
+      "step": 4206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217004811669253e-06,
+      "loss": 0.9216,
+      "step": 4207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212575261245351e-06,
+      "loss": 0.9473,
+      "step": 4208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208147417604665e-06,
+      "loss": 0.8923,
+      "step": 4209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2037212820530285e-06,
+      "loss": 0.8911,
+      "step": 4210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19929685589576e-06,
+      "loss": 0.9424,
+      "step": 4211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194874140437675e-06,
+      "loss": 0.8342,
+      "step": 4212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190453136983091e-06,
+      "loss": 0.9487,
+      "step": 4213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186033846835814e-06,
+      "loss": 0.8918,
+      "step": 4214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181616271299153e-06,
+      "loss": 0.8809,
+      "step": 4215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1772004116759004e-06,
+      "loss": 0.5856,
+      "step": 4216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172786269268345e-06,
+      "loss": 0.9194,
+      "step": 4217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1683738453782795e-06,
+      "loss": 0.9099,
+      "step": 4218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16396314130698e-06,
+      "loss": 0.9194,
+      "step": 4219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1595541583552125e-06,
+      "loss": 0.6039,
+      "step": 4220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155146897823249e-06,
+      "loss": 0.9163,
+      "step": 4221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150741361010837e-06,
+      "loss": 0.9209,
+      "step": 4222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1463375492172285e-06,
+      "loss": 0.9014,
+      "step": 4223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141935463741168e-06,
+      "loss": 0.9492,
+      "step": 4224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137535105880872e-06,
+      "loss": 0.9243,
+      "step": 4225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1331364769340665e-06,
+      "loss": 0.9534,
+      "step": 4226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128739578197965e-06,
+      "loss": 0.865,
+      "step": 4227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124344410969263e-06,
+      "loss": 0.9087,
+      "step": 4228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119950976544158e-06,
+      "loss": 0.9346,
+      "step": 4229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1155592762183135e-06,
+      "loss": 0.9551,
+      "step": 4230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111169311286905e-06,
+      "loss": 0.9006,
+      "step": 4231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106781083044591e-06,
+      "loss": 0.9216,
+      "step": 4232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102394592785509e-06,
+      "loss": 0.9031,
+      "step": 4233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098009841803288e-06,
+      "loss": 0.9097,
+      "step": 4234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093626831391051e-06,
+      "loss": 0.8918,
+      "step": 4235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089245562841396e-06,
+      "loss": 0.9663,
+      "step": 4236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0848660374464185e-06,
+      "loss": 0.9475,
+      "step": 4237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080488256497694e-06,
+      "loss": 0.9009,
+      "step": 4238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076112221286278e-06,
+      "loss": 0.8708,
+      "step": 4239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071737933102726e-06,
+      "loss": 0.8689,
+      "step": 4240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067365393237064e-06,
+      "loss": 0.8955,
+      "step": 4241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062994602978812e-06,
+      "loss": 0.8684,
+      "step": 4242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05862556361697e-06,
+      "loss": 0.8726,
+      "step": 4243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054258276440016e-06,
+      "loss": 0.8911,
+      "step": 4244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049892742735924e-06,
+      "loss": 0.9468,
+      "step": 4245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045528963792139e-06,
+      "loss": 0.6417,
+      "step": 4246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041166940895599e-06,
+      "loss": 0.9114,
+      "step": 4247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036806675332715e-06,
+      "loss": 0.9304,
+      "step": 4248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03244816838938e-06,
+      "loss": 0.9487,
+      "step": 4249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028091421350975e-06,
+      "loss": 0.9377,
+      "step": 4250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023736435502367e-06,
+      "loss": 0.8965,
+      "step": 4251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019383212127879e-06,
+      "loss": 0.8652,
+      "step": 4252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015031752511338e-06,
+      "loss": 0.988,
+      "step": 4253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0106820579360485e-06,
+      "loss": 0.9214,
+      "step": 4254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006334129684781e-06,
+      "loss": 0.8838,
+      "step": 4255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001987969039801e-06,
+      "loss": 0.9419,
+      "step": 4256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99764357728284e-06,
+      "loss": 0.64,
+      "step": 4257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993300955695112e-06,
+      "loss": 0.9407,
+      "step": 4258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9889601055573145e-06,
+      "loss": 0.9062,
+      "step": 4259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984621028149617e-06,
+      "loss": 0.96,
+      "step": 4260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98028372475166e-06,
+      "loss": 0.8875,
+      "step": 4261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975948196642578e-06,
+      "loss": 0.9519,
+      "step": 4262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9716144451009655e-06,
+      "loss": 0.6194,
+      "step": 4263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967282471404903e-06,
+      "loss": 0.9204,
+      "step": 4264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962952276831943e-06,
+      "loss": 0.8683,
+      "step": 4265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958623862659108e-06,
+      "loss": 0.8687,
+      "step": 4266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954297230162908e-06,
+      "loss": 0.9841,
+      "step": 4267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9499723806193145e-06,
+      "loss": 0.9182,
+      "step": 4268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9456493153037835e-06,
+      "loss": 0.9983,
+      "step": 4269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941328035491245e-06,
+      "loss": 0.9309,
+      "step": 4270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937008542456085e-06,
+      "loss": 0.8762,
+      "step": 4271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932690837472184e-06,
+      "loss": 0.9546,
+      "step": 4272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9283749218128885e-06,
+      "loss": 0.8794,
+      "step": 4273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924060796751012e-06,
+      "loss": 0.8665,
+      "step": 4274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91974846355884e-06,
+      "loss": 0.6136,
+      "step": 4275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91543792350814e-06,
+      "loss": 0.9402,
+      "step": 4276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911129177870139e-06,
+      "loss": 0.9197,
+      "step": 4277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9068222279155425e-06,
+      "loss": 0.936,
+      "step": 4278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902517074914522e-06,
+      "loss": 0.9053,
+      "step": 4279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898213720136716e-06,
+      "loss": 0.6481,
+      "step": 4280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893912164851246e-06,
+      "loss": 0.9023,
+      "step": 4281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889612410326685e-06,
+      "loss": 0.9548,
+      "step": 4282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885314457831091e-06,
+      "loss": 0.928,
+      "step": 4283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881018308631981e-06,
+      "loss": 0.9568,
+      "step": 4284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876723963996338e-06,
+      "loss": 0.8945,
+      "step": 4285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872431425190626e-06,
+      "loss": 0.8879,
+      "step": 4286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8681406934807585e-06,
+      "loss": 0.9617,
+      "step": 4287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863851770132134e-06,
+      "loss": 0.8634,
+      "step": 4288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859564656409606e-06,
+      "loss": 0.8757,
+      "step": 4289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855279353577493e-06,
+      "loss": 0.9216,
+      "step": 4290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850995862899587e-06,
+      "loss": 0.9402,
+      "step": 4291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84671418563915e-06,
+      "loss": 0.9216,
+      "step": 4292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842434323058888e-06,
+      "loss": 0.6663,
+      "step": 4293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838156276420994e-06,
+      "loss": 0.9587,
+      "step": 4294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833880046987114e-06,
+      "loss": 0.885,
+      "step": 4295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8296056360183605e-06,
+      "loss": 0.9683,
+      "step": 4296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82533304477532e-06,
+      "loss": 0.8925,
+      "step": 4297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821062274518017e-06,
+      "loss": 0.9136,
+      "step": 4298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816793326505961e-06,
+      "loss": 0.9241,
+      "step": 4299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8125262019981224e-06,
+      "loss": 0.9448,
+      "step": 4300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8082609022529226e-06,
+      "loss": 0.9451,
+      "step": 4301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039974285282567e-06,
+      "loss": 0.9414,
+      "step": 4302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799735782081473e-06,
+      "loss": 0.9011,
+      "step": 4303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7954759641693795e-06,
+      "loss": 0.8889,
+      "step": 4304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7912179760482582e-06,
+      "loss": 0.9075,
+      "step": 4305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7869618189738377e-06,
+      "loss": 0.9473,
+      "step": 4306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7827074942013085e-06,
+      "loss": 0.8718,
+      "step": 4307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7784550029853297e-06,
+      "loss": 0.9263,
+      "step": 4308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774204346580008e-06,
+      "loss": 0.5446,
+      "step": 4309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769955526238921e-06,
+      "loss": 0.8662,
+      "step": 4310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7657085432150942e-06,
+      "loss": 0.9102,
+      "step": 4311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7614633987610116e-06,
+      "loss": 0.8933,
+      "step": 4312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7572200941286284e-06,
+      "loss": 0.8943,
+      "step": 4313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7529786305693384e-06,
+      "loss": 0.8633,
+      "step": 4314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7487390093340102e-06,
+      "loss": 0.8979,
+      "step": 4315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7445012316729556e-06,
+      "loss": 0.9529,
+      "step": 4316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7402652988359454e-06,
+      "loss": 0.989,
+      "step": 4317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736031212072211e-06,
+      "loss": 0.6322,
+      "step": 4318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7317989726304416e-06,
+      "loss": 0.8821,
+      "step": 4319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7275685817587738e-06,
+      "loss": 0.874,
+      "step": 4320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723340040704797e-06,
+      "loss": 0.9402,
+      "step": 4321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7191133507155697e-06,
+      "loss": 0.8967,
+      "step": 4322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7148885130375858e-06,
+      "loss": 0.8555,
+      "step": 4323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7106655289168105e-06,
+      "loss": 0.9509,
+      "step": 4324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70644439959865e-06,
+      "loss": 0.8489,
+      "step": 4325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702225126327965e-06,
+      "loss": 0.9163,
+      "step": 4326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6980077103490773e-06,
+      "loss": 0.9355,
+      "step": 4327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6937921529057497e-06,
+      "loss": 0.8962,
+      "step": 4328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6895784552412084e-06,
+      "loss": 0.908,
+      "step": 4329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6853666185981206e-06,
+      "loss": 0.8813,
+      "step": 4330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6811566442186064e-06,
+      "loss": 0.9185,
+      "step": 4331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6769485333442466e-06,
+      "loss": 0.8943,
+      "step": 4332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6727422872160614e-06,
+      "loss": 0.9275,
+      "step": 4333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668537907074522e-06,
+      "loss": 0.9026,
+      "step": 4334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6643353941595582e-06,
+      "loss": 0.8889,
+      "step": 4335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6601347497105366e-06,
+      "loss": 0.9041,
+      "step": 4336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6559359749662828e-06,
+      "loss": 0.9141,
+      "step": 4337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6517390711650737e-06,
+      "loss": 0.905,
+      "step": 4338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.647544039544615e-06,
+      "loss": 0.9373,
+      "step": 4339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643350881342085e-06,
+      "loss": 0.9453,
+      "step": 4340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6391595977940884e-06,
+      "loss": 0.9192,
+      "step": 4341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6349701901366916e-06,
+      "loss": 0.9155,
+      "step": 4342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6307826596054108e-06,
+      "loss": 0.896,
+      "step": 4343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6265970074351843e-06,
+      "loss": 0.9302,
+      "step": 4344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6224132348604213e-06,
+      "loss": 0.9558,
+      "step": 4345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618231343114972e-06,
+      "loss": 0.9082,
+      "step": 4346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.614051333432124e-06,
+      "loss": 0.9939,
+      "step": 4347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6098732070446084e-06,
+      "loss": 0.884,
+      "step": 4348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6056969651846176e-06,
+      "loss": 0.9509,
+      "step": 4349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6015226090837664e-06,
+      "loss": 0.9133,
+      "step": 4350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5973501399731326e-06,
+      "loss": 0.9036,
+      "step": 4351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593179559083225e-06,
+      "loss": 0.9031,
+      "step": 4352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589010867643997e-06,
+      "loss": 0.8735,
+      "step": 4353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5848440668848538e-06,
+      "loss": 0.9302,
+      "step": 4354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580679158034629e-06,
+      "loss": 0.9395,
+      "step": 4355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5765161423216145e-06,
+      "loss": 0.5742,
+      "step": 4356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572355020973529e-06,
+      "loss": 0.9888,
+      "step": 4357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568195795217537e-06,
+      "loss": 0.8884,
+      "step": 4358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564038466280252e-06,
+      "loss": 0.9631,
+      "step": 4359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5598830353877158e-06,
+      "loss": 0.8801,
+      "step": 4360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5557295037654227e-06,
+      "loss": 0.9258,
+      "step": 4361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5515778726382967e-06,
+      "loss": 0.8943,
+      "step": 4362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5474281432307024e-06,
+      "loss": 0.8726,
+      "step": 4363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54328031676645e-06,
+      "loss": 0.8853,
+      "step": 4364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391343944687906e-06,
+      "loss": 0.9246,
+      "step": 4365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5349903775603955e-06,
+      "loss": 0.9045,
+      "step": 4366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.530848267263394e-06,
+      "loss": 0.6516,
+      "step": 4367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5267080647993468e-06,
+      "loss": 0.864,
+      "step": 4368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522569771389246e-06,
+      "loss": 0.9043,
+      "step": 4369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5184333882535314e-06,
+      "loss": 0.9038,
+      "step": 4370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514298916612071e-06,
+      "loss": 0.8931,
+      "step": 4371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510166357684168e-06,
+      "loss": 0.9102,
+      "step": 4372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5060357126885725e-06,
+      "loss": 0.8833,
+      "step": 4373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5019069828434547e-06,
+      "loss": 0.9097,
+      "step": 4374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497780169366436e-06,
+      "loss": 0.9016,
+      "step": 4375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493655273474561e-06,
+      "loss": 0.9155,
+      "step": 4376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489532296384309e-06,
+      "loss": 0.9431,
+      "step": 4377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4854112393116047e-06,
+      "loss": 0.8663,
+      "step": 4378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481292103471794e-06,
+      "loss": 0.917,
+      "step": 4379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4771748900796597e-06,
+      "loss": 0.8623,
+      "step": 4380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473059600349424e-06,
+      "loss": 0.9219,
+      "step": 4381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4689462354947312e-06,
+      "loss": 0.9333,
+      "step": 4382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4648347967286655e-06,
+      "loss": 0.938,
+      "step": 4383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4607252852637494e-06,
+      "loss": 0.5989,
+      "step": 4384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.456617702311914e-06,
+      "loss": 0.9084,
+      "step": 4385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4525120490845444e-06,
+      "loss": 0.9116,
+      "step": 4386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448408326792451e-06,
+      "loss": 0.9326,
+      "step": 4387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4443065366458683e-06,
+      "loss": 0.855,
+      "step": 4388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4402066798544664e-06,
+      "loss": 0.9407,
+      "step": 4389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43610875762734e-06,
+      "loss": 0.8859,
+      "step": 4390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432012771173021e-06,
+      "loss": 0.6025,
+      "step": 4391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427918721699469e-06,
+      "loss": 0.5733,
+      "step": 4392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423826610414066e-06,
+      "loss": 0.9214,
+      "step": 4393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4197364385236255e-06,
+      "loss": 0.8795,
+      "step": 4394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415648207234393e-06,
+      "loss": 0.9822,
+      "step": 4395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4115619177520343e-06,
+      "loss": 0.826,
+      "step": 4396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4074775712816533e-06,
+      "loss": 0.9768,
+      "step": 4397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40339516902777e-06,
+      "loss": 0.8765,
+      "step": 4398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3993147121943324e-06,
+      "loss": 0.842,
+      "step": 4399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395236201984725e-06,
+      "loss": 0.9458,
+      "step": 4400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3911596396017433e-06,
+      "loss": 0.9185,
+      "step": 4401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387085026247624e-06,
+      "loss": 0.9912,
+      "step": 4402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3830123631240163e-06,
+      "loss": 0.8435,
+      "step": 4403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378941651431996e-06,
+      "loss": 0.9431,
+      "step": 4404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374872892372074e-06,
+      "loss": 0.9209,
+      "step": 4405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370806087144174e-06,
+      "loss": 0.6333,
+      "step": 4406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366741236947644e-06,
+      "loss": 0.938,
+      "step": 4407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3626783429812635e-06,
+      "loss": 0.9065,
+      "step": 4408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3586174064432262e-06,
+      "loss": 0.885,
+      "step": 4409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.354558428531154e-06,
+      "loss": 0.8953,
+      "step": 4410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3505014104420976e-06,
+      "loss": 0.8662,
+      "step": 4411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.346446353372508e-06,
+      "loss": 0.8918,
+      "step": 4412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342393258518278e-06,
+      "loss": 0.6367,
+      "step": 4413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338342127074721e-06,
+      "loss": 0.9531,
+      "step": 4414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334292960236556e-06,
+      "loss": 0.9062,
+      "step": 4415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330245759197941e-06,
+      "loss": 0.8923,
+      "step": 4416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326200525152441e-06,
+      "loss": 0.6249,
+      "step": 4417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322157259293044e-06,
+      "loss": 0.8853,
+      "step": 4418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.318115962812165e-06,
+      "loss": 0.9165,
+      "step": 4419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31407663690163e-06,
+      "loss": 0.9409,
+      "step": 4420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3100392827526807e-06,
+      "loss": 0.8799,
+      "step": 4421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306003901555991e-06,
+      "loss": 0.6305,
+      "step": 4422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3019704945016385e-06,
+      "loss": 0.9263,
+      "step": 4423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297939062779131e-06,
+      "loss": 0.6028,
+      "step": 4424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2939096075773835e-06,
+      "loss": 0.896,
+      "step": 4425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2898821300847294e-06,
+      "loss": 0.8801,
+      "step": 4426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2858566314889295e-06,
+      "loss": 0.663,
+      "step": 4427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281833112977145e-06,
+      "loss": 0.9023,
+      "step": 4428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2778115757359695e-06,
+      "loss": 0.8638,
+      "step": 4429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2737920209513994e-06,
+      "loss": 0.8831,
+      "step": 4430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269774449808849e-06,
+      "loss": 0.939,
+      "step": 4431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2657588634931516e-06,
+      "loss": 0.8604,
+      "step": 4432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.261745263188557e-06,
+      "loss": 0.9058,
+      "step": 4433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2577336500787237e-06,
+      "loss": 0.926,
+      "step": 4434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253724025346723e-06,
+      "loss": 0.9849,
+      "step": 4435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249716390175046e-06,
+      "loss": 0.6321,
+      "step": 4436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2457107457455915e-06,
+      "loss": 0.9084,
+      "step": 4437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241707093239681e-06,
+      "loss": 0.9661,
+      "step": 4438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23770543383803e-06,
+      "loss": 0.9211,
+      "step": 4439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2337057687207816e-06,
+      "loss": 0.5925,
+      "step": 4440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229708099067491e-06,
+      "loss": 0.9878,
+      "step": 4441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225712426057115e-06,
+      "loss": 0.8704,
+      "step": 4442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2217187508680314e-06,
+      "loss": 0.9185,
+      "step": 4443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217727074678023e-06,
+      "loss": 0.9258,
+      "step": 4444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2137373986642807e-06,
+      "loss": 0.918,
+      "step": 4445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2097497240034147e-06,
+      "loss": 0.8424,
+      "step": 4446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205764051871435e-06,
+      "loss": 0.8987,
+      "step": 4447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201780383443772e-06,
+      "loss": 0.8486,
+      "step": 4448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1977987198952543e-06,
+      "loss": 0.957,
+      "step": 4449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193819062400122e-06,
+      "loss": 0.918,
+      "step": 4450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1898414121320277e-06,
+      "loss": 0.9233,
+      "step": 4451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185865770264035e-06,
+      "loss": 0.9277,
+      "step": 4452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1818921379685997e-06,
+      "loss": 0.8984,
+      "step": 4453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177920516417603e-06,
+      "loss": 0.5891,
+      "step": 4454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1739509067823192e-06,
+      "loss": 0.9475,
+      "step": 4455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1699833102334397e-06,
+      "loss": 0.8889,
+      "step": 4456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166017727941062e-06,
+      "loss": 0.8701,
+      "step": 4457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1620541610746747e-06,
+      "loss": 0.9734,
+      "step": 4458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1580926108031884e-06,
+      "loss": 0.9119,
+      "step": 4459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154133078294915e-06,
+      "loss": 0.8522,
+      "step": 4460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1501755647175693e-06,
+      "loss": 0.9629,
+      "step": 4461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1462200712382662e-06,
+      "loss": 0.9321,
+      "step": 4462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142266599023537e-06,
+      "loss": 0.8545,
+      "step": 4463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1383151492393015e-06,
+      "loss": 0.9446,
+      "step": 4464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1343657230508985e-06,
+      "loss": 0.9106,
+      "step": 4465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130418321623061e-06,
+      "loss": 0.9346,
+      "step": 4466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1264729461199208e-06,
+      "loss": 0.8512,
+      "step": 4467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1225295977050265e-06,
+      "loss": 0.8955,
+      "step": 4468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1185882775413123e-06,
+      "loss": 0.9065,
+      "step": 4469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114648986791129e-06,
+      "loss": 0.8992,
+      "step": 4470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11071172661622e-06,
+      "loss": 0.9429,
+      "step": 4471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1067764981777292e-06,
+      "loss": 0.8777,
+      "step": 4472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1028433026362093e-06,
+      "loss": 0.9165,
+      "step": 4473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098912141151602e-06,
+      "loss": 0.9741,
+      "step": 4474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0949830148832627e-06,
+      "loss": 0.9595,
+      "step": 4475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910559249899364e-06,
+      "loss": 0.9265,
+      "step": 4476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0871308726297667e-06,
+      "loss": 0.9885,
+      "step": 4477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083207858960302e-06,
+      "loss": 0.96,
+      "step": 4478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079286885138497e-06,
+      "loss": 0.8931,
+      "step": 4479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0753679523206803e-06,
+      "loss": 0.9016,
+      "step": 4480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0714510616626012e-06,
+      "loss": 0.8555,
+      "step": 4481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067536214319402e-06,
+      "loss": 0.8904,
+      "step": 4482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0636234114456144e-06,
+      "loss": 0.9419,
+      "step": 4483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059712654195176e-06,
+      "loss": 0.9009,
+      "step": 4484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055803943721417e-06,
+      "loss": 0.9346,
+      "step": 4485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051897281177061e-06,
+      "loss": 0.906,
+      "step": 4486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047992667714237e-06,
+      "loss": 0.6088,
+      "step": 4487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044090104484456e-06,
+      "loss": 0.9019,
+      "step": 4488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0401895926386417e-06,
+      "loss": 0.6281,
+      "step": 4489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0362911333270983e-06,
+      "loss": 0.8894,
+      "step": 4490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032394727699527e-06,
+      "loss": 0.887,
+      "step": 4491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0285003769050314e-06,
+      "loss": 0.9077,
+      "step": 4492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024608082092102e-06,
+      "loss": 0.9182,
+      "step": 4493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0207178444086206e-06,
+      "loss": 0.8877,
+      "step": 4494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0168296650018736e-06,
+      "loss": 0.8789,
+      "step": 4495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012943545018526e-06,
+      "loss": 0.8748,
+      "step": 4496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0090594856046473e-06,
+      "loss": 0.9087,
+      "step": 4497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0051774879056995e-06,
+      "loss": 0.9897,
+      "step": 4498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.00129755306652e-06,
+      "loss": 0.8718,
+      "step": 4499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997419682231355e-06,
+      "loss": 0.8741,
+      "step": 4500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993543876543842e-06,
+      "loss": 0.9097,
+      "step": 4501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9896701371469983e-06,
+      "loss": 0.9553,
+      "step": 4502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9857984651832396e-06,
+      "loss": 0.9592,
+      "step": 4503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981928861794364e-06,
+      "loss": 0.9087,
+      "step": 4504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978061328121573e-06,
+      "loss": 0.9478,
+      "step": 4505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9741958653054494e-06,
+      "loss": 0.9138,
+      "step": 4506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9703324744859653e-06,
+      "loss": 0.916,
+      "step": 4507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966471156802477e-06,
+      "loss": 0.8645,
+      "step": 4508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962611913393744e-06,
+      "loss": 0.9414,
+      "step": 4509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958754745397896e-06,
+      "loss": 0.9182,
+      "step": 4510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9548996539524676e-06,
+      "loss": 0.8423,
+      "step": 4511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9510466401943703e-06,
+      "loss": 0.8733,
+      "step": 4512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947195705259901e-06,
+      "loss": 0.9167,
+      "step": 4513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943346850284754e-06,
+      "loss": 0.897,
+      "step": 4514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9395000764040007e-06,
+      "loss": 0.968,
+      "step": 4515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935655384752105e-06,
+      "loss": 0.9021,
+      "step": 4516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9318127764629135e-06,
+      "loss": 0.9404,
+      "step": 4517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9279722526696552e-06,
+      "loss": 0.8977,
+      "step": 4518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9241338145049546e-06,
+      "loss": 0.9543,
+      "step": 4519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9202974631008084e-06,
+      "loss": 0.8931,
+      "step": 4520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9164631995886095e-06,
+      "loss": 0.9321,
+      "step": 4521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9126310250991275e-06,
+      "loss": 0.9429,
+      "step": 4522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908800940762514e-06,
+      "loss": 0.8875,
+      "step": 4523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904972947708312e-06,
+      "loss": 0.627,
+      "step": 4524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9011470470654512e-06,
+      "loss": 0.8833,
+      "step": 4525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8973232399622242e-06,
+      "loss": 0.8503,
+      "step": 4526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8935015275263235e-06,
+      "loss": 0.9458,
+      "step": 4527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889681910884825e-06,
+      "loss": 0.9106,
+      "step": 4528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8858643911641726e-06,
+      "loss": 0.8977,
+      "step": 4529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882048969490208e-06,
+      "loss": 0.8704,
+      "step": 4530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8782356469881424e-06,
+      "loss": 0.8953,
+      "step": 4531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874424424782569e-06,
+      "loss": 0.8765,
+      "step": 4532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8706153039974705e-06,
+      "loss": 0.9263,
+      "step": 4533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8668082857562006e-06,
+      "loss": 0.9041,
+      "step": 4534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863003371181493e-06,
+      "loss": 0.8406,
+      "step": 4535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85920056139547e-06,
+      "loss": 0.8647,
+      "step": 4536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855399857519623e-06,
+      "loss": 0.9607,
+      "step": 4537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85160126067483e-06,
+      "loss": 0.5917,
+      "step": 4538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847804771981343e-06,
+      "loss": 0.9287,
+      "step": 4539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440103925587904e-06,
+      "loss": 0.9165,
+      "step": 4540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.840218123526186e-06,
+      "loss": 0.8562,
+      "step": 4541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8364279660019135e-06,
+      "loss": 0.9126,
+      "step": 4542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8326399211037413e-06,
+      "loss": 0.9189,
+      "step": 4543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8288539899488086e-06,
+      "loss": 0.9282,
+      "step": 4544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825070173653631e-06,
+      "loss": 0.9451,
+      "step": 4545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821288473334104e-06,
+      "loss": 0.8796,
+      "step": 4546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8175088901055026e-06,
+      "loss": 0.9175,
+      "step": 4547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8137314250824686e-06,
+      "loss": 0.9041,
+      "step": 4548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8099560793790204e-06,
+      "loss": 0.9656,
+      "step": 4549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806182854108561e-06,
+      "loss": 0.8909,
+      "step": 4550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8024117503838543e-06,
+      "loss": 0.9807,
+      "step": 4551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798642769317055e-06,
+      "loss": 0.9224,
+      "step": 4552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794875912019669e-06,
+      "loss": 0.9602,
+      "step": 4553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791111179602597e-06,
+      "loss": 0.9128,
+      "step": 4554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7873485731761074e-06,
+      "loss": 0.8556,
+      "step": 4555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783588093849834e-06,
+      "loss": 0.9558,
+      "step": 4556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7798297427327938e-06,
+      "loss": 0.9775,
+      "step": 4557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776073520933369e-06,
+      "loss": 0.9038,
+      "step": 4558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7723194295593113e-06,
+      "loss": 0.866,
+      "step": 4559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7685674697177568e-06,
+      "loss": 0.9663,
+      "step": 4560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7648176425151972e-06,
+      "loss": 0.6187,
+      "step": 4561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76106994905751e-06,
+      "loss": 0.9241,
+      "step": 4562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7573243904499338e-06,
+      "loss": 0.9595,
+      "step": 4563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.753580967797076e-06,
+      "loss": 0.8887,
+      "step": 4564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7498396822029216e-06,
+      "loss": 0.8309,
+      "step": 4565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746100534770829e-06,
+      "loss": 0.9066,
+      "step": 4566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7423635266035053e-06,
+      "loss": 0.894,
+      "step": 4567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.738628658803052e-06,
+      "loss": 0.9192,
+      "step": 4568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734895932470919e-06,
+      "loss": 0.8821,
+      "step": 4569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7311653487079394e-06,
+      "loss": 0.8921,
+      "step": 4570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7274369086143127e-06,
+      "loss": 0.9253,
+      "step": 4571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7237106132895897e-06,
+      "loss": 0.9475,
+      "step": 4572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719986463832708e-06,
+      "loss": 0.5752,
+      "step": 4573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7162644613419687e-06,
+      "loss": 0.9297,
+      "step": 4574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712544606915032e-06,
+      "loss": 0.8557,
+      "step": 4575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708826901648928e-06,
+      "loss": 0.8776,
+      "step": 4576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705111346640058e-06,
+      "loss": 0.947,
+      "step": 4577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7013979429841807e-06,
+      "loss": 0.8628,
+      "step": 4578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69768669177643e-06,
+      "loss": 0.9092,
+      "step": 4579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6939775941112965e-06,
+      "loss": 0.9363,
+      "step": 4580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690270651082638e-06,
+      "loss": 0.8762,
+      "step": 4581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686565863783681e-06,
+      "loss": 0.9165,
+      "step": 4582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68286323330701e-06,
+      "loss": 0.8916,
+      "step": 4583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67916276074458e-06,
+      "loss": 0.9075,
+      "step": 4584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6754644471877054e-06,
+      "loss": 0.9517,
+      "step": 4585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6717682937270605e-06,
+      "loss": 0.8684,
+      "step": 4586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668074301452693e-06,
+      "loss": 0.8623,
+      "step": 4587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6643824714540002e-06,
+      "loss": 0.9358,
+      "step": 4588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6606928048197555e-06,
+      "loss": 0.5982,
+      "step": 4589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657005302638085e-06,
+      "loss": 0.8887,
+      "step": 4590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6533199659964737e-06,
+      "loss": 0.8914,
+      "step": 4591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649636795981777e-06,
+      "loss": 0.9553,
+      "step": 4592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6459557936802104e-06,
+      "loss": 0.957,
+      "step": 4593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6422769601773437e-06,
+      "loss": 0.9678,
+      "step": 4594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638600296558108e-06,
+      "loss": 0.9658,
+      "step": 4595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6349258039068026e-06,
+      "loss": 0.8982,
+      "step": 4596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6312534833070745e-06,
+      "loss": 0.916,
+      "step": 4597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6275833358419447e-06,
+      "loss": 0.6033,
+      "step": 4598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6239153625937786e-06,
+      "loss": 0.8962,
+      "step": 4599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620249564644307e-06,
+      "loss": 0.9414,
+      "step": 4600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616585943074623e-06,
+      "loss": 0.9172,
+      "step": 4601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612924498965169e-06,
+      "loss": 0.9478,
+      "step": 4602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.609265233395757e-06,
+      "loss": 0.8889,
+      "step": 4603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605608147445544e-06,
+      "loss": 0.9136,
+      "step": 4604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6019532421930505e-06,
+      "loss": 0.7888,
+      "step": 4605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598300518716156e-06,
+      "loss": 1.0161,
+      "step": 4606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594649978092093e-06,
+      "loss": 0.9766,
+      "step": 4607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5910016213974475e-06,
+      "loss": 0.9473,
+      "step": 4608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587355449708171e-06,
+      "loss": 0.9277,
+      "step": 4609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58371146409956e-06,
+      "loss": 0.8569,
+      "step": 4610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580069665646271e-06,
+      "loss": 0.9863,
+      "step": 4611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576430055422324e-06,
+      "loss": 0.9155,
+      "step": 4612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5727926345010724e-06,
+      "loss": 0.8911,
+      "step": 4613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5691574039552423e-06,
+      "loss": 0.8762,
+      "step": 4614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5655243648569116e-06,
+      "loss": 0.8809,
+      "step": 4615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561893518277506e-06,
+      "loss": 0.9128,
+      "step": 4616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558264865287806e-06,
+      "loss": 0.9255,
+      "step": 4617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5546384069579446e-06,
+      "loss": 0.9165,
+      "step": 4618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55101414435741e-06,
+      "loss": 0.6177,
+      "step": 4619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547392078555048e-06,
+      "loss": 0.9509,
+      "step": 4620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5437722106190454e-06,
+      "loss": 0.8933,
+      "step": 4621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540154541616945e-06,
+      "loss": 0.9226,
+      "step": 4622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5365390726156456e-06,
+      "loss": 0.9258,
+      "step": 4623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532925804681391e-06,
+      "loss": 0.8958,
+      "step": 4624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5293147388797813e-06,
+      "loss": 0.9268,
+      "step": 4625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5257058762757647e-06,
+      "loss": 0.9111,
+      "step": 4626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5220992179336344e-06,
+      "loss": 0.8779,
+      "step": 4627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5184947649170455e-06,
+      "loss": 0.8667,
+      "step": 4628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.8679,
+      "step": 4629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511292479111819e-06,
+      "loss": 0.887,
+      "step": 4630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5076946484472277e-06,
+      "loss": 0.6046,
+      "step": 4631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5040990273562572e-06,
+      "loss": 0.8683,
+      "step": 4632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500505616899307e-06,
+      "loss": 0.8486,
+      "step": 4633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496914418136113e-06,
+      "loss": 0.8982,
+      "step": 4634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4933254321257694e-06,
+      "loss": 0.5966,
+      "step": 4635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489738659926708e-06,
+      "loss": 0.9734,
+      "step": 4636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486154102596713e-06,
+      "loss": 0.9019,
+      "step": 4637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4825717611929144e-06,
+      "loss": 0.6324,
+      "step": 4638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478991636771797e-06,
+      "loss": 0.9258,
+      "step": 4639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475413730389171e-06,
+      "loss": 0.8933,
+      "step": 4640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471838043100211e-06,
+      "loss": 0.9045,
+      "step": 4641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468264575959436e-06,
+      "loss": 0.8857,
+      "step": 4642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4646933300206977e-06,
+      "loss": 0.9121,
+      "step": 4643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461124306337207e-06,
+      "loss": 0.9636,
+      "step": 4644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.457557505961512e-06,
+      "loss": 1.0066,
+      "step": 4645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4539929299455024e-06,
+      "loss": 0.8608,
+      "step": 4646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4504305793404204e-06,
+      "loss": 0.9224,
+      "step": 4647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446870455196847e-06,
+      "loss": 0.968,
+      "step": 4648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443312558564701e-06,
+      "loss": 0.9082,
+      "step": 4649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439756890493258e-06,
+      "loss": 0.9048,
+      "step": 4650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4362034520311216e-06,
+      "loss": 0.8574,
+      "step": 4651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4326522442262522e-06,
+      "loss": 0.9343,
+      "step": 4652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429103268125941e-06,
+      "loss": 0.9099,
+      "step": 4653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425556524776821e-06,
+      "loss": 0.9541,
+      "step": 4654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4220120152248783e-06,
+      "loss": 0.9329,
+      "step": 4655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418469740515427e-06,
+      "loss": 0.9351,
+      "step": 4656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4149297016931317e-06,
+      "loss": 0.906,
+      "step": 4657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4113918998019927e-06,
+      "loss": 0.9277,
+      "step": 4658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4078563358853467e-06,
+      "loss": 0.5838,
+      "step": 4659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4043230109858805e-06,
+      "loss": 0.9199,
+      "step": 4660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400791926145617e-06,
+      "loss": 0.8997,
+      "step": 4661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397263082405914e-06,
+      "loss": 0.9067,
+      "step": 4662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3937364808074683e-06,
+      "loss": 0.623,
+      "step": 4663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390212122390323e-06,
+      "loss": 0.9541,
+      "step": 4664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3866900081938517e-06,
+      "loss": 0.9365,
+      "step": 4665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383170139256774e-06,
+      "loss": 0.9082,
+      "step": 4666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37965251661714e-06,
+      "loss": 0.9153,
+      "step": 4667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376137141312338e-06,
+      "loss": 0.938,
+      "step": 4668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372624014379099e-06,
+      "loss": 0.8582,
+      "step": 4669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3691131368534858e-06,
+      "loss": 0.9636,
+      "step": 4670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365604509770901e-06,
+      "loss": 0.9023,
+      "step": 4671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3620981341660833e-06,
+      "loss": 0.5916,
+      "step": 4672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3585940110731008e-06,
+      "loss": 0.8525,
+      "step": 4673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3550921415253704e-06,
+      "loss": 0.5894,
+      "step": 4674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3515925265556293e-06,
+      "loss": 0.9119,
+      "step": 4675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348095167195964e-06,
+      "loss": 0.9041,
+      "step": 4676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3446000644777856e-06,
+      "loss": 0.6134,
+      "step": 4677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341107219431842e-06,
+      "loss": 0.5717,
+      "step": 4678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337616633088219e-06,
+      "loss": 0.9304,
+      "step": 4679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3341283064763343e-06,
+      "loss": 0.8738,
+      "step": 4680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330642240624933e-06,
+      "loss": 0.9219,
+      "step": 4681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.327158436562107e-06,
+      "loss": 0.8865,
+      "step": 4682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3236768953152665e-06,
+      "loss": 0.8884,
+      "step": 4683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320197617911163e-06,
+      "loss": 0.916,
+      "step": 4684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316720605375886e-06,
+      "loss": 0.9004,
+      "step": 4685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313245858734838e-06,
+      "loss": 0.9336,
+      "step": 4686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30977337901277e-06,
+      "loss": 0.9395,
+      "step": 4687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306303167233761e-06,
+      "loss": 0.9641,
+      "step": 4688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3028352244212173e-06,
+      "loss": 0.9385,
+      "step": 4689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2993695515978767e-06,
+      "loss": 0.8136,
+      "step": 4690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2959061497858125e-06,
+      "loss": 0.9062,
+      "step": 4691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292445020006422e-06,
+      "loss": 0.9182,
+      "step": 4692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2889861632804377e-06,
+      "loss": 0.9109,
+      "step": 4693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2855295806279188e-06,
+      "loss": 0.9072,
+      "step": 4694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2820752730682494e-06,
+      "loss": 0.915,
+      "step": 4695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2786232416201548e-06,
+      "loss": 0.9136,
+      "step": 4696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275173487301676e-06,
+      "loss": 0.8921,
+      "step": 4697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2717260111301943e-06,
+      "loss": 0.5869,
+      "step": 4698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268280814122409e-06,
+      "loss": 0.8862,
+      "step": 4699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2648378972943495e-06,
+      "loss": 0.9253,
+      "step": 4700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26139726166138e-06,
+      "loss": 0.8596,
+      "step": 4701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2579589082381813e-06,
+      "loss": 0.8965,
+      "step": 4702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2545228380387706e-06,
+      "loss": 0.9819,
+      "step": 4703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251089052076487e-06,
+      "loss": 0.6542,
+      "step": 4704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247657551363992e-06,
+      "loss": 0.9756,
+      "step": 4705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.24422833691328e-06,
+      "loss": 0.9138,
+      "step": 4706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2408014097356732e-06,
+      "loss": 0.8606,
+      "step": 4707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2373767708418116e-06,
+      "loss": 0.9028,
+      "step": 4708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233954421241661e-06,
+      "loss": 0.9971,
+      "step": 4709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2305343619445185e-06,
+      "loss": 0.9268,
+      "step": 4710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227116593958999e-06,
+      "loss": 0.8773,
+      "step": 4711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223701118293048e-06,
+      "loss": 0.3084,
+      "step": 4712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22028793595393e-06,
+      "loss": 0.936,
+      "step": 4713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2168770479482315e-06,
+      "loss": 0.9041,
+      "step": 4714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213468455281872e-06,
+      "loss": 0.875,
+      "step": 4715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2100621589600813e-06,
+      "loss": 0.8567,
+      "step": 4716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206658159987424e-06,
+      "loss": 0.8707,
+      "step": 4717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2032564593677773e-06,
+      "loss": 0.9004,
+      "step": 4718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1998570581043453e-06,
+      "loss": 0.939,
+      "step": 4719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196459957199657e-06,
+      "loss": 0.5913,
+      "step": 4720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193065157655556e-06,
+      "loss": 0.8921,
+      "step": 4721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189672660473211e-06,
+      "loss": 0.8796,
+      "step": 4722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1862824666531135e-06,
+      "loss": 0.8914,
+      "step": 4723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1828945771950704e-06,
+      "loss": 0.8789,
+      "step": 4724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179508993098214e-06,
+      "loss": 0.9021,
+      "step": 4725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1761257153610004e-06,
+      "loss": 0.8694,
+      "step": 4726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17274474498119e-06,
+      "loss": 0.9263,
+      "step": 4727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1693660829558804e-06,
+      "loss": 0.8921,
+      "step": 4728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165989730281475e-06,
+      "loss": 0.8926,
+      "step": 4729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1626156879537084e-06,
+      "loss": 0.9558,
+      "step": 4730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159243956967624e-06,
+      "loss": 0.8522,
+      "step": 4731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155874538317585e-06,
+      "loss": 0.9375,
+      "step": 4732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152507432997277e-06,
+      "loss": 0.5917,
+      "step": 4733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149142641999703e-06,
+      "loss": 0.9332,
+      "step": 4734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1457801663171805e-06,
+      "loss": 0.899,
+      "step": 4735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1424200069413416e-06,
+      "loss": 0.6085,
+      "step": 4736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139062164863144e-06,
+      "loss": 0.658,
+      "step": 4737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135706641072851e-06,
+      "loss": 0.9382,
+      "step": 4738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132353436560055e-06,
+      "loss": 0.9224,
+      "step": 4739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.129002552313655e-06,
+      "loss": 0.9458,
+      "step": 4740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125653989321863e-06,
+      "loss": 0.8955,
+      "step": 4741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12230774857222e-06,
+      "loss": 0.9131,
+      "step": 4742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118963831051567e-06,
+      "loss": 0.8652,
+      "step": 4743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1156222377460723e-06,
+      "loss": 0.894,
+      "step": 4744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1122829696412108e-06,
+      "loss": 0.8499,
+      "step": 4745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.10894602772177e-06,
+      "loss": 0.9106,
+      "step": 4746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105611412971863e-06,
+      "loss": 0.9138,
+      "step": 4747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102279126374902e-06,
+      "loss": 0.897,
+      "step": 4748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098949168913624e-06,
+      "loss": 0.9243,
+      "step": 4749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.095621541570075e-06,
+      "loss": 0.8361,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092296245325609e-06,
+      "loss": 0.9099,
+      "step": 4751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0889732811608996e-06,
+      "loss": 0.8826,
+      "step": 4752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085652650055935e-06,
+      "loss": 0.8938,
+      "step": 4753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0823343529900007e-06,
+      "loss": 0.9463,
+      "step": 4754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0790183909417096e-06,
+      "loss": 0.8799,
+      "step": 4755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0757047648889816e-06,
+      "loss": 0.8054,
+      "step": 4756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.07239347580904e-06,
+      "loss": 0.8628,
+      "step": 4757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069084524678432e-06,
+      "loss": 0.9133,
+      "step": 4758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0657779124730036e-06,
+      "loss": 0.8621,
+      "step": 4759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.062473640167915e-06,
+      "loss": 0.8574,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0591717087376416e-06,
+      "loss": 0.8706,
+      "step": 4761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558721191559606e-06,
+      "loss": 0.8906,
+      "step": 4762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0525748723959595e-06,
+      "loss": 0.937,
+      "step": 4763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049279969430044e-06,
+      "loss": 0.9365,
+      "step": 4764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045987411229913e-06,
+      "loss": 0.9387,
+      "step": 4765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0426971987665912e-06,
+      "loss": 0.9905,
+      "step": 4766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0394093330103992e-06,
+      "loss": 0.8718,
+      "step": 4767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036123814930967e-06,
+      "loss": 0.9128,
+      "step": 4768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0328406454972395e-06,
+      "loss": 0.8547,
+      "step": 4769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0295598256774598e-06,
+      "loss": 0.884,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0262813564391846e-06,
+      "loss": 0.873,
+      "step": 4771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023005238749276e-06,
+      "loss": 0.8553,
+      "step": 4772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019731473573897e-06,
+      "loss": 0.9104,
+      "step": 4773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0164600618785247e-06,
+      "loss": 0.8835,
+      "step": 4774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013191004627941e-06,
+      "loss": 0.9295,
+      "step": 4775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0099243027862303e-06,
+      "loss": 0.907,
+      "step": 4776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0066599573167824e-06,
+      "loss": 0.8943,
+      "step": 4777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003397969182289e-06,
+      "loss": 0.9167,
+      "step": 4778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000138339344756e-06,
+      "loss": 0.6053,
+      "step": 4779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9968810687654893e-06,
+      "loss": 0.9094,
+      "step": 4780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9936261584050974e-06,
+      "loss": 0.9346,
+      "step": 4781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9903736092234904e-06,
+      "loss": 0.8829,
+      "step": 4782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9871234221798895e-06,
+      "loss": 0.9185,
+      "step": 4783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9838755982328108e-06,
+      "loss": 0.3271,
+      "step": 4784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9806301383400837e-06,
+      "loss": 0.9021,
+      "step": 4785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9773870434588306e-06,
+      "loss": 0.9373,
+      "step": 4786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741463145454786e-06,
+      "loss": 0.9333,
+      "step": 4787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970907952555764e-06,
+      "loss": 0.9041,
+      "step": 4788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9676719584447134e-06,
+      "loss": 0.8164,
+      "step": 4789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9644383331666684e-06,
+      "loss": 0.949,
+      "step": 4790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961207077675261e-06,
+      "loss": 0.603,
+      "step": 4791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9579781929234275e-06,
+      "loss": 0.9302,
+      "step": 4792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9547516798634104e-06,
+      "loss": 0.8711,
+      "step": 4793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9515275394467446e-06,
+      "loss": 0.885,
+      "step": 4794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9483057726242694e-06,
+      "loss": 0.8616,
+      "step": 4795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945086380346126e-06,
+      "loss": 0.929,
+      "step": 4796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941869363561749e-06,
+      "loss": 0.9941,
+      "step": 4797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9386547232198795e-06,
+      "loss": 0.9026,
+      "step": 4798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9354424602685586e-06,
+      "loss": 0.9026,
+      "step": 4799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932232575655113e-06,
+      "loss": 0.8774,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9290250703261824e-06,
+      "loss": 0.95,
+      "step": 4801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9258199452277037e-06,
+      "loss": 0.9126,
+      "step": 4802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9226172013049027e-06,
+      "loss": 0.9126,
+      "step": 4803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9194168395023083e-06,
+      "loss": 0.8806,
+      "step": 4804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91621886076375e-06,
+      "loss": 0.9749,
+      "step": 4805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913023266032348e-06,
+      "loss": 0.9336,
+      "step": 4806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.8997,
+      "step": 4807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9066392323600003e-06,
+      "loss": 0.9324,
+      "step": 4808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903450795301781e-06,
+      "loss": 0.9331,
+      "step": 4809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9002647460161826e-06,
+      "loss": 0.5807,
+      "step": 4810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8970810854428046e-06,
+      "loss": 0.8882,
+      "step": 4811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8938998145205557e-06,
+      "loss": 0.8053,
+      "step": 4812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8907209341876266e-06,
+      "loss": 0.9219,
+      "step": 4813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8875444453815073e-06,
+      "loss": 0.8726,
+      "step": 4814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8843703490389885e-06,
+      "loss": 0.9287,
+      "step": 4815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8811986460961461e-06,
+      "loss": 0.8804,
+      "step": 4816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8780293374883584e-06,
+      "loss": 0.9031,
+      "step": 4817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8748624241502934e-06,
+      "loss": 0.9363,
+      "step": 4818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.871697907015907e-06,
+      "loss": 0.9131,
+      "step": 4819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8685357870184605e-06,
+      "loss": 0.8923,
+      "step": 4820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8653760650905028e-06,
+      "loss": 0.9053,
+      "step": 4821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8622187421638737e-06,
+      "loss": 0.9021,
+      "step": 4822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8590638191697031e-06,
+      "loss": 0.9128,
+      "step": 4823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.855911297038422e-06,
+      "loss": 0.9736,
+      "step": 4824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8527611766997444e-06,
+      "loss": 0.9143,
+      "step": 4825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8496134590826864e-06,
+      "loss": 0.8621,
+      "step": 4826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464681451155387e-06,
+      "loss": 0.8936,
+      "step": 4827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8433252357258991e-06,
+      "loss": 0.8591,
+      "step": 4828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8401847318406518e-06,
+      "loss": 0.9236,
+      "step": 4829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8370466343859672e-06,
+      "loss": 0.9065,
+      "step": 4830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8339109442873115e-06,
+      "loss": 0.8318,
+      "step": 4831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8307776624694383e-06,
+      "loss": 0.927,
+      "step": 4832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8276467898563887e-06,
+      "loss": 0.8726,
+      "step": 4833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8245183273714984e-06,
+      "loss": 0.8298,
+      "step": 4834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8213922759373903e-06,
+      "loss": 0.6337,
+      "step": 4835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8182686364759704e-06,
+      "loss": 0.9771,
+      "step": 4836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8151474099084466e-06,
+      "loss": 0.863,
+      "step": 4837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8120285971552987e-06,
+      "loss": 0.9153,
+      "step": 4838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808912199136309e-06,
+      "loss": 0.8772,
+      "step": 4839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8057982167705446e-06,
+      "loss": 0.8733,
+      "step": 4840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8026866509763497e-06,
+      "loss": 0.9333,
+      "step": 4841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799577502671368e-06,
+      "loss": 0.9329,
+      "step": 4842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7964707727725238e-06,
+      "loss": 0.958,
+      "step": 4843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7933664621960334e-06,
+      "loss": 0.936,
+      "step": 4844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7902645718573953e-06,
+      "loss": 0.8994,
+      "step": 4845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787165102671391e-06,
+      "loss": 0.95,
+      "step": 4846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7840680555520961e-06,
+      "loss": 0.8333,
+      "step": 4847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809734314128712e-06,
+      "loss": 0.8804,
+      "step": 4848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.777881231166355e-06,
+      "loss": 0.8879,
+      "step": 4849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774791455724475e-06,
+      "loss": 0.6155,
+      "step": 4850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7717041059984486e-06,
+      "loss": 0.9595,
+      "step": 4851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7686191828987676e-06,
+      "loss": 0.8904,
+      "step": 4852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7655366873352197e-06,
+      "loss": 1.0051,
+      "step": 4853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762456620216869e-06,
+      "loss": 0.9199,
+      "step": 4854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7593789824520624e-06,
+      "loss": 0.9087,
+      "step": 4855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.756303774948439e-06,
+      "loss": 0.989,
+      "step": 4856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7532309986129092e-06,
+      "loss": 0.9006,
+      "step": 4857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7501606543516792e-06,
+      "loss": 0.9187,
+      "step": 4858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7470927430702277e-06,
+      "loss": 0.9072,
+      "step": 4859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7440272656733182e-06,
+      "loss": 0.639,
+      "step": 4860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7409642230650037e-06,
+      "loss": 0.9397,
+      "step": 4861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7379036161486063e-06,
+      "loss": 0.5863,
+      "step": 4862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7348454458267428e-06,
+      "loss": 0.9353,
+      "step": 4863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731789713001304e-06,
+      "loss": 0.8835,
+      "step": 4864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7287364185734602e-06,
+      "loss": 0.9438,
+      "step": 4865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7256855634436675e-06,
+      "loss": 0.874,
+      "step": 4866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7226371485116667e-06,
+      "loss": 0.9597,
+      "step": 4867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7195911746764627e-06,
+      "loss": 0.7935,
+      "step": 4868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716547642836356e-06,
+      "loss": 0.9607,
+      "step": 4869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.713506553888924e-06,
+      "loss": 0.9187,
+      "step": 4870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7104679087310172e-06,
+      "loss": 0.8933,
+      "step": 4871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074317082587755e-06,
+      "loss": 0.9321,
+      "step": 4872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704397953367607e-06,
+      "loss": 0.6143,
+      "step": 4873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013666449522025e-06,
+      "loss": 0.8525,
+      "step": 4874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6983377839065373e-06,
+      "loss": 0.8586,
+      "step": 4875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6953113711238578e-06,
+      "loss": 0.9749,
+      "step": 4876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.692287407496689e-06,
+      "loss": 0.929,
+      "step": 4877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689265893916837e-06,
+      "loss": 0.5845,
+      "step": 4878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686246831275382e-06,
+      "loss": 0.926,
+      "step": 4879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683230220462686e-06,
+      "loss": 0.9724,
+      "step": 4880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6802160623683838e-06,
+      "loss": 0.8901,
+      "step": 4881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6772043578813846e-06,
+      "loss": 0.9482,
+      "step": 4882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674195107889881e-06,
+      "loss": 0.8918,
+      "step": 4883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6711883132813355e-06,
+      "loss": 0.9006,
+      "step": 4884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668183974942491e-06,
+      "loss": 0.8564,
+      "step": 4885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6651820937593677e-06,
+      "loss": 0.9014,
+      "step": 4886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6621826706172494e-06,
+      "loss": 0.9224,
+      "step": 4887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6591857064007067e-06,
+      "loss": 0.9124,
+      "step": 4888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6561912019935843e-06,
+      "loss": 0.6187,
+      "step": 4889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6531991582789964e-06,
+      "loss": 0.8152,
+      "step": 4890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6502095761393332e-06,
+      "loss": 0.9258,
+      "step": 4891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6472224564562577e-06,
+      "loss": 0.8521,
+      "step": 4892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6442378001107106e-06,
+      "loss": 0.5914,
+      "step": 4893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6412556079829066e-06,
+      "loss": 0.9563,
+      "step": 4894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.638275880952328e-06,
+      "loss": 0.9446,
+      "step": 4895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6352986198977327e-06,
+      "loss": 0.8555,
+      "step": 4896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632323825697154e-06,
+      "loss": 0.9045,
+      "step": 4897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293514992278935e-06,
+      "loss": 0.8816,
+      "step": 4898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626381641366529e-06,
+      "loss": 0.927,
+      "step": 4899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6234142529889084e-06,
+      "loss": 0.9397,
+      "step": 4900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6204493349701477e-06,
+      "loss": 0.9189,
+      "step": 4901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6174868881846429e-06,
+      "loss": 0.9043,
+      "step": 4902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6145269135060514e-06,
+      "loss": 0.97,
+      "step": 4903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6115694118073112e-06,
+      "loss": 0.9548,
+      "step": 4904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6086143839606238e-06,
+      "loss": 0.8549,
+      "step": 4905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6056618308374606e-06,
+      "loss": 0.9016,
+      "step": 4906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6027117533085723e-06,
+      "loss": 0.8911,
+      "step": 4907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59976415224397e-06,
+      "loss": 0.8638,
+      "step": 4908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5968190285129349e-06,
+      "loss": 0.9189,
+      "step": 4909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5938763829840265e-06,
+      "loss": 0.9067,
+      "step": 4910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5909362165250609e-06,
+      "loss": 0.8599,
+      "step": 4911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587998530003133e-06,
+      "loss": 0.9302,
+      "step": 4912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5850633242846082e-06,
+      "loss": 0.9258,
+      "step": 4913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5821306002351045e-06,
+      "loss": 0.8804,
+      "step": 4914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5792003587195237e-06,
+      "loss": 0.8844,
+      "step": 4915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5762726006020325e-06,
+      "loss": 0.9019,
+      "step": 4916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5733473267460609e-06,
+      "loss": 0.6388,
+      "step": 4917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570424538014307e-06,
+      "loss": 0.8406,
+      "step": 4918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5675042352687397e-06,
+      "loss": 0.9119,
+      "step": 4919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5645864193705896e-06,
+      "loss": 0.9221,
+      "step": 4920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5616710911803611e-06,
+      "loss": 0.8645,
+      "step": 4921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558758251557817e-06,
+      "loss": 0.9072,
+      "step": 4922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.55584790136199e-06,
+      "loss": 0.9053,
+      "step": 4923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5529400414511809e-06,
+      "loss": 0.8867,
+      "step": 4924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5500346726829495e-06,
+      "loss": 0.8899,
+      "step": 4925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5471317959141296e-06,
+      "loss": 0.9397,
+      "step": 4926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.544231412000814e-06,
+      "loss": 0.8882,
+      "step": 4927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5413335217983594e-06,
+      "loss": 0.9753,
+      "step": 4928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5384381261613924e-06,
+      "loss": 0.9307,
+      "step": 4929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5355452259437986e-06,
+      "loss": 0.9131,
+      "step": 4930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5326548219987326e-06,
+      "loss": 0.9922,
+      "step": 4931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.52976691517861e-06,
+      "loss": 0.6268,
+      "step": 4932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5268815063351072e-06,
+      "loss": 0.8712,
+      "step": 4933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5239985963191683e-06,
+      "loss": 0.9453,
+      "step": 4934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5211181859810032e-06,
+      "loss": 0.8892,
+      "step": 4935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5182402761700776e-06,
+      "loss": 0.9441,
+      "step": 4936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5153648677351196e-06,
+      "loss": 0.8677,
+      "step": 4937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5124919615241284e-06,
+      "loss": 0.8157,
+      "step": 4938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5096215583843554e-06,
+      "loss": 0.8406,
+      "step": 4939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5067536591623233e-06,
+      "loss": 0.938,
+      "step": 4940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5038882647038034e-06,
+      "loss": 0.9255,
+      "step": 4941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5010253758538396e-06,
+      "loss": 0.9419,
+      "step": 4942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4981649934567365e-06,
+      "loss": 0.873,
+      "step": 4943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4953071183560508e-06,
+      "loss": 0.9395,
+      "step": 4944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49245175139461e-06,
+      "loss": 0.6083,
+      "step": 4945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895988934144956e-06,
+      "loss": 0.8752,
+      "step": 4946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4867485452570473e-06,
+      "loss": 0.9446,
+      "step": 4947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.483900707762873e-06,
+      "loss": 0.8837,
+      "step": 4948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4810553817718343e-06,
+      "loss": 0.8953,
+      "step": 4949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4782125681230497e-06,
+      "loss": 0.8909,
+      "step": 4950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4753722676549042e-06,
+      "loss": 0.8716,
+      "step": 4951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4725344812050336e-06,
+      "loss": 0.877,
+      "step": 4952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469699209610338e-06,
+      "loss": 0.9338,
+      "step": 4953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.46686645370698e-06,
+      "loss": 0.9268,
+      "step": 4954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4640362143303644e-06,
+      "loss": 0.9097,
+      "step": 4955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4612084923151703e-06,
+      "loss": 0.5782,
+      "step": 4956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4583832884953241e-06,
+      "loss": 0.9346,
+      "step": 4957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4555606037040159e-06,
+      "loss": 0.8887,
+      "step": 4958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4527404387736943e-06,
+      "loss": 0.9111,
+      "step": 4959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4499227945360505e-06,
+      "loss": 0.9093,
+      "step": 4960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.44710767182205e-06,
+      "loss": 0.9229,
+      "step": 4961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4442950714619075e-06,
+      "loss": 0.8928,
+      "step": 4962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4414849942850927e-06,
+      "loss": 0.9058,
+      "step": 4963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4386774411203285e-06,
+      "loss": 0.88,
+      "step": 4964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4358724127956036e-06,
+      "loss": 0.9265,
+      "step": 4965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4330699101381496e-06,
+      "loss": 0.913,
+      "step": 4966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430269933974463e-06,
+      "loss": 0.9221,
+      "step": 4967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4274724851302912e-06,
+      "loss": 0.9265,
+      "step": 4968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4246775644306333e-06,
+      "loss": 0.9263,
+      "step": 4969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4218851726997507e-06,
+      "loss": 0.9175,
+      "step": 4970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4190953107611483e-06,
+      "loss": 0.5989,
+      "step": 4971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4163079794375977e-06,
+      "loss": 0.9358,
+      "step": 4972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.413523179551115e-06,
+      "loss": 0.8616,
+      "step": 4973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4107409119229675e-06,
+      "loss": 0.8845,
+      "step": 4974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4079611773736879e-06,
+      "loss": 0.8936,
+      "step": 4975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051839767230479e-06,
+      "loss": 0.8645,
+      "step": 4976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4024093107900838e-06,
+      "loss": 0.6263,
+      "step": 4977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3996371803930753e-06,
+      "loss": 0.6163,
+      "step": 4978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3968675863495572e-06,
+      "loss": 0.6475,
+      "step": 4979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.39410052947632e-06,
+      "loss": 0.9329,
+      "step": 4980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3913360105894048e-06,
+      "loss": 0.9087,
+      "step": 4981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3885740305040962e-06,
+      "loss": 0.5745,
+      "step": 4982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3858145900349385e-06,
+      "loss": 0.8544,
+      "step": 4983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3830576899957292e-06,
+      "loss": 0.9287,
+      "step": 4984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3803033311995072e-06,
+      "loss": 0.9177,
+      "step": 4985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775515144585705e-06,
+      "loss": 0.8423,
+      "step": 4986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3748022405844619e-06,
+      "loss": 0.8816,
+      "step": 4987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3720555103879751e-06,
+      "loss": 0.9153,
+      "step": 4988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.369311324679159e-06,
+      "loss": 0.8772,
+      "step": 4989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366569684267306e-06,
+      "loss": 0.864,
+      "step": 4990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363830589960956e-06,
+      "loss": 0.632,
+      "step": 4991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610940425679076e-06,
+      "loss": 0.8838,
+      "step": 4992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358360042895198e-06,
+      "loss": 0.8682,
+      "step": 4993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355628591749122e-06,
+      "loss": 0.949,
+      "step": 4994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3528996899352155e-06,
+      "loss": 0.9802,
+      "step": 4995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3501733382582627e-06,
+      "loss": 0.8887,
+      "step": 4996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347449537522305e-06,
+      "loss": 0.875,
+      "step": 4997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3447282885306201e-06,
+      "loss": 0.9136,
+      "step": 4998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3420095920857378e-06,
+      "loss": 0.8538,
+      "step": 4999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3392934489894427e-06,
+      "loss": 0.9185,
+      "step": 5000
+    }
+  ],
+  "max_steps": 5964,
+  "num_train_epochs": 1,
+  "total_flos": 2.0213264621038993e+19,
+  "trial_name": null,
+  "trial_params": null
+}