diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14623 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999229939935315,
+  "global_step": 2433,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 1.658,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 1.8738,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.1654,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.1991,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.0743,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.1493,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.1059,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.0695,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.0905,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.2188,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.1695,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.0543,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.0257,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 1.001,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 0.9035,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 0.9979,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.0299,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.0091,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 0.9827,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 1.0258,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.0672,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 1.0086,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 1.0374,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.978,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 1.0097,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.0713,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 1.0203,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 1.0834,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.0166,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 1.0958,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 0.962,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 0.9332,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 0.9713,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 1.0534,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 0.9566,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 1.0108,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.8976,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 1.0651,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 1.0302,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 1.0534,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 0.9088,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 1.0085,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 1.0221,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 1.0166,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 0.9398,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 1.0277,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 1.0425,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.8892,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 0.9969,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.9618,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 1.0422,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.9439,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 1.0713,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 1.1741,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 1.1307,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 1.0006,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.912,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.8973,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 1.0237,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.9566,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 0.9816,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.9593,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 1.0471,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 1.0543,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 0.9724,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 1.005,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 0.9581,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.9424,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 1.0022,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 1.0554,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 1.1595,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.9725,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.9838,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999113975605e-05,
+      "loss": 1.0296,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996455903989e-05,
+      "loss": 1.0783,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992025789863e-05,
+      "loss": 1.0363,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999858236410775e-05,
+      "loss": 0.9063,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999778494686226e-05,
+      "loss": 0.8785,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999681032866287e-05,
+      "loss": 0.9943,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999956585112367e-05,
+      "loss": 0.9244,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999432949662483e-05,
+      "loss": 0.8962,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999928232871823e-05,
+      "loss": 1.0743,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999911398855782e-05,
+      "loss": 0.974,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999892792947956e-05,
+      "loss": 1.0517,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998724151813157e-05,
+      "loss": 0.9491,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998502655919713e-05,
+      "loss": 1.0521,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999826344219173e-05,
+      "loss": 1.0169,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999800651105311e-05,
+      "loss": 1.0079,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997731862959143e-05,
+      "loss": 1.0357,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999743949839652e-05,
+      "loss": 0.9522,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997129417883326e-05,
+      "loss": 0.9206,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996801621969037e-05,
+      "loss": 0.9974,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999645611123453e-05,
+      "loss": 1.0022,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996092886292055e-05,
+      "loss": 1.0249,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995711947785276e-05,
+      "loss": 1.072,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995313296389226e-05,
+      "loss": 1.0365,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999489693281034e-05,
+      "loss": 1.015,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994462857786433e-05,
+      "loss": 0.8519,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994011072086707e-05,
+      "loss": 1.0599,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999354157651175e-05,
+      "loss": 1.0633,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993054371893526e-05,
+      "loss": 1.0066,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999254945909539e-05,
+      "loss": 1.0184,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999202683901207e-05,
+      "loss": 1.0657,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991486512569677e-05,
+      "loss": 0.9894,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990928480725694e-05,
+      "loss": 1.0008,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999035274446898e-05,
+      "loss": 1.0979,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989759304819765e-05,
+      "loss": 0.9782,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989148162829663e-05,
+      "loss": 1.0324,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988519319581637e-05,
+      "loss": 1.1834,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998787277619003e-05,
+      "loss": 1.1209,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987208533800552e-05,
+      "loss": 1.1269,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986526593590275e-05,
+      "loss": 1.0442,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998582695676762e-05,
+      "loss": 1.0206,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998510962457239e-05,
+      "loss": 0.926,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984374598275722e-05,
+      "loss": 0.9785,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983621879180125e-05,
+      "loss": 0.91,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998285146861945e-05,
+      "loss": 1.0386,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982063367958907e-05,
+      "loss": 0.9223,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981257578595047e-05,
+      "loss": 1.0241,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980434101955762e-05,
+      "loss": 0.9925,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99795929395003e-05,
+      "loss": 1.0568,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978734092719244e-05,
+      "loss": 1.028,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977857563134503e-05,
+      "loss": 1.1134,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976963352299337e-05,
+      "loss": 1.0476,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997605146179833e-05,
+      "loss": 1.0234,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975121893247396e-05,
+      "loss": 0.9512,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974174648293774e-05,
+      "loss": 1.1104,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973209728616032e-05,
+      "loss": 1.197,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997222713592405e-05,
+      "loss": 1.1371,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971226871959037e-05,
+      "loss": 1.1173,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970208938493504e-05,
+      "loss": 1.0441,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969173337331283e-05,
+      "loss": 1.0247,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968120070307503e-05,
+      "loss": 1.0264,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996704913928861e-05,
+      "loss": 1.0072,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965960546172346e-05,
+      "loss": 1.1115,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964854292887747e-05,
+      "loss": 0.8799,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963730381395154e-05,
+      "loss": 1.0802,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962588813686187e-05,
+      "loss": 0.9735,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961429591783764e-05,
+      "loss": 1.0063,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996025271774208e-05,
+      "loss": 0.9453,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959058193646618e-05,
+      "loss": 1.1692,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9957846021614126e-05,
+      "loss": 0.971,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956616203792636e-05,
+      "loss": 1.0007,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955368742361445e-05,
+      "loss": 1.0786,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954103639531116e-05,
+      "loss": 0.9693,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952820897543468e-05,
+      "loss": 1.0557,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951520518671587e-05,
+      "loss": 1.014,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950202505219808e-05,
+      "loss": 0.9715,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948866859523717e-05,
+      "loss": 1.0267,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994751358395014e-05,
+      "loss": 1.0039,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946142680897145e-05,
+      "loss": 0.9758,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944754152794044e-05,
+      "loss": 1.1135,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943348002101374e-05,
+      "loss": 1.0018,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941924231310903e-05,
+      "loss": 1.0246,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994048284294562e-05,
+      "loss": 1.0571,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939023839559745e-05,
+      "loss": 1.0445,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993754722373869e-05,
+      "loss": 0.967,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99360529980991e-05,
+      "loss": 0.999,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993454116528881e-05,
+      "loss": 0.9723,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933011727986865e-05,
+      "loss": 0.9921,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931464688903502e-05,
+      "loss": 1.0482,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929900050780147e-05,
+      "loss": 0.9949,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928317816389416e-05,
+      "loss": 0.9308,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992671798853511e-05,
+      "loss": 1.0839,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925100570052194e-05,
+      "loss": 1.0279,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923465563806825e-05,
+      "loss": 1.1507,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921812972696298e-05,
+      "loss": 1.0505,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920142799649098e-05,
+      "loss": 0.9731,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918455047624847e-05,
+      "loss": 1.0994,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916749719614326e-05,
+      "loss": 1.0291,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915026818639457e-05,
+      "loss": 1.094,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913286347753306e-05,
+      "loss": 1.0935,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911528310040073e-05,
+      "loss": 0.9176,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909752708615088e-05,
+      "loss": 1.0972,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99079595466248e-05,
+      "loss": 1.0049,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990614882724678e-05,
+      "loss": 0.9358,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990432055368971e-05,
+      "loss": 1.0514,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902474729193385e-05,
+      "loss": 1.0593,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990061135702869e-05,
+      "loss": 0.8965,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989873044049762e-05,
+      "loss": 0.9694,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989683198293324e-05,
+      "loss": 1.1043,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894915987699718e-05,
+      "loss": 1.0442,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892982458192286e-05,
+      "loss": 0.9898,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891031397837258e-05,
+      "loss": 0.9948,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889062810092002e-05,
+      "loss": 1.0379,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887076698444953e-05,
+      "loss": 1.071,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885073066415596e-05,
+      "loss": 0.9517,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883051917554473e-05,
+      "loss": 1.0254,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881013255443152e-05,
+      "loss": 1.0169,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987895708369424e-05,
+      "loss": 1.0932,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 1.1318,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874792225889223e-05,
+      "loss": 0.9104,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872683547213446e-05,
+      "loss": 1.0188,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870557373660733e-05,
+      "loss": 0.9453,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986841370899876e-05,
+      "loss": 0.9985,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866252557026215e-05,
+      "loss": 0.9495,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864073921572756e-05,
+      "loss": 0.9413,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861877806499033e-05,
+      "loss": 0.9551,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859664215696676e-05,
+      "loss": 0.9801,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857433153088267e-05,
+      "loss": 1.163,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855184622627362e-05,
+      "loss": 1.0419,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852918628298466e-05,
+      "loss": 1.1131,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850635174117033e-05,
+      "loss": 1.0732,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984833426412945e-05,
+      "loss": 0.998,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846015902413053e-05,
+      "loss": 1.1185,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984368009307608e-05,
+      "loss": 1.0758,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98413268402577e-05,
+      "loss": 0.8996,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838956148128004e-05,
+      "loss": 1.0708,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836568020887963e-05,
+      "loss": 1.1631,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834162462769454e-05,
+      "loss": 1.104,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983173947803525e-05,
+      "loss": 1.1173,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829299070978997e-05,
+      "loss": 0.9888,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982684124592521e-05,
+      "loss": 1.0857,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824366007229284e-05,
+      "loss": 1.0165,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982187335927745e-05,
+      "loss": 0.9773,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819363306486814e-05,
+      "loss": 0.9627,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816835853305306e-05,
+      "loss": 1.0736,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814291004211695e-05,
+      "loss": 0.9905,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811728763715587e-05,
+      "loss": 0.9823,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809149136357387e-05,
+      "loss": 0.9941,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806552126708322e-05,
+      "loss": 1.0074,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980393773937042e-05,
+      "loss": 0.9188,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980130597897651e-05,
+      "loss": 0.9262,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798656850190192e-05,
+      "loss": 1.0517,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795990357705853e-05,
+      "loss": 1.007,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979330650624865e-05,
+      "loss": 1.06,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97906053005745e-05,
+      "loss": 1.1018,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787886745470067e-05,
+      "loss": 1.0161,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978515084575276e-05,
+      "loss": 0.9865,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782397606270738e-05,
+      "loss": 1.2085,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977962703190287e-05,
+      "loss": 0.9255,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776839127558744e-05,
+      "loss": 1.0249,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774033898178668e-05,
+      "loss": 0.9838,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771211348733644e-05,
+      "loss": 0.9712,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976837148422537e-05,
+      "loss": 0.9958,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976551430968622e-05,
+      "loss": 1.0248,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976263983017925e-05,
+      "loss": 1.1418,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759748050798176e-05,
+      "loss": 0.9808,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756838976667373e-05,
+      "loss": 1.1026,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753912612941867e-05,
+      "loss": 1.1482,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750968964807305e-05,
+      "loss": 1.1188,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748008037479988e-05,
+      "loss": 1.0532,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745029836206813e-05,
+      "loss": 1.0273,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742034366265308e-05,
+      "loss": 1.0363,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739021632963584e-05,
+      "loss": 1.0821,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735991641640354e-05,
+      "loss": 0.9579,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732944397664915e-05,
+      "loss": 1.0749,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729879906437124e-05,
+      "loss": 0.9643,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726798173387417e-05,
+      "loss": 1.0482,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723699203976768e-05,
+      "loss": 1.0518,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720583003696708e-05,
+      "loss": 1.0277,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971744957806929e-05,
+      "loss": 1.0539,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97142989326471e-05,
+      "loss": 1.1261,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971113107301324e-05,
+      "loss": 1.0665,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707946004781305e-05,
+      "loss": 1.0359,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704743733595394e-05,
+      "loss": 1.0673,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701524265130088e-05,
+      "loss": 0.9193,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969828760509044e-05,
+      "loss": 1.0486,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695033759211972e-05,
+      "loss": 0.9321,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969176273326066e-05,
+      "loss": 1.0751,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688474533032916e-05,
+      "loss": 1.0429,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685169164355594e-05,
+      "loss": 0.9218,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681846633085968e-05,
+      "loss": 1.1199,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678506945111727e-05,
+      "loss": 1.0038,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675150106350957e-05,
+      "loss": 0.987,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671776122752145e-05,
+      "loss": 0.9921,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668385000294156e-05,
+      "loss": 1.1113,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966497674498622e-05,
+      "loss": 1.023,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661551362867926e-05,
+      "loss": 1.038,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658108860009234e-05,
+      "loss": 1.0056,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654649242510412e-05,
+      "loss": 1.0078,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965117251650208e-05,
+      "loss": 0.9425,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647678688145163e-05,
+      "loss": 0.8794,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644167763630892e-05,
+      "loss": 0.9562,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640639749180804e-05,
+      "loss": 1.1024,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637094651046707e-05,
+      "loss": 1.0845,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963353247551069e-05,
+      "loss": 1.0339,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96299532288851e-05,
+      "loss": 1.0999,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626356917512538e-05,
+      "loss": 1.0032,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9622743547765845e-05,
+      "loss": 1.0988,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619113126048086e-05,
+      "loss": 1.1032,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615465658792546e-05,
+      "loss": 0.9792,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611801152462715e-05,
+      "loss": 0.9861,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608119613552278e-05,
+      "loss": 1.0892,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96044210485851e-05,
+      "loss": 1.0133,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960070546411522e-05,
+      "loss": 1.0414,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596972866726835e-05,
+      "loss": 0.9747,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9593223263034288e-05,
+      "loss": 0.9685,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958945665968206e-05,
+      "loss": 0.8743,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585673063344753e-05,
+      "loss": 1.0342,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581872480727095e-05,
+      "loss": 1.0507,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578054918563893e-05,
+      "loss": 1.0207,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574220383620054e-05,
+      "loss": 1.0882,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570368882690572e-05,
+      "loss": 0.9782,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566500422600482e-05,
+      "loss": 1.1303,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956261501020489e-05,
+      "loss": 1.2799,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558712652388932e-05,
+      "loss": 1.0535,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955479335606778e-05,
+      "loss": 1.0697,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955085712818662e-05,
+      "loss": 1.1006,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546903975720636e-05,
+      "loss": 1.0432,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954293390567501e-05,
+      "loss": 1.0783,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538946925084898e-05,
+      "loss": 1.0622,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534943041015425e-05,
+      "loss": 0.8598,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953092226056167e-05,
+      "loss": 1.0235,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526884590848646e-05,
+      "loss": 1.0579,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952283003903131e-05,
+      "loss": 0.8823,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951875861229452e-05,
+      "loss": 0.9815,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514670317853043e-05,
+      "loss": 1.0394,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 1.1034,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506443154864536e-05,
+      "loss": 1.0242,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502304300896442e-05,
+      "loss": 0.9295,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498148608381506e-05,
+      "loss": 0.9371,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493976084683814e-05,
+      "loss": 0.9686,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489786737197286e-05,
+      "loss": 1.0366,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485580573345645e-05,
+      "loss": 1.1104,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481357600582425e-05,
+      "loss": 1.0952,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477117826390934e-05,
+      "loss": 1.0771,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947286125828426e-05,
+      "loss": 1.003,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468587903805256e-05,
+      "loss": 1.0663,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464297770526507e-05,
+      "loss": 1.0059,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459990866050337e-05,
+      "loss": 1.0338,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94556671980088e-05,
+      "loss": 1.0562,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451326774063636e-05,
+      "loss": 1.1058,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94469696019063e-05,
+      "loss": 1.1528,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442595689257898e-05,
+      "loss": 1.0276,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438205043869232e-05,
+      "loss": 1.0071,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943379767352073e-05,
+      "loss": 1.0381,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429373586022472e-05,
+      "loss": 1.0016,
+      "step": 328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424932789214158e-05,
+      "loss": 0.9238,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942047529096509e-05,
+      "loss": 1.0245,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416001099174183e-05,
+      "loss": 1.0655,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941151022176991e-05,
+      "loss": 1.0373,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407002666710334e-05,
+      "loss": 1.0664,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402478441983058e-05,
+      "loss": 1.0436,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397937555605235e-05,
+      "loss": 0.9819,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393380015623535e-05,
+      "loss": 1.1562,
+      "step": 336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388805830114132e-05,
+      "loss": 0.8764,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938421500718272e-05,
+      "loss": 1.1058,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937960755496445e-05,
+      "loss": 0.9259,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937498348162396e-05,
+      "loss": 0.9354,
+      "step": 340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937034279535533e-05,
+      "loss": 0.9948,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936568550438208e-05,
+      "loss": 1.0582,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361011616957165e-05,
+      "loss": 1.0361,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935632114136293e-05,
+      "loss": 0.9712,
+      "step": 344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351614085911134e-05,
+      "loss": 1.1029,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346890458942914e-05,
+      "loss": 0.9734,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342150268828754e-05,
+      "loss": 1.0093,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933739352396851e-05,
+      "loss": 1.0836,
+      "step": 348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933262023279137e-05,
+      "loss": 1.0266,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327830403755832e-05,
+      "loss": 1.1339,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323024045349704e-05,
+      "loss": 0.9362,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318201166090097e-05,
+      "loss": 0.9422,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9313361774523387e-05,
+      "loss": 1.1174,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308505879225207e-05,
+      "loss": 0.8944,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930363348880044e-05,
+      "loss": 1.0049,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92987446118832e-05,
+      "loss": 0.993,
+      "step": 356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929383925713682e-05,
+      "loss": 1.0758,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288917433253823e-05,
+      "loss": 1.024,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283979148955927e-05,
+      "loss": 1.0144,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927902441299401e-05,
+      "loss": 1.0515,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92740532341481e-05,
+      "loss": 1.1693,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269065621227376e-05,
+      "loss": 1.1387,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264061583070126e-05,
+      "loss": 1.0449,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259041128543754e-05,
+      "loss": 0.9822,
+      "step": 364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925400426654475e-05,
+      "loss": 1.0456,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248951005998678e-05,
+      "loss": 0.99,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243881355860163e-05,
+      "loss": 0.9867,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.1193,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9233692922769497e-05,
+      "loss": 1.0152,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228574157871745e-05,
+      "loss": 0.9656,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922343903949032e-05,
+      "loss": 1.0229,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218287576724903e-05,
+      "loss": 1.0,
+      "step": 372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921311977870413e-05,
+      "loss": 1.0122,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92079356545856e-05,
+      "loss": 1.0574,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920273521355583e-05,
+      "loss": 0.9855,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197518464830255e-05,
+      "loss": 0.8773,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192285417653208e-05,
+      "loss": 1.0066,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187036081297907e-05,
+      "loss": 1.1173,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181770465066426e-05,
+      "loss": 1.0675,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91764885782897e-05,
+      "loss": 1.0952,
+      "step": 380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917119043032749e-05,
+      "loss": 0.898,
+      "step": 381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165876030568365e-05,
+      "loss": 1.0363,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916054538842971e-05,
+      "loss": 1.1136,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155198513357677e-05,
+      "loss": 1.0828,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149835414827193e-05,
+      "loss": 0.9551,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144456102341928e-05,
+      "loss": 1.0687,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139060585434287e-05,
+      "loss": 0.9852,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913364887366539e-05,
+      "loss": 0.9951,
+      "step": 388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912822097662505e-05,
+      "loss": 1.1121,
+      "step": 389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122776903931776e-05,
+      "loss": 0.9408,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117316665232715e-05,
+      "loss": 1.1265,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911184027020369e-05,
+      "loss": 0.9619,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106347728549134e-05,
+      "loss": 0.9748,
+      "step": 393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100839050002098e-05,
+      "loss": 1.0213,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095314244324233e-05,
+      "loss": 1.136,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908977332130576e-05,
+      "loss": 1.0657,
+      "step": 396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908421629076547e-05,
+      "loss": 0.9701,
+      "step": 397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078643162550686e-05,
+      "loss": 1.0995,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073053946537265e-05,
+      "loss": 0.9926,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067448652629573e-05,
+      "loss": 1.1041,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061827290760466e-05,
+      "loss": 0.9751,
+      "step": 401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056189870891266e-05,
+      "loss": 1.0071,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905053640301176e-05,
+      "loss": 1.0108,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044866897140165e-05,
+      "loss": 0.9498,
+      "step": 404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039181363323128e-05,
+      "loss": 1.1555,
+      "step": 405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033479811635687e-05,
+      "loss": 1.0037,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027762252181272e-05,
+      "loss": 1.0307,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022028695091678e-05,
+      "loss": 1.1141,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016279150527044e-05,
+      "loss": 0.9423,
+      "step": 409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901051362867585e-05,
+      "loss": 0.9885,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004732139754875e-05,
+      "loss": 1.0891,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998934694009207e-05,
+      "loss": 0.8694,
+      "step": 412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993121301712194e-05,
+      "loss": 1.0453,
+      "step": 413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898729197316546e-05,
+      "loss": 1.064,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898144671869885e-05,
+      "loss": 1.0784,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975585548670444e-05,
+      "loss": 1.0275,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896970847346653e-05,
+      "loss": 1.0602,
+      "step": 417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896381550350156e-05,
+      "loss": 0.986,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957906649218167e-05,
+      "loss": 0.9449,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951981921087133e-05,
+      "loss": 1.0912,
+      "step": 420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946041329607364e-05,
+      "loss": 0.9684,
+      "step": 421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940084885305875e-05,
+      "loss": 1.0854,
+      "step": 422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934112598737777e-05,
+      "loss": 0.9689,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928124480486258e-05,
+      "loss": 1.108,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892212054116255e-05,
+      "loss": 0.9416,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8916100791405925e-05,
+      "loss": 1.0297,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.8853,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89040139032911e-05,
+      "loss": 1.0542,
+      "step": 428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889794678635145e-05,
+      "loss": 0.9289,
+      "step": 429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891863901815962e-05,
+      "loss": 0.9652,
+      "step": 430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88857652604638e-05,
+      "loss": 0.8988,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879650873102055e-05,
+      "loss": 1.0342,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873520750565716e-05,
+      "loss": 0.9505,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886737490371767e-05,
+      "loss": 0.9446,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861213343448645e-05,
+      "loss": 1.1764,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885503608067724e-05,
+      "loss": 1.2689,
+      "step": 436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884884312634985e-05,
+      "loss": 1.0402,
+      "step": 437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842634491440704e-05,
+      "loss": 1.0064,
+      "step": 438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836410186951805e-05,
+      "loss": 1.1046,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883017022391292e-05,
+      "loss": 0.9783,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823914613381568e-05,
+      "loss": 0.9758,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8817643366443e-05,
+      "loss": 0.9861,
+      "step": 442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811356494210166e-05,
+      "loss": 1.0803,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805054007823716e-05,
+      "loss": 1.0105,
+      "step": 444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798735918451963e-05,
+      "loss": 1.0224,
+      "step": 445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792402237290865e-05,
+      "loss": 1.0147,
+      "step": 446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878605297556402e-05,
+      "loss": 1.057,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779688144522625e-05,
+      "loss": 0.9985,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773307755445468e-05,
+      "loss": 0.9799,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8766911819638917e-05,
+      "loss": 1.0228,
+      "step": 450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876050034843688e-05,
+      "loss": 0.9089,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754073353200796e-05,
+      "loss": 0.9996,
+      "step": 452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874763084531961e-05,
+      "loss": 0.9506,
+      "step": 453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741172836209773e-05,
+      "loss": 1.0606,
+      "step": 454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873469933731518e-05,
+      "loss": 0.9391,
+      "step": 455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872821036010719e-05,
+      "loss": 0.9784,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872170591608459e-05,
+      "loss": 0.9766,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871518601677357e-05,
+      "loss": 0.9846,
+      "step": 458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708650673727708e-05,
+      "loss": 1.0851,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702099898527955e-05,
+      "loss": 0.8785,
+      "step": 460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86955337027826e-05,
+      "loss": 1.0858,
+      "step": 461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688952098127265e-05,
+      "loss": 1.1099,
+      "step": 462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682355096224873e-05,
+      "loss": 1.0563,
+      "step": 463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675742708765633e-05,
+      "loss": 1.0595,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866911494746702e-05,
+      "loss": 0.8897,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866247182407375e-05,
+      "loss": 1.0525,
+      "step": 466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655813350357764e-05,
+      "loss": 0.9812,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649139538118196e-05,
+      "loss": 1.0569,
+      "step": 468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642450399181373e-05,
+      "loss": 1.156,
+      "step": 469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635745945400772e-05,
+      "loss": 1.0206,
+      "step": 470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862902618865701e-05,
+      "loss": 0.9696,
+      "step": 471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862229114085783e-05,
+      "loss": 1.0654,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615540813938063e-05,
+      "loss": 0.9939,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608775219859618e-05,
+      "loss": 1.008,
+      "step": 474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601994370611452e-05,
+      "loss": 0.9662,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859519827820957e-05,
+      "loss": 0.9245,
+      "step": 476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588386954696972e-05,
+      "loss": 1.0275,
+      "step": 477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581560412143663e-05,
+      "loss": 0.9314,
+      "step": 478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85747186626466e-05,
+      "loss": 0.9696,
+      "step": 479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567861718329705e-05,
+      "loss": 1.1141,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856098959134381e-05,
+      "loss": 0.9806,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855410229386667e-05,
+      "loss": 0.9459,
+      "step": 482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547199838102904e-05,
+      "loss": 1.0808,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540282236284005e-05,
+      "loss": 0.9773,
+      "step": 484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533349500668295e-05,
+      "loss": 1.1119,
+      "step": 485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526401643540924e-05,
+      "loss": 1.0087,
+      "step": 486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519438677213834e-05,
+      "loss": 1.0375,
+      "step": 487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851246061402574e-05,
+      "loss": 1.0478,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850546746634211e-05,
+      "loss": 1.0229,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498459246555143e-05,
+      "loss": 0.9939,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849143596708375e-05,
+      "loss": 1.0645,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484397640373517e-05,
+      "loss": 0.8781,
+      "step": 492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477344278896708e-05,
+      "loss": 0.9516,
+      "step": 493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470275895152228e-05,
+      "loss": 1.0237,
+      "step": 494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846319250166559e-05,
+      "loss": 1.069,
+      "step": 495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456094110988914e-05,
+      "loss": 1.0403,
+      "step": 496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84489807357009e-05,
+      "loss": 1.0142,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441852388406788e-05,
+      "loss": 1.0581,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8434709081738364e-05,
+      "loss": 0.9616,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427550828353912e-05,
+      "loss": 1.028,
+      "step": 500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420377640938204e-05,
+      "loss": 1.0888,
+      "step": 501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413189532202488e-05,
+      "loss": 1.0171,
+      "step": 502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840598651488443e-05,
+      "loss": 1.0198,
+      "step": 503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398768601748143e-05,
+      "loss": 0.9213,
+      "step": 504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839153580558411e-05,
+      "loss": 0.9659,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384288139209204e-05,
+      "loss": 1.0121,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837702561546664e-05,
+      "loss": 0.9429,
+      "step": 507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369748247225965e-05,
+      "loss": 0.9574,
+      "step": 508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362456047383032e-05,
+      "loss": 0.8838,
+      "step": 509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355149028859975e-05,
+      "loss": 1.0912,
+      "step": 510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347827204605187e-05,
+      "loss": 1.0289,
+      "step": 511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83404905875933e-05,
+      "loss": 0.9732,
+      "step": 512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833313919082515e-05,
+      "loss": 0.9304,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832577302732778e-05,
+      "loss": 0.9851,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8318392110154387e-05,
+      "loss": 0.9973,
+      "step": 515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310996452384312e-05,
+      "loss": 1.033,
+      "step": 516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303586067123028e-05,
+      "loss": 1.1081,
+      "step": 517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82961609675021e-05,
+      "loss": 1.0485,
+      "step": 518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828872116667916e-05,
+      "loss": 1.0581,
+      "step": 519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82812666778379e-05,
+      "loss": 1.0565,
+      "step": 520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273797514188043e-05,
+      "loss": 1.0394,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266313688965307e-05,
+      "loss": 0.9207,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258815215431395e-05,
+      "loss": 1.1348,
+      "step": 523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825130210687397e-05,
+      "loss": 1.0388,
+      "step": 524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824377437660663e-05,
+      "loss": 1.0929,
+      "step": 525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236232037968873e-05,
+      "loss": 0.9733,
+      "step": 526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228675104326096e-05,
+      "loss": 1.1005,
+      "step": 527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221103589069553e-05,
+      "loss": 1.0583,
+      "step": 528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821351750561634e-05,
+      "loss": 1.0498,
+      "step": 529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820591686740936e-05,
+      "loss": 1.0468,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198301687917325e-05,
+      "loss": 1.0576,
+      "step": 531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190671980634698e-05,
+      "loss": 1.0863,
+      "step": 532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818302775908169e-05,
+      "loss": 0.9544,
+      "step": 533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175369036804243e-05,
+      "loss": 1.0569,
+      "step": 534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167695827373982e-05,
+      "loss": 0.9525,
+      "step": 535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160008144388212e-05,
+      "loss": 1.044,
+      "step": 536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8152306001469875e-05,
+      "loss": 0.8825,
+      "step": 537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814458941226755e-05,
+      "loss": 1.0051,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136858390455406e-05,
+      "loss": 0.964,
+      "step": 539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129112949733193e-05,
+      "loss": 1.2041,
+      "step": 540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121353103826213e-05,
+      "loss": 1.0948,
+      "step": 541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113578866485288e-05,
+      "loss": 0.9717,
+      "step": 542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810579025148674e-05,
+      "loss": 1.0894,
+      "step": 543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097987272632384e-05,
+      "loss": 0.963,
+      "step": 544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0476,
+      "step": 545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082338278690704e-05,
+      "loss": 1.1185,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807449229133416e-05,
+      "loss": 1.0054,
+      "step": 547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066631995583318e-05,
+      "loss": 1.0131,
+      "step": 548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058757405367003e-05,
+      "loss": 1.1011,
+      "step": 549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805086853463938e-05,
+      "loss": 1.0835,
+      "step": 550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042965397379904e-05,
+      "loss": 0.9314,
+      "step": 551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035048007593322e-05,
+      "loss": 1.0555,
+      "step": 552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8027116379309637e-05,
+      "loss": 1.0529,
+      "step": 553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019170526584083e-05,
+      "loss": 0.9156,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011210463497095e-05,
+      "loss": 1.0872,
+      "step": 555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003236204154296e-05,
+      "loss": 0.9799,
+      "step": 556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799524776268646e-05,
+      "loss": 1.0613,
+      "step": 557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987245153249496e-05,
+      "loss": 0.8917,
+      "step": 558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979228390024417e-05,
+      "loss": 1.006,
+      "step": 559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971197487217322e-05,
+      "loss": 1.07,
+      "step": 560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796315245905936e-05,
+      "loss": 0.9515,
+      "step": 561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795509331980672e-05,
+      "loss": 0.9812,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947020083740575e-05,
+      "loss": 1.137,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7938932765167107e-05,
+      "loss": 1.1004,
+      "step": 564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930831378417437e-05,
+      "loss": 1.0575,
+      "step": 565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792271593784761e-05,
+      "loss": 1.1174,
+      "step": 566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914586457838592e-05,
+      "loss": 1.0646,
+      "step": 567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906442952796212e-05,
+      "loss": 1.0319,
+      "step": 568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898285437151163e-05,
+      "loss": 1.0151,
+      "step": 569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890113925358954e-05,
+      "loss": 1.0234,
+      "step": 570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788192843189991e-05,
+      "loss": 1.2047,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7873728971279116e-05,
+      "loss": 1.0391,
+      "step": 572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786551555802643e-05,
+      "loss": 0.9231,
+      "step": 573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857288206696405e-05,
+      "loss": 1.0342,
+      "step": 574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784904693186832e-05,
+      "loss": 0.971,
+      "step": 575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840791748146112e-05,
+      "loss": 1.0648,
+      "step": 576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783252267015837e-05,
+      "loss": 1.0146,
+      "step": 577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824239712558303e-05,
+      "loss": 0.978,
+      "step": 578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815942890023716e-05,
+      "loss": 1.144,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7807632217256988e-05,
+      "loss": 0.9478,
+      "step": 580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779930770898503e-05,
+      "loss": 1.0221,
+      "step": 581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790969379959276e-05,
+      "loss": 1.1083,
+      "step": 582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778261724495566e-05,
+      "loss": 0.9185,
+      "step": 583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774251318774568e-05,
+      "loss": 1.0508,
+      "step": 584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776587161624083e-05,
+      "loss": 0.9467,
+      "step": 585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757478152203683e-05,
+      "loss": 1.1174,
+      "step": 586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749070941536763e-05,
+      "loss": 0.985,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774064999913805e-05,
+      "loss": 1.0967,
+      "step": 588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732215339929874e-05,
+      "loss": 0.9661,
+      "step": 589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772376697885885e-05,
+      "loss": 0.983,
+      "step": 590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715304930895894e-05,
+      "loss": 1.1112,
+      "step": 591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706829211036172e-05,
+      "loss": 0.9431,
+      "step": 592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698339834299064e-05,
+      "loss": 0.91,
+      "step": 593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689836815728164e-05,
+      "loss": 0.9477,
+      "step": 594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681320170391236e-05,
+      "loss": 0.9234,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672789913380192e-05,
+      "loss": 0.9149,
+      "step": 596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664246059811058e-05,
+      "loss": 1.012,
+      "step": 597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765568862482397e-05,
+      "loss": 1.1162,
+      "step": 598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647117623583107e-05,
+      "loss": 1.1264,
+      "step": 599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638533071276712e-05,
+      "loss": 0.9201,
+      "step": 600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629934983117025e-05,
+      "loss": 0.9787,
+      "step": 601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762132337434028e-05,
+      "loss": 1.0927,
+      "step": 602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612698260206668e-05,
+      "loss": 0.9746,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604059656000313e-05,
+      "loss": 1.0678,
+      "step": 604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759540757702924e-05,
+      "loss": 0.9678,
+      "step": 605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586742038625357e-05,
+      "loss": 0.9799,
+      "step": 606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757806305614442e-05,
+      "loss": 1.0382,
+      "step": 607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7569370644966007e-05,
+      "loss": 1.1101,
+      "step": 608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560664820493502e-05,
+      "loss": 1.1365,
+      "step": 609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551945598154044e-05,
+      "loss": 1.115,
+      "step": 610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754321299339852e-05,
+      "loss": 0.9999,
+      "step": 611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753446702170154e-05,
+      "loss": 0.999,
+      "step": 612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525707698561383e-05,
+      "loss": 1.0083,
+      "step": 613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516935039500007e-05,
+      "loss": 1.0636,
+      "step": 614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750814906006298e-05,
+      "loss": 1.0376,
+      "step": 615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499349775819497e-05,
+      "loss": 1.1103,
+      "step": 616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490537202362313e-05,
+      "loss": 1.0945,
+      "step": 617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481711355307735e-05,
+      "loss": 1.0328,
+      "step": 618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7472872250295603e-05,
+      "loss": 1.0303,
+      "step": 619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464019902989234e-05,
+      "loss": 1.0486,
+      "step": 620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455154329075427e-05,
+      "loss": 1.0767,
+      "step": 621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744627554426441e-05,
+      "loss": 0.9857,
+      "step": 622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437383564289816e-05,
+      "loss": 1.0462,
+      "step": 623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428478404908675e-05,
+      "loss": 1.042,
+      "step": 624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741956008190136e-05,
+      "loss": 1.1192,
+      "step": 625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410628611071576e-05,
+      "loss": 1.0238,
+      "step": 626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401684008246326e-05,
+      "loss": 1.0647,
+      "step": 627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739272628927588e-05,
+      "loss": 1.0534,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7383755470033756e-05,
+      "loss": 1.0037,
+      "step": 629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374771566416684e-05,
+      "loss": 1.0622,
+      "step": 630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365774594344572e-05,
+      "loss": 1.0817,
+      "step": 631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73567645697605e-05,
+      "loss": 1.0678,
+      "step": 632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347741508630673e-05,
+      "loss": 0.956,
+      "step": 633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338705426944393e-05,
+      "loss": 0.9309,
+      "step": 634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329656340714037e-05,
+      "loss": 0.9983,
+      "step": 635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320594265975025e-05,
+      "loss": 0.9068,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73115192187858e-05,
+      "loss": 0.8823,
+      "step": 637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302431215227782e-05,
+      "loss": 0.9902,
+      "step": 638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293330271405367e-05,
+      "loss": 0.9584,
+      "step": 639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284216403445865e-05,
+      "loss": 0.9598,
+      "step": 640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275089627499493e-05,
+      "loss": 1.0631,
+      "step": 641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265949959739345e-05,
+      "loss": 0.982,
+      "step": 642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725679741636136e-05,
+      "loss": 0.9639,
+      "step": 643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247632013584296e-05,
+      "loss": 1.033,
+      "step": 644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7238453767649683e-05,
+      "loss": 1.037,
+      "step": 645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229262694821825e-05,
+      "loss": 1.0494,
+      "step": 646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220058811387754e-05,
+      "loss": 1.0285,
+      "step": 647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210842133657197e-05,
+      "loss": 0.9003,
+      "step": 648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720161267796256e-05,
+      "loss": 0.9587,
+      "step": 649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192370460658888e-05,
+      "loss": 0.8851,
+      "step": 650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183115498123843e-05,
+      "loss": 0.9959,
+      "step": 651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7173847806757662e-05,
+      "loss": 1.024,
+      "step": 652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7164567402983153e-05,
+      "loss": 1.1369,
+      "step": 653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155274303245642e-05,
+      "loss": 1.0514,
+      "step": 654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714596852401296e-05,
+      "loss": 0.9785,
+      "step": 655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136650081775395e-05,
+      "loss": 1.1178,
+      "step": 656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127318993045686e-05,
+      "loss": 1.0403,
+      "step": 657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117975274358975e-05,
+      "loss": 1.0075,
+      "step": 658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108618942272786e-05,
+      "loss": 1.0077,
+      "step": 659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099250013367e-05,
+      "loss": 1.0364,
+      "step": 660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7089868504243816e-05,
+      "loss": 0.9999,
+      "step": 661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080474431527724e-05,
+      "loss": 1.0608,
+      "step": 662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.9611,
+      "step": 663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061648661926068e-05,
+      "loss": 1.082,
+      "step": 664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705221699840069e-05,
+      "loss": 1.0361,
+      "step": 665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042772838002704e-05,
+      "loss": 0.974,
+      "step": 666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033316197467634e-05,
+      "loss": 1.045,
+      "step": 667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70238470935531e-05,
+      "loss": 1.0623,
+      "step": 668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701436554303882e-05,
+      "loss": 0.9795,
+      "step": 669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004871562726563e-05,
+      "loss": 1.0353,
+      "step": 670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699536516944013e-05,
+      "loss": 1.0013,
+      "step": 671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69858463800253e-05,
+      "loss": 1.1328,
+      "step": 672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976315211349848e-05,
+      "loss": 1.0368,
+      "step": 673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966771680303462e-05,
+      "loss": 0.9242,
+      "step": 674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957215803797748e-05,
+      "loss": 1.0086,
+      "step": 675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947647598766183e-05,
+      "loss": 1.0575,
+      "step": 676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938067082164093e-05,
+      "loss": 1.0118,
+      "step": 677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692847427096862e-05,
+      "loss": 0.9457,
+      "step": 678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6918869182178698e-05,
+      "loss": 0.9153,
+      "step": 679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909251832815005e-05,
+      "loss": 0.9573,
+      "step": 680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899622239919965e-05,
+      "loss": 1.0043,
+      "step": 681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889980420557674e-05,
+      "loss": 0.9114,
+      "step": 682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880326391813917e-05,
+      "loss": 0.957,
+      "step": 683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6870660170796094e-05,
+      "loss": 0.9858,
+      "step": 684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860981774633228e-05,
+      "loss": 1.0881,
+      "step": 685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851291220475908e-05,
+      "loss": 1.0752,
+      "step": 686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841588525496268e-05,
+      "loss": 1.0071,
+      "step": 687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683187370688795e-05,
+      "loss": 0.9215,
+      "step": 688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822146781866097e-05,
+      "loss": 0.8374,
+      "step": 689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812407767667293e-05,
+      "loss": 1.03,
+      "step": 690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680265668154954e-05,
+      "loss": 1.0224,
+      "step": 691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679289354079224e-05,
+      "loss": 0.957,
+      "step": 692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6783118362696162e-05,
+      "loss": 1.0607,
+      "step": 693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773331164583393e-05,
+      "loss": 1.0057,
+      "step": 694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763531963797325e-05,
+      "loss": 1.0486,
+      "step": 695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675372077770262e-05,
+      "loss": 1.103,
+      "step": 696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743897623685178e-05,
+      "loss": 1.0367,
+      "step": 697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734062519152113e-05,
+      "loss": 1.1671,
+      "step": 698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724215481531704e-05,
+      "loss": 1.0899,
+      "step": 699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714356528273382e-05,
+      "loss": 0.9339,
+      "step": 700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6704485676847695e-05,
+      "loss": 1.0635,
+      "step": 701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6694602944746275e-05,
+      "loss": 0.9791,
+      "step": 702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684708349481808e-05,
+      "loss": 1.0137,
+      "step": 703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674801908587988e-05,
+      "loss": 0.9719,
+      "step": 704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666488363961952e-05,
+      "loss": 1.0617,
+      "step": 705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654953560152063e-05,
+      "loss": 0.9899,
+      "step": 706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6645011687782196e-05,
+      "loss": 1.027,
+      "step": 707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635058040127408e-05,
+      "loss": 1.0079,
+      "step": 708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662509263482604e-05,
+      "loss": 0.9832,
+      "step": 709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6615115489537285e-05,
+      "loss": 0.9349,
+      "step": 710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6605126621941127e-05,
+      "loss": 0.9694,
+      "step": 711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595126049738328e-05,
+      "loss": 0.9241,
+      "step": 712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658511379065039e-05,
+      "loss": 0.9774,
+      "step": 713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657508986241952e-05,
+      "loss": 0.9507,
+      "step": 714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565054282808617e-05,
+      "loss": 1.0445,
+      "step": 715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555007069601208e-05,
+      "loss": 1.0163,
+      "step": 716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544948240601453e-05,
+      "loss": 0.9039,
+      "step": 717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.653487781363408e-05,
+      "loss": 1.0291,
+      "step": 718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524795806544384e-05,
+      "loss": 1.0336,
+      "step": 719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514702237198172e-05,
+      "loss": 1.1045,
+      "step": 720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504597123481737e-05,
+      "loss": 1.0446,
+      "step": 721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6494480483301836e-05,
+      "loss": 0.9759,
+      "step": 722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484352334585654e-05,
+      "loss": 1.0232,
+      "step": 723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474212695280756e-05,
+      "loss": 1.0596,
+      "step": 724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6464061583355088e-05,
+      "loss": 1.0333,
+      "step": 725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6453899016796903e-05,
+      "loss": 1.0234,
+      "step": 726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443725013614772e-05,
+      "loss": 0.9545,
+      "step": 727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433539591837527e-05,
+      "loss": 1.0994,
+      "step": 728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423342769514227e-05,
+      "loss": 0.9335,
+      "step": 729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413134564714142e-05,
+      "loss": 1.0766,
+      "step": 730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640291499552671e-05,
+      "loss": 1.0006,
+      "step": 731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392684080061503e-05,
+      "loss": 1.1148,
+      "step": 732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382441836448203e-05,
+      "loss": 0.9362,
+      "step": 733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637218828283657e-05,
+      "loss": 0.9155,
+      "step": 734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636192343739639e-05,
+      "loss": 0.9952,
+      "step": 735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635164731831748e-05,
+      "loss": 0.9724,
+      "step": 736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341359943809626e-05,
+      "loss": 1.0119,
+      "step": 737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633106133210255e-05,
+      "loss": 0.8702,
+      "step": 738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63207515014459e-05,
+      "loss": 1.0013,
+      "step": 739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310430470109196e-05,
+      "loss": 1.0206,
+      "step": 740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300098256381807e-05,
+      "loss": 1.0876,
+      "step": 741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.628975487857293e-05,
+      "loss": 1.0504,
+      "step": 742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627940035501152e-05,
+      "loss": 0.9877,
+      "step": 743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626903470404631e-05,
+      "loss": 1.0494,
+      "step": 744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625865794404573e-05,
+      "loss": 0.9823,
+      "step": 745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248270093397915e-05,
+      "loss": 0.8848,
+      "step": 746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237871170510636e-05,
+      "loss": 1.0968,
+      "step": 747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62274611938113e-05,
+      "loss": 0.901,
+      "step": 748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621704018174688e-05,
+      "loss": 1.039,
+      "step": 749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206608152783924e-05,
+      "loss": 1.0096,
+      "step": 750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6196165125408507e-05,
+      "loss": 0.9397,
+      "step": 751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185711118126164e-05,
+      "loss": 1.0388,
+      "step": 752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617524614946192e-05,
+      "loss": 1.0634,
+      "step": 753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164770237960204e-05,
+      "loss": 1.0586,
+      "step": 754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154283402184846e-05,
+      "loss": 1.0547,
+      "step": 755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614378566071903e-05,
+      "loss": 1.0706,
+      "step": 756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133277032165264e-05,
+      "loss": 0.9817,
+      "step": 757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122757535145346e-05,
+      "loss": 1.0162,
+      "step": 758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611222718830035e-05,
+      "loss": 0.9974,
+      "step": 759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101686010290556e-05,
+      "loss": 1.0385,
+      "step": 760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091134019795447e-05,
+      "loss": 0.9456,
+      "step": 761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080571235513666e-05,
+      "loss": 1.0276,
+      "step": 762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606999767616298e-05,
+      "loss": 0.9206,
+      "step": 763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605941336048025e-05,
+      "loss": 0.9756,
+      "step": 764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604881830722141e-05,
+      "loss": 0.9648,
+      "step": 765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60382125351614e-05,
+      "loss": 1.0213,
+      "step": 766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027596063094174e-05,
+      "loss": 1.0014,
+      "step": 767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016968909832632e-05,
+      "loss": 1.1058,
+      "step": 768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600633109420861e-05,
+      "loss": 0.9459,
+      "step": 769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995682635072843e-05,
+      "loss": 1.09,
+      "step": 770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985023551294907e-05,
+      "loss": 1.1158,
+      "step": 771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597435386176323e-05,
+      "loss": 0.9966,
+      "step": 772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5963673585385016e-05,
+      "loss": 0.9845,
+      "step": 773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952982741086238e-05,
+      "loss": 0.9748,
+      "step": 774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5942281347811596e-05,
+      "loss": 1.0895,
+      "step": 775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931569424524477e-05,
+      "loss": 1.1871,
+      "step": 776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920846990206934e-05,
+      "loss": 0.986,
+      "step": 777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591011406385964e-05,
+      "loss": 1.052,
+      "step": 778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589937066450187e-05,
+      "loss": 0.9269,
+      "step": 779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888616811171452e-05,
+      "loss": 1.0525,
+      "step": 780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 1.1237,
+      "step": 781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586707781883656e-05,
+      "loss": 1.042,
+      "step": 782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5856292718000235e-05,
+      "loss": 0.9764,
+      "step": 783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584549723952748e-05,
+      "loss": 0.9498,
+      "step": 784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5834691402548415e-05,
+      "loss": 0.9808,
+      "step": 785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5823875226211507e-05,
+      "loss": 1.0024,
+      "step": 786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813048729683543e-05,
+      "loss": 1.0588,
+      "step": 787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802211932149614e-05,
+      "loss": 0.8983,
+      "step": 788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791364852813047e-05,
+      "loss": 1.0963,
+      "step": 789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780507510895398e-05,
+      "loss": 0.9529,
+      "step": 790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5769639925636404e-05,
+      "loss": 0.9525,
+      "step": 791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575876211629396e-05,
+      "loss": 0.9296,
+      "step": 792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747874102144073e-05,
+      "loss": 0.9804,
+      "step": 793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736975902480832e-05,
+      "loss": 1.0092,
+      "step": 794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726067536616383e-05,
+      "loss": 0.9943,
+      "step": 795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571514902388088e-05,
+      "loss": 0.9165,
+      "step": 796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704220383622464e-05,
+      "loss": 0.9893,
+      "step": 797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5693281635207214e-05,
+      "loss": 0.9976,
+      "step": 798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5682332798019137e-05,
+      "loss": 1.0344,
+      "step": 799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567137389146009e-05,
+      "loss": 1.0415,
+      "step": 800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660404934949798e-05,
+      "loss": 1.0315,
+      "step": 801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564942594792579e-05,
+      "loss": 0.9606,
+      "step": 802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563843694984336e-05,
+      "loss": 0.9618,
+      "step": 803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5627437960175556e-05,
+      "loss": 1.0324,
+      "step": 804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616428998413122e-05,
+      "loss": 0.9187,
+      "step": 805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605410084064468e-05,
+      "loss": 1.0448,
+      "step": 806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5594381236655665e-05,
+      "loss": 1.0227,
+      "step": 807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558334247573035e-05,
+      "loss": 1.0774,
+      "step": 808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5572293820849754e-05,
+      "loss": 1.0876,
+      "step": 809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561235291592635e-05,
+      "loss": 0.9874,
+      "step": 810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550166907555243e-05,
+      "loss": 1.0704,
+      "step": 811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5539088688351295e-05,
+      "loss": 1.1851,
+      "step": 812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528000653611935e-05,
+      "loss": 0.6687,
+      "step": 813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55169028229857e-05,
+      "loss": 0.6852,
+      "step": 814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5505795216138498e-05,
+      "loss": 0.5976,
+      "step": 815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549467785275354e-05,
+      "loss": 0.5681,
+      "step": 816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483550752531337e-05,
+      "loss": 0.631,
+      "step": 817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472413935189656e-05,
+      "loss": 0.6254,
+      "step": 818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546126742046348e-05,
+      "loss": 0.5024,
+      "step": 819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450111228104976e-05,
+      "loss": 0.5781,
+      "step": 820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5438945377883463e-05,
+      "loss": 0.6256,
+      "step": 821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542776988958537e-05,
+      "loss": 0.5193,
+      "step": 822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541658478301421e-05,
+      "loss": 0.5569,
+      "step": 823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405390077990538e-05,
+      "loss": 0.5445,
+      "step": 824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5394185794351914e-05,
+      "loss": 0.5204,
+      "step": 825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382971951952878e-05,
+      "loss": 0.5262,
+      "step": 826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371748570664906e-05,
+      "loss": 0.5566,
+      "step": 827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360515670376373e-05,
+      "loss": 0.5698,
+      "step": 828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349273270992537e-05,
+      "loss": 0.5174,
+      "step": 829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5338021392435462e-05,
+      "loss": 0.4506,
+      "step": 830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326760054644045e-05,
+      "loss": 0.5495,
+      "step": 831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315489277573906e-05,
+      "loss": 0.5458,
+      "step": 832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304209081197425e-05,
+      "loss": 0.6553,
+      "step": 833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5292919485503662e-05,
+      "loss": 0.611,
+      "step": 834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281620510498322e-05,
+      "loss": 0.5755,
+      "step": 835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5270312176203742e-05,
+      "loss": 0.5727,
+      "step": 836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5258994502658846e-05,
+      "loss": 0.5053,
+      "step": 837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247667509919104e-05,
+      "loss": 0.5938,
+      "step": 838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52363312180565e-05,
+      "loss": 0.7924,
+      "step": 839
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5224985647159489e-05,
+      "loss": 0.5243,
+      "step": 840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213630817332985e-05,
+      "loss": 0.5438,
+      "step": 841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202266748698298e-05,
+      "loss": 0.5526,
+      "step": 842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190893461393108e-05,
+      "loss": 0.6025,
+      "step": 843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517951097557144e-05,
+      "loss": 0.561,
+      "step": 844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5168119311403611e-05,
+      "loss": 0.5186,
+      "step": 845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156718489076208e-05,
+      "loss": 0.5673,
+      "step": 846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145308528792045e-05,
+      "loss": 0.6513,
+      "step": 847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133889450770122e-05,
+      "loss": 0.5702,
+      "step": 848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512246127524561e-05,
+      "loss": 0.5445,
+      "step": 849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511102402246979e-05,
+      "loss": 0.6269,
+      "step": 850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099577712710036e-05,
+      "loss": 0.4926,
+      "step": 851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508812236624976e-05,
+      "loss": 0.5888,
+      "step": 852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50766580033884e-05,
+      "loss": 0.4709,
+      "step": 853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506518464444137e-05,
+      "loss": 0.5209,
+      "step": 854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505370230974001e-05,
+      "loss": 0.4759,
+      "step": 855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042211019631588e-05,
+      "loss": 0.6702,
+      "step": 856
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030710794479226e-05,
+      "loss": 0.5046,
+      "step": 857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5019201654661886e-05,
+      "loss": 0.5031,
+      "step": 858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5007683620574322e-05,
+      "loss": 0.6509,
+      "step": 859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4996156712627059e-05,
+      "loss": 0.5351,
+      "step": 860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4984620951246333e-05,
+      "loss": 0.5281,
+      "step": 861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973076356874081e-05,
+      "loss": 0.5847,
+      "step": 862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961522949967887e-05,
+      "loss": 0.4243,
+      "step": 863
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949960751000944e-05,
+      "loss": 0.5879,
+      "step": 864
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938389780462044e-05,
+      "loss": 0.5711,
+      "step": 865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926810058855508e-05,
+      "loss": 0.5433,
+      "step": 866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915221606701162e-05,
+      "loss": 0.5825,
+      "step": 867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4903624444534317e-05,
+      "loss": 0.5057,
+      "step": 868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4892018592905702e-05,
+      "loss": 0.5675,
+      "step": 869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488040407238146e-05,
+      "loss": 0.4703,
+      "step": 870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486878090354308e-05,
+      "loss": 0.5622,
+      "step": 871
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4857149106987393e-05,
+      "loss": 0.5979,
+      "step": 872
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845508703326504e-05,
+      "loss": 0.5317,
+      "step": 873
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833859713187777e-05,
+      "loss": 0.5876,
+      "step": 874
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482220215721379e-05,
+      "loss": 0.4549,
+      "step": 875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810536056062307e-05,
+      "loss": 0.5158,
+      "step": 876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798861430406221e-05,
+      "loss": 0.6084,
+      "step": 877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4787178300933543e-05,
+      "loss": 0.5854,
+      "step": 878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775486688347346e-05,
+      "loss": 0.5037,
+      "step": 879
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476378661336574e-05,
+      "loss": 0.4829,
+      "step": 880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4752078096721827e-05,
+      "loss": 0.6024,
+      "step": 881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4740361159163668e-05,
+      "loss": 0.597,
+      "step": 882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728635821454255e-05,
+      "loss": 0.5802,
+      "step": 883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716902104371449e-05,
+      "loss": 0.5768,
+      "step": 884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4705160028707976e-05,
+      "loss": 0.4529,
+      "step": 885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4693409615271365e-05,
+      "loss": 0.6421,
+      "step": 886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681650884883923e-05,
+      "loss": 0.4661,
+      "step": 887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4669883858382689e-05,
+      "loss": 0.615,
+      "step": 888
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658108556619417e-05,
+      "loss": 0.4985,
+      "step": 889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646325000460509e-05,
+      "loss": 0.5652,
+      "step": 890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4634533210787006e-05,
+      "loss": 0.4635,
+      "step": 891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622733208494526e-05,
+      "loss": 0.5324,
+      "step": 892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461092501449326e-05,
+      "loss": 0.5913,
+      "step": 893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599108649707899e-05,
+      "loss": 0.5431,
+      "step": 894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587284135077614e-05,
+      "loss": 0.5692,
+      "step": 895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575451491556027e-05,
+      "loss": 0.6694,
+      "step": 896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563610740111163e-05,
+      "loss": 0.5441,
+      "step": 897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551761901725402e-05,
+      "loss": 0.599,
+      "step": 898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.5007,
+      "step": 899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528040048132376e-05,
+      "loss": 0.5522,
+      "step": 900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516167074961394e-05,
+      "loss": 0.5332,
+      "step": 901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450428609892201e-05,
+      "loss": 0.6242,
+      "step": 902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4492397141067888e-05,
+      "loss": 0.575,
+      "step": 903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4480500222466849e-05,
+      "loss": 0.5436,
+      "step": 904
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4468595364200808e-05,
+      "loss": 0.5151,
+      "step": 905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456682587365759e-05,
+      "loss": 0.5247,
+      "step": 906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4444761913071721e-05,
+      "loss": 0.6823,
+      "step": 907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432833362442708e-05,
+      "loss": 0.6233,
+      "step": 908
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420896956616698e-05,
+      "loss": 0.5607,
+      "step": 909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408952716745583e-05,
+      "loss": 0.5813,
+      "step": 910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4397000663995139e-05,
+      "loss": 0.606,
+      "step": 911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4385040819544988e-05,
+      "loss": 0.5589,
+      "step": 912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373073204588556e-05,
+      "loss": 0.5539,
+      "step": 913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361097840333037e-05,
+      "loss": 0.5728,
+      "step": 914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434911474799937e-05,
+      "loss": 0.5567,
+      "step": 915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337123948822172e-05,
+      "loss": 0.5565,
+      "step": 916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325125464049725e-05,
+      "loss": 0.6242,
+      "step": 917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313119314943933e-05,
+      "loss": 0.5333,
+      "step": 918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301105522780275e-05,
+      "loss": 0.51,
+      "step": 919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289084108847777e-05,
+      "loss": 0.6843,
+      "step": 920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427705509444897e-05,
+      "loss": 0.5723,
+      "step": 921
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265018500899856e-05,
+      "loss": 0.5378,
+      "step": 922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252974349529871e-05,
+      "loss": 0.6137,
+      "step": 923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240922661681826e-05,
+      "loss": 0.6309,
+      "step": 924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228863458711915e-05,
+      "loss": 0.5622,
+      "step": 925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4216796761989621e-05,
+      "loss": 0.6125,
+      "step": 926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204722592897728e-05,
+      "loss": 0.4548,
+      "step": 927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419264097283225e-05,
+      "loss": 0.4341,
+      "step": 928
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180551923202406e-05,
+      "loss": 0.5962,
+      "step": 929
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168455465430585e-05,
+      "loss": 0.6933,
+      "step": 930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156351620952293e-05,
+      "loss": 0.5052,
+      "step": 931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144240411216144e-05,
+      "loss": 0.6242,
+      "step": 932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132121857683782e-05,
+      "loss": 0.609,
+      "step": 933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119995981829884e-05,
+      "loss": 0.5992,
+      "step": 934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107862805142084e-05,
+      "loss": 0.4893,
+      "step": 935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095722349120977e-05,
+      "loss": 0.56,
+      "step": 936
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083574635280029e-05,
+      "loss": 0.5354,
+      "step": 937
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071419685145587e-05,
+      "loss": 0.7124,
+      "step": 938
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405925752025682e-05,
+      "loss": 0.5729,
+      "step": 939
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047088162165673e-05,
+      "loss": 0.5088,
+      "step": 940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403491163243684e-05,
+      "loss": 0.5273,
+      "step": 941
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402272795264773e-05,
+      "loss": 0.6047,
+      "step": 942
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010537144388416e-05,
+      "loss": 0.5869,
+      "step": 943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399833922926161e-05,
+      "loss": 0.5757,
+      "step": 944
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986134228882607e-05,
+      "loss": 0.5671,
+      "step": 945
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973922164879263e-05,
+      "loss": 0.4423,
+      "step": 946
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961703058891955e-05,
+      "loss": 0.6312,
+      "step": 947
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949476932573531e-05,
+      "loss": 0.6056,
+      "step": 948
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937243807589291e-05,
+      "loss": 0.5232,
+      "step": 949
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925003705616917e-05,
+      "loss": 0.524,
+      "step": 950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912756648346477e-05,
+      "loss": 0.4864,
+      "step": 951
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900502657480352e-05,
+      "loss": 0.553,
+      "step": 952
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.388824175473321e-05,
+      "loss": 0.5696,
+      "step": 953
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875973961831965e-05,
+      "loss": 0.5393,
+      "step": 954
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863699300515754e-05,
+      "loss": 0.5701,
+      "step": 955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851417792535866e-05,
+      "loss": 0.5353,
+      "step": 956
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383912945965574e-05,
+      "loss": 0.647,
+      "step": 957
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.5723,
+      "step": 958
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814532406308922e-05,
+      "loss": 0.5151,
+      "step": 959
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380222372942941e-05,
+      "loss": 0.561,
+      "step": 960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789908314823932e-05,
+      "loss": 0.583,
+      "step": 961
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777586184316016e-05,
+      "loss": 0.5595,
+      "step": 962
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3765257359741065e-05,
+      "loss": 0.5149,
+      "step": 963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3752921862946364e-05,
+      "loss": 0.5862,
+      "step": 964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740579715791017e-05,
+      "loss": 0.5593,
+      "step": 965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728230940145911e-05,
+      "loss": 0.5631,
+      "step": 966
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371587555789367e-05,
+      "loss": 0.6061,
+      "step": 967
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703513590928647e-05,
+      "loss": 0.5657,
+      "step": 968
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691145061156843e-05,
+      "loss": 0.6639,
+      "step": 969
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678769990495899e-05,
+      "loss": 0.6061,
+      "step": 970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366638840087504e-05,
+      "loss": 0.6093,
+      "step": 971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365400031423505e-05,
+      "loss": 0.6102,
+      "step": 972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641605752528225e-05,
+      "loss": 0.5884,
+      "step": 973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629204737718328e-05,
+      "loss": 0.5254,
+      "step": 974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616797291780563e-05,
+      "loss": 0.5393,
+      "step": 975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604383436701536e-05,
+      "loss": 0.5531,
+      "step": 976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3591963194479198e-05,
+      "loss": 0.6008,
+      "step": 977
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579536587122828e-05,
+      "loss": 0.5299,
+      "step": 978
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567103636652976e-05,
+      "loss": 0.5843,
+      "step": 979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554664365101438e-05,
+      "loss": 0.5744,
+      "step": 980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542218794511212e-05,
+      "loss": 0.5031,
+      "step": 981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529766946936456e-05,
+      "loss": 0.542,
+      "step": 982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351730884444245e-05,
+      "loss": 0.5276,
+      "step": 983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504844509105562e-05,
+      "loss": 0.5635,
+      "step": 984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492373963013199e-05,
+      "loss": 0.4981,
+      "step": 985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479897228263781e-05,
+      "loss": 0.6569,
+      "step": 986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3467414326966685e-05,
+      "loss": 0.6171,
+      "step": 987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454925281242225e-05,
+      "loss": 0.4989,
+      "step": 988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442430113221602e-05,
+      "loss": 0.5932,
+      "step": 989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342992884504686e-05,
+      "loss": 0.5803,
+      "step": 990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417421498870854e-05,
+      "loss": 0.5895,
+      "step": 991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3404908096857216e-05,
+      "loss": 0.6557,
+      "step": 992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392388661180303e-05,
+      "loss": 0.5851,
+      "step": 993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3379863214025169e-05,
+      "loss": 0.6023,
+      "step": 994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367331777587509e-05,
+      "loss": 0.5347,
+      "step": 995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335479437407365e-05,
+      "loss": 0.6429,
+      "step": 996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342251025700474e-05,
+      "loss": 0.5198,
+      "step": 997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329701754695412e-05,
+      "loss": 0.5803,
+      "step": 998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317146583296385e-05,
+      "loss": 0.5882,
+      "step": 999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304585533751766e-05,
+      "loss": 0.5271,
+      "step": 1000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292018628320346e-05,
+      "loss": 0.5584,
+      "step": 1001
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32794458892713e-05,
+      "loss": 0.6215,
+      "step": 1002
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266867338884131e-05,
+      "loss": 0.5837,
+      "step": 1003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254282999448647e-05,
+      "loss": 0.6337,
+      "step": 1004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241692893264909e-05,
+      "loss": 0.558,
+      "step": 1005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32290970426432e-05,
+      "loss": 0.4728,
+      "step": 1006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216495469903983e-05,
+      "loss": 0.5645,
+      "step": 1007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203888197377857e-05,
+      "loss": 0.6095,
+      "step": 1008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191275247405525e-05,
+      "loss": 0.5947,
+      "step": 1009
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178656642337755e-05,
+      "loss": 0.5517,
+      "step": 1010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166032404535326e-05,
+      "loss": 0.712,
+      "step": 1011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153402556369001e-05,
+      "loss": 0.6198,
+      "step": 1012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314076712021949e-05,
+      "loss": 0.475,
+      "step": 1013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128126118477402e-05,
+      "loss": 0.5553,
+      "step": 1014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115479573543213e-05,
+      "loss": 0.5729,
+      "step": 1015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102827507827209e-05,
+      "loss": 0.5815,
+      "step": 1016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.5046,
+      "step": 1017
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3077506903739829e-05,
+      "loss": 0.5311,
+      "step": 1018
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064838410237799e-05,
+      "loss": 0.6729,
+      "step": 1019
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305216448569257e-05,
+      "loss": 0.5671,
+      "step": 1020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039485152562951e-05,
+      "loss": 0.6124,
+      "step": 1021
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026800433317348e-05,
+      "loss": 0.581,
+      "step": 1022
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30141103504337e-05,
+      "loss": 0.5247,
+      "step": 1023
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001414926399447e-05,
+      "loss": 0.5531,
+      "step": 1024
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988714183711504e-05,
+      "loss": 0.6003,
+      "step": 1025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2976008144876211e-05,
+      "loss": 0.5571,
+      "step": 1026
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296329683240928e-05,
+      "loss": 0.6207,
+      "step": 1027
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950580268835784e-05,
+      "loss": 0.6789,
+      "step": 1028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937858476690089e-05,
+      "loss": 0.5324,
+      "step": 1029
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925131478515833e-05,
+      "loss": 0.5551,
+      "step": 1030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291239929686588e-05,
+      "loss": 0.4953,
+      "step": 1031
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899661954302277e-05,
+      "loss": 0.5855,
+      "step": 1032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886919473396212e-05,
+      "loss": 0.6074,
+      "step": 1033
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2874171876727988e-05,
+      "loss": 0.5353,
+      "step": 1034
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861419186886963e-05,
+      "loss": 0.5915,
+      "step": 1035
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848661426471532e-05,
+      "loss": 0.5486,
+      "step": 1036
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2835898618089064e-05,
+      "loss": 0.5106,
+      "step": 1037
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2823130784355882e-05,
+      "loss": 0.5592,
+      "step": 1038
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810357947897205e-05,
+      "loss": 0.5262,
+      "step": 1039
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797580131347127e-05,
+      "loss": 0.5883,
+      "step": 1040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784797357348562e-05,
+      "loss": 0.5447,
+      "step": 1041
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772009648553208e-05,
+      "loss": 0.5459,
+      "step": 1042
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759217027621507e-05,
+      "loss": 0.5576,
+      "step": 1043
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274641951722261e-05,
+      "loss": 0.5609,
+      "step": 1044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733617140034329e-05,
+      "loss": 0.5318,
+      "step": 1045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2720809918743102e-05,
+      "loss": 0.6397,
+      "step": 1046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707997876043952e-05,
+      "loss": 0.5759,
+      "step": 1047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695181034640435e-05,
+      "loss": 0.5347,
+      "step": 1048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268235941724463e-05,
+      "loss": 0.5401,
+      "step": 1049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669533046577063e-05,
+      "loss": 0.5342,
+      "step": 1050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656701945366689e-05,
+      "loss": 0.5112,
+      "step": 1051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2643866136350847e-05,
+      "loss": 0.5222,
+      "step": 1052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631025642275212e-05,
+      "loss": 0.5332,
+      "step": 1053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618180485893775e-05,
+      "loss": 0.6541,
+      "step": 1054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605330689968771e-05,
+      "loss": 0.5035,
+      "step": 1055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592476277270671e-05,
+      "loss": 0.5342,
+      "step": 1056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257961727057812e-05,
+      "loss": 0.581,
+      "step": 1057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566753692677902e-05,
+      "loss": 0.5817,
+      "step": 1058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553885566364907e-05,
+      "loss": 0.5856,
+      "step": 1059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2541012914442088e-05,
+      "loss": 0.5536,
+      "step": 1060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528135759720403e-05,
+      "loss": 0.5319,
+      "step": 1061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515254125018803e-05,
+      "loss": 0.4743,
+      "step": 1062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502368033164176e-05,
+      "loss": 0.4541,
+      "step": 1063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24894775069913e-05,
+      "loss": 0.5831,
+      "step": 1064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476582569342819e-05,
+      "loss": 0.5395,
+      "step": 1065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2463683243069192e-05,
+      "loss": 0.542,
+      "step": 1066
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450779551028651e-05,
+      "loss": 0.5681,
+      "step": 1067
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437871516087174e-05,
+      "loss": 0.5889,
+      "step": 1068
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424959161118425e-05,
+      "loss": 0.5713,
+      "step": 1069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412042509003728e-05,
+      "loss": 0.5874,
+      "step": 1070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2399121582632018e-05,
+      "loss": 0.6203,
+      "step": 1071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386196404899808e-05,
+      "loss": 0.5156,
+      "step": 1072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373266998711152e-05,
+      "loss": 0.4884,
+      "step": 1073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360333386977574e-05,
+      "loss": 0.6521,
+      "step": 1074
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2347395592618075e-05,
+      "loss": 0.5497,
+      "step": 1075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334453638559057e-05,
+      "loss": 0.5648,
+      "step": 1076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232150754773429e-05,
+      "loss": 0.5857,
+      "step": 1077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2308557343084881e-05,
+      "loss": 0.5554,
+      "step": 1078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295603047559226e-05,
+      "loss": 0.5455,
+      "step": 1079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2282644684112964e-05,
+      "loss": 0.5518,
+      "step": 1080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269682275708951e-05,
+      "loss": 0.5088,
+      "step": 1081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225671584531721e-05,
+      "loss": 0.5572,
+      "step": 1082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2243745415914882e-05,
+      "loss": 0.536,
+      "step": 1083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2230771010486204e-05,
+      "loss": 0.5277,
+      "step": 1084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2217792652022452e-05,
+      "loss": 0.5343,
+      "step": 1085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204810363521919e-05,
+      "loss": 0.6137,
+      "step": 1086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191824167989845e-05,
+      "loss": 0.5242,
+      "step": 1087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178834088438404e-05,
+      "loss": 0.595,
+      "step": 1088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165840147886656e-05,
+      "loss": 0.5137,
+      "step": 1089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152842369360489e-05,
+      "loss": 0.6098,
+      "step": 1090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2139840775892606e-05,
+      "loss": 0.5694,
+      "step": 1091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126835390522466e-05,
+      "loss": 0.5768,
+      "step": 1092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113826236296245e-05,
+      "loss": 0.5193,
+      "step": 1093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21008133362668e-05,
+      "loss": 0.4776,
+      "step": 1094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087796713493618e-05,
+      "loss": 0.5436,
+      "step": 1095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074776391042797e-05,
+      "loss": 0.567,
+      "step": 1096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061752391986982e-05,
+      "loss": 0.5557,
+      "step": 1097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048724739405337e-05,
+      "loss": 0.5333,
+      "step": 1098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2035693456383493e-05,
+      "loss": 0.551,
+      "step": 1099
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202265856601352e-05,
+      "loss": 0.5993,
+      "step": 1100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009620091393885e-05,
+      "loss": 0.5892,
+      "step": 1101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1996578055629395e-05,
+      "loss": 0.5147,
+      "step": 1102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1983532481831179e-05,
+      "loss": 0.4809,
+      "step": 1103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970483393116626e-05,
+      "loss": 0.5957,
+      "step": 1104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957430812609361e-05,
+      "loss": 0.5448,
+      "step": 1105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944374763439189e-05,
+      "loss": 0.5733,
+      "step": 1106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1931315268742075e-05,
+      "loss": 0.5776,
+      "step": 1107
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1918252351660066e-05,
+      "loss": 0.5024,
+      "step": 1108
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905186035341304e-05,
+      "loss": 0.5114,
+      "step": 1109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189211634293993e-05,
+      "loss": 0.5298,
+      "step": 1110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187904329761608e-05,
+      "loss": 0.5016,
+      "step": 1111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865966922535826e-05,
+      "loss": 0.4817,
+      "step": 1112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852887240871145e-05,
+      "loss": 0.4268,
+      "step": 1113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183980427579987e-05,
+      "loss": 0.5752,
+      "step": 1114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1826718050505653e-05,
+      "loss": 0.4588,
+      "step": 1115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813628588177923e-05,
+      "loss": 0.4916,
+      "step": 1116
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1800535912011846e-05,
+      "loss": 0.655,
+      "step": 1117
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1787440045208287e-05,
+      "loss": 0.5271,
+      "step": 1118
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1774341010973753e-05,
+      "loss": 0.5527,
+      "step": 1119
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761238832520384e-05,
+      "loss": 0.5261,
+      "step": 1120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748133533065864e-05,
+      "loss": 0.6096,
+      "step": 1121
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1735025135833436e-05,
+      "loss": 0.5587,
+      "step": 1122
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721913664051814e-05,
+      "loss": 0.5811,
+      "step": 1123
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708799140955165e-05,
+      "loss": 0.5232,
+      "step": 1124
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695681589783065e-05,
+      "loss": 0.6073,
+      "step": 1125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1682561033780457e-05,
+      "loss": 0.4646,
+      "step": 1126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16694374961976e-05,
+      "loss": 0.6386,
+      "step": 1127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165631100029005e-05,
+      "loss": 0.4819,
+      "step": 1128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643181569318596e-05,
+      "loss": 0.5645,
+      "step": 1129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1630049226549227e-05,
+      "loss": 0.5028,
+      "step": 1130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16169139952531e-05,
+      "loss": 0.6587,
+      "step": 1131
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603775898706479e-05,
+      "loss": 0.5249,
+      "step": 1132
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590634960190722e-05,
+      "loss": 0.4885,
+      "step": 1133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577491202992204e-05,
+      "loss": 0.5353,
+      "step": 1134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.5796,
+      "step": 1135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155119532571737e-05,
+      "loss": 0.5553,
+      "step": 1136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1538043252238629e-05,
+      "loss": 0.5264,
+      "step": 1137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15248884532722e-05,
+      "loss": 0.5136,
+      "step": 1138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511730952129037e-05,
+      "loss": 0.5225,
+      "step": 1139
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498570772124863e-05,
+      "loss": 0.5899,
+      "step": 1140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485407936580169e-05,
+      "loss": 0.5824,
+      "step": 1141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1472242468820136e-05,
+      "loss": 0.5478,
+      "step": 1142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459074392174619e-05,
+      "loss": 0.546,
+      "step": 1143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1445903729978088e-05,
+      "loss": 0.5521,
+      "step": 1144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432730505569597e-05,
+      "loss": 0.3808,
+      "step": 1145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419554742292753e-05,
+      "loss": 0.5555,
+      "step": 1146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406376463495643e-05,
+      "loss": 0.5659,
+      "step": 1147
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393195692530823e-05,
+      "loss": 0.5034,
+      "step": 1148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380012452755259e-05,
+      "loss": 0.534,
+      "step": 1149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366826767530296e-05,
+      "loss": 0.592,
+      "step": 1150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353638660221616e-05,
+      "loss": 0.5388,
+      "step": 1151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1340448154199186e-05,
+      "loss": 0.5289,
+      "step": 1152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327255272837221e-05,
+      "loss": 0.5205,
+      "step": 1153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131406003951416e-05,
+      "loss": 0.5794,
+      "step": 1154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130086247761259e-05,
+      "loss": 0.5576,
+      "step": 1155
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1287662610519244e-05,
+      "loss": 0.549,
+      "step": 1156
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274460461624925e-05,
+      "loss": 0.4726,
+      "step": 1157
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261256054324488e-05,
+      "loss": 0.5337,
+      "step": 1158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1248049412016782e-05,
+      "loss": 0.5183,
+      "step": 1159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234840558104628e-05,
+      "loss": 0.615,
+      "step": 1160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221629515994754e-05,
+      "loss": 0.5413,
+      "step": 1161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208416309097775e-05,
+      "loss": 0.5226,
+      "step": 1162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1195200960828138e-05,
+      "loss": 0.6069,
+      "step": 1163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1181983494604082e-05,
+      "loss": 0.578,
+      "step": 1164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168763933847608e-05,
+      "loss": 0.5077,
+      "step": 1165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155542301984415e-05,
+      "loss": 0.5418,
+      "step": 1166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142318622443883e-05,
+      "loss": 0.4442,
+      "step": 1167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129092918659019e-05,
+      "loss": 0.5971,
+      "step": 1168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115865214066414e-05,
+      "loss": 0.5145,
+      "step": 1169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102635532106204e-05,
+      "loss": 0.5157,
+      "step": 1170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108940389622204e-05,
+      "loss": 0.5566,
+      "step": 1171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1076170329861012e-05,
+      "loss": 0.5325,
+      "step": 1172
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062934856473655e-05,
+      "loss": 0.5184,
+      "step": 1173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049697499513871e-05,
+      "loss": 0.5819,
+      "step": 1174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036458282438905e-05,
+      "loss": 0.5604,
+      "step": 1175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102321722870929e-05,
+      "loss": 0.5077,
+      "step": 1176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009974361788822e-05,
+      "loss": 0.5139,
+      "step": 1177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099672970514451e-05,
+      "loss": 0.5208,
+      "step": 1178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983483282246527e-05,
+      "loss": 0.5535,
+      "step": 1179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0970235116568186e-05,
+      "loss": 0.5893,
+      "step": 1180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095698523158588e-05,
+      "loss": 0.5307,
+      "step": 1181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094373365077905e-05,
+      "loss": 0.4983,
+      "step": 1182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930480397630146e-05,
+      "loss": 0.445,
+      "step": 1183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917225495624581e-05,
+      "loss": 0.4914,
+      "step": 1184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903968968250682e-05,
+      "loss": 0.4853,
+      "step": 1185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0890710838999671e-05,
+      "loss": 0.5507,
+      "step": 1186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087745113136559e-05,
+      "loss": 0.5086,
+      "step": 1187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864189868845296e-05,
+      "loss": 0.7193,
+      "step": 1188
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085092707493839e-05,
+      "loss": 0.6319,
+      "step": 1189
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837662773147189e-05,
+      "loss": 0.5104,
+      "step": 1190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824396986976681e-05,
+      "loss": 0.5457,
+      "step": 1191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811129739934494e-05,
+      "loss": 0.5824,
+      "step": 1192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797861055530832e-05,
+      "loss": 0.5203,
+      "step": 1193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784590957278452e-05,
+      "loss": 0.6037,
+      "step": 1194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771319468692613e-05,
+      "loss": 0.4794,
+      "step": 1195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0758046613291043e-05,
+      "loss": 0.6196,
+      "step": 1196
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744772414593889e-05,
+      "loss": 0.5942,
+      "step": 1197
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731496896123676e-05,
+      "loss": 0.4483,
+      "step": 1198
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0718220081405277e-05,
+      "loss": 0.6162,
+      "step": 1199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704941993965849e-05,
+      "loss": 0.4529,
+      "step": 1200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0691662657334815e-05,
+      "loss": 0.6396,
+      "step": 1201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678382095043807e-05,
+      "loss": 0.4539,
+      "step": 1202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665100330626625e-05,
+      "loss": 0.6123,
+      "step": 1203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651817387619206e-05,
+      "loss": 0.53,
+      "step": 1204
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638533289559574e-05,
+      "loss": 0.5906,
+      "step": 1205
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06252480599878e-05,
+      "loss": 0.5797,
+      "step": 1206
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611961722445955e-05,
+      "loss": 0.6582,
+      "step": 1207
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598674300478085e-05,
+      "loss": 0.488,
+      "step": 1208
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0585385817630137e-05,
+      "loss": 0.5822,
+      "step": 1209
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572096297449967e-05,
+      "loss": 0.4932,
+      "step": 1210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558805763487242e-05,
+      "loss": 0.5382,
+      "step": 1211
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545514239293437e-05,
+      "loss": 0.5278,
+      "step": 1212
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532221748421786e-05,
+      "loss": 0.6017,
+      "step": 1213
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518928314427233e-05,
+      "loss": 0.5503,
+      "step": 1214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505633960866384e-05,
+      "loss": 0.5056,
+      "step": 1215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492338711297488e-05,
+      "loss": 0.5756,
+      "step": 1216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047904258928037e-05,
+      "loss": 0.5415,
+      "step": 1217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465745618376417e-05,
+      "loss": 0.5843,
+      "step": 1218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452447822148499e-05,
+      "loss": 0.5269,
+      "step": 1219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043914922416097e-05,
+      "loss": 0.6205,
+      "step": 1220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425849847979586e-05,
+      "loss": 0.6223,
+      "step": 1221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412549717171497e-05,
+      "loss": 0.5524,
+      "step": 1222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399248855305178e-05,
+      "loss": 0.5667,
+      "step": 1223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385947285950407e-05,
+      "loss": 0.5062,
+      "step": 1224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372645032678215e-05,
+      "loss": 0.5965,
+      "step": 1225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359342119060844e-05,
+      "loss": 0.4743,
+      "step": 1226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346038568671708e-05,
+      "loss": 0.6031,
+      "step": 1227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332734405085343e-05,
+      "loss": 0.5637,
+      "step": 1228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031942965187738e-05,
+      "loss": 0.5094,
+      "step": 1229
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0306124332624484e-05,
+      "loss": 0.667,
+      "step": 1230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029281847090434e-05,
+      "loss": 0.5506,
+      "step": 1231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279512090295574e-05,
+      "loss": 0.4839,
+      "step": 1232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026620521437775e-05,
+      "loss": 0.5498,
+      "step": 1233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252897866731295e-05,
+      "loss": 0.6048,
+      "step": 1234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239590070937483e-05,
+      "loss": 0.4795,
+      "step": 1235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226281850578377e-05,
+      "loss": 0.5554,
+      "step": 1236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0212973229236787e-05,
+      "loss": 0.5215,
+      "step": 1237
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199664230496247e-05,
+      "loss": 0.5171,
+      "step": 1238
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186354877940948e-05,
+      "loss": 0.5311,
+      "step": 1239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173045195155712e-05,
+      "loss": 0.495,
+      "step": 1240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159735205725949e-05,
+      "loss": 0.5537,
+      "step": 1241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146424933237608e-05,
+      "loss": 0.5556,
+      "step": 1242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013311440127714e-05,
+      "loss": 0.5388,
+      "step": 1243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119803633431459e-05,
+      "loss": 0.5613,
+      "step": 1244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0106492653287893e-05,
+      "loss": 0.533,
+      "step": 1245
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093181484434151e-05,
+      "loss": 0.5268,
+      "step": 1246
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079870150458274e-05,
+      "loss": 0.6224,
+      "step": 1247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006655867494859e-05,
+      "loss": 0.5594,
+      "step": 1248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053247081493684e-05,
+      "loss": 0.4897,
+      "step": 1249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039935393682358e-05,
+      "loss": 0.4742,
+      "step": 1250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026623635103563e-05,
+      "loss": 0.5325,
+      "step": 1251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013311829346389e-05,
+      "loss": 0.6013,
+      "step": 1252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.5623,
+      "step": 1253
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986688170653616e-06,
+      "loss": 0.4787,
+      "step": 1254
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973376364896438e-06,
+      "loss": 0.63,
+      "step": 1255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960064606317647e-06,
+      "loss": 0.4757,
+      "step": 1256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946752918506319e-06,
+      "loss": 0.5611,
+      "step": 1257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933441325051414e-06,
+      "loss": 0.5185,
+      "step": 1258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920129849541731e-06,
+      "loss": 0.508,
+      "step": 1259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906818515565849e-06,
+      "loss": 0.5856,
+      "step": 1260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893507346712112e-06,
+      "loss": 0.5983,
+      "step": 1261
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880196366568546e-06,
+      "loss": 0.4987,
+      "step": 1262
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866885598722865e-06,
+      "loss": 0.6237,
+      "step": 1263
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853575066762395e-06,
+      "loss": 0.5622,
+      "step": 1264
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.840264794274053e-06,
+      "loss": 0.543,
+      "step": 1265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826954804844288e-06,
+      "loss": 0.5344,
+      "step": 1266
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813645122059054e-06,
+      "loss": 0.5382,
+      "step": 1267
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800335769503756e-06,
+      "loss": 0.568,
+      "step": 1268
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787026770763216e-06,
+      "loss": 0.4701,
+      "step": 1269
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773718149421627e-06,
+      "loss": 0.4735,
+      "step": 1270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760409929062518e-06,
+      "loss": 0.5715,
+      "step": 1271
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747102133268709e-06,
+      "loss": 0.5006,
+      "step": 1272
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733794785622254e-06,
+      "loss": 0.5082,
+      "step": 1273
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72048790970443e-06,
+      "loss": 0.5209,
+      "step": 1274
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707181529095663e-06,
+      "loss": 0.7047,
+      "step": 1275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.693875667375518e-06,
+      "loss": 0.5197,
+      "step": 1276
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680570348122626e-06,
+      "loss": 0.6685,
+      "step": 1277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667265594914662e-06,
+      "loss": 0.6546,
+      "step": 1278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653961431328295e-06,
+      "loss": 0.5609,
+      "step": 1279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640657880939157e-06,
+      "loss": 0.6543,
+      "step": 1280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627354967321785e-06,
+      "loss": 0.5296,
+      "step": 1281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614052714049597e-06,
+      "loss": 0.54,
+      "step": 1282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600751144694827e-06,
+      "loss": 0.5812,
+      "step": 1283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587450282828508e-06,
+      "loss": 0.547,
+      "step": 1284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574150152020415e-06,
+      "loss": 0.5773,
+      "step": 1285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.560850775839034e-06,
+      "loss": 0.5702,
+      "step": 1286
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.5475521778515e-06,
+      "loss": 0.5487,
+      "step": 1287
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534254381623588e-06,
+      "loss": 0.5106,
+      "step": 1288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.520957410719632e-06,
+      "loss": 0.4711,
+      "step": 1289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507661288702515e-06,
+      "loss": 0.538,
+      "step": 1290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494366039133619e-06,
+      "loss": 0.6421,
+      "step": 1291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481071685572769e-06,
+      "loss": 0.5767,
+      "step": 1292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467778251578217e-06,
+      "loss": 0.5717,
+      "step": 1293
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454485760706564e-06,
+      "loss": 0.5049,
+      "step": 1294
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441194236512763e-06,
+      "loss": 0.6325,
+      "step": 1295
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427903702550034e-06,
+      "loss": 0.5534,
+      "step": 1296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414614182369862e-06,
+      "loss": 0.6109,
+      "step": 1297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401325699521922e-06,
+      "loss": 0.5426,
+      "step": 1298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388038277554046e-06,
+      "loss": 0.5317,
+      "step": 1299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374751940012203e-06,
+      "loss": 0.6109,
+      "step": 1300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361466710440428e-06,
+      "loss": 0.5378,
+      "step": 1301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.348182612380796e-06,
+      "loss": 0.5346,
+      "step": 1302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334899669373379e-06,
+      "loss": 0.4505,
+      "step": 1303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3216179049562e-06,
+      "loss": 0.6156,
+      "step": 1304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.308337342665188e-06,
+      "loss": 0.5438,
+      "step": 1305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295058006034153e-06,
+      "loss": 0.5866,
+      "step": 1306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.281779918594723e-06,
+      "loss": 0.554,
+      "step": 1307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268503103876324e-06,
+      "loss": 0.5989,
+      "step": 1308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255227585406116e-06,
+      "loss": 0.6214,
+      "step": 1309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.241953386708962e-06,
+      "loss": 0.5744,
+      "step": 1310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22868053130739e-06,
+      "loss": 0.5998,
+      "step": 1311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215409042721553e-06,
+      "loss": 0.4885,
+      "step": 1312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202138944469168e-06,
+      "loss": 0.5322,
+      "step": 1313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188870260065507e-06,
+      "loss": 0.625,
+      "step": 1314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17560301302332e-06,
+      "loss": 0.5804,
+      "step": 1315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162337226852813e-06,
+      "loss": 0.508,
+      "step": 1316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149072925061614e-06,
+      "loss": 0.4975,
+      "step": 1317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.135810131154707e-06,
+      "loss": 0.5131,
+      "step": 1318
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122548868634416e-06,
+      "loss": 0.5623,
+      "step": 1319
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109289161000334e-06,
+      "loss": 0.5186,
+      "step": 1320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096031031749321e-06,
+      "loss": 0.5396,
+      "step": 1321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082774504375422e-06,
+      "loss": 0.6223,
+      "step": 1322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069519602369856e-06,
+      "loss": 0.5696,
+      "step": 1323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056266349220951e-06,
+      "loss": 0.6405,
+      "step": 1324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043014768414125e-06,
+      "loss": 0.4751,
+      "step": 1325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.029764883431818e-06,
+      "loss": 0.5887,
+      "step": 1326
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.016516717753474e-06,
+      "loss": 0.5308,
+      "step": 1327
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003270294855491e-06,
+      "loss": 0.6018,
+      "step": 1328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99002563821118e-06,
+      "loss": 0.4905,
+      "step": 1329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976782771290715e-06,
+      "loss": 0.5102,
+      "step": 1330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9635417175611e-06,
+      "loss": 0.5821,
+      "step": 1331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95030250048613e-06,
+      "loss": 0.5619,
+      "step": 1332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937065143526349e-06,
+      "loss": 0.5037,
+      "step": 1333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92382967013899e-06,
+      "loss": 0.4878,
+      "step": 1334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910596103777965e-06,
+      "loss": 0.5217,
+      "step": 1335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897364467893797e-06,
+      "loss": 0.479,
+      "step": 1336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884134785933588e-06,
+      "loss": 0.4989,
+      "step": 1337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870907081340983e-06,
+      "loss": 0.5119,
+      "step": 1338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857681377556117e-06,
+      "loss": 0.6321,
+      "step": 1339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844457698015588e-06,
+      "loss": 0.4308,
+      "step": 1340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831236066152397e-06,
+      "loss": 0.5542,
+      "step": 1341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818016505395921e-06,
+      "loss": 0.585,
+      "step": 1342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.804799039171863e-06,
+      "loss": 0.5585,
+      "step": 1343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791583690902226e-06,
+      "loss": 0.5806,
+      "step": 1344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778370484005245e-06,
+      "loss": 0.4797,
+      "step": 1345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765159441895376e-06,
+      "loss": 0.5148,
+      "step": 1346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751950587983221e-06,
+      "loss": 0.5539,
+      "step": 1347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738743945675515e-06,
+      "loss": 0.5215,
+      "step": 1348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725539538375078e-06,
+      "loss": 0.5428,
+      "step": 1349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712337389480758e-06,
+      "loss": 0.4676,
+      "step": 1350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.699137522387415e-06,
+      "loss": 0.4829,
+      "step": 1351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685939960485846e-06,
+      "loss": 0.5762,
+      "step": 1352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672744727162782e-06,
+      "loss": 0.5142,
+      "step": 1353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.659551845800818e-06,
+      "loss": 0.5234,
+      "step": 1354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646361339778386e-06,
+      "loss": 0.5152,
+      "step": 1355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633173232469707e-06,
+      "loss": 0.5449,
+      "step": 1356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619987547244746e-06,
+      "loss": 0.5545,
+      "step": 1357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606804307469182e-06,
+      "loss": 0.5465,
+      "step": 1358
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59362353650436e-06,
+      "loss": 0.6104,
+      "step": 1359
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580445257707246e-06,
+      "loss": 0.4816,
+      "step": 1360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567269494430404e-06,
+      "loss": 0.6053,
+      "step": 1361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554096270021917e-06,
+      "loss": 0.6355,
+      "step": 1362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540925607825385e-06,
+      "loss": 0.4961,
+      "step": 1363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.527757531179866e-06,
+      "loss": 0.5935,
+      "step": 1364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514592063419833e-06,
+      "loss": 0.5945,
+      "step": 1365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501429227875137e-06,
+      "loss": 0.5842,
+      "step": 1366
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488269047870968e-06,
+      "loss": 0.5052,
+      "step": 1367
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.475111546727802e-06,
+      "loss": 0.5677,
+      "step": 1368
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461956747761375e-06,
+      "loss": 0.5367,
+      "step": 1369
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448804674282633e-06,
+      "loss": 0.6271,
+      "step": 1370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.6137,
+      "step": 1371
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422508797007798e-06,
+      "loss": 0.6369,
+      "step": 1372
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.409365039809282e-06,
+      "loss": 0.5954,
+      "step": 1373
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396224101293523e-06,
+      "loss": 0.5645,
+      "step": 1374
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.383086004746903e-06,
+      "loss": 0.5453,
+      "step": 1375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.369950773450774e-06,
+      "loss": 0.5044,
+      "step": 1376
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356818430681409e-06,
+      "loss": 0.6318,
+      "step": 1377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343688999709953e-06,
+      "loss": 0.533,
+      "step": 1378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330562503802402e-06,
+      "loss": 0.6221,
+      "step": 1379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317438966219546e-06,
+      "loss": 0.4671,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304318410216937e-06,
+      "loss": 0.5248,
+      "step": 1381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291200859044836e-06,
+      "loss": 0.5349,
+      "step": 1382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.278086335948191e-06,
+      "loss": 0.5211,
+      "step": 1383
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.264974864166566e-06,
+      "loss": 0.5882,
+      "step": 1384
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.251866466934137e-06,
+      "loss": 0.6037,
+      "step": 1385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23876116747962e-06,
+      "loss": 0.6253,
+      "step": 1386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225658989026245e-06,
+      "loss": 0.6069,
+      "step": 1387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.212559954791718e-06,
+      "loss": 0.5798,
+      "step": 1388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.199464087988158e-06,
+      "loss": 0.5389,
+      "step": 1389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18637141182208e-06,
+      "loss": 0.5249,
+      "step": 1390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.173281949494352e-06,
+      "loss": 0.5362,
+      "step": 1391
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.160195724200132e-06,
+      "loss": 0.5533,
+      "step": 1392
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147112759128859e-06,
+      "loss": 0.5616,
+      "step": 1393
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134033077464177e-06,
+      "loss": 0.4979,
+      "step": 1394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120956702383922e-06,
+      "loss": 0.4791,
+      "step": 1395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.107883657060072e-06,
+      "loss": 0.4978,
+      "step": 1396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094813964658698e-06,
+      "loss": 0.5612,
+      "step": 1397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.081747648339939e-06,
+      "loss": 0.7394,
+      "step": 1398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.068684731257932e-06,
+      "loss": 0.5848,
+      "step": 1399
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055625236560813e-06,
+      "loss": 0.5882,
+      "step": 1400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.042569187390642e-06,
+      "loss": 0.5685,
+      "step": 1401
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029516606883376e-06,
+      "loss": 0.5368,
+      "step": 1402
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01646751816882e-06,
+      "loss": 0.5684,
+      "step": 1403
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.003421944370607e-06,
+      "loss": 0.4981,
+      "step": 1404
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.990379908606118e-06,
+      "loss": 0.5292,
+      "step": 1405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977341433986481e-06,
+      "loss": 0.7444,
+      "step": 1406
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964306543616509e-06,
+      "loss": 0.4861,
+      "step": 1407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951275260594666e-06,
+      "loss": 0.562,
+      "step": 1408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.938247608013021e-06,
+      "loss": 0.5561,
+      "step": 1409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925223608957205e-06,
+      "loss": 0.4738,
+      "step": 1410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912203286506386e-06,
+      "loss": 0.6379,
+      "step": 1411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899186663733204e-06,
+      "loss": 0.5213,
+      "step": 1412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886173763703757e-06,
+      "loss": 0.5223,
+      "step": 1413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873164609477537e-06,
+      "loss": 0.588,
+      "step": 1414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.860159224107397e-06,
+      "loss": 0.5685,
+      "step": 1415
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.847157630639513e-06,
+      "loss": 0.5309,
+      "step": 1416
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834159852113347e-06,
+      "loss": 0.624,
+      "step": 1417
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.821165911561596e-06,
+      "loss": 0.5482,
+      "step": 1418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808175832010158e-06,
+      "loss": 0.4744,
+      "step": 1419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795189636478084e-06,
+      "loss": 0.5059,
+      "step": 1420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78220734797755e-06,
+      "loss": 0.528,
+      "step": 1421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.7692289895138e-06,
+      "loss": 0.5184,
+      "step": 1422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756254584085121e-06,
+      "loss": 0.4594,
+      "step": 1423
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.743284154682792e-06,
+      "loss": 0.4871,
+      "step": 1424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73031772429105e-06,
+      "loss": 0.5061,
+      "step": 1425
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71735531588704e-06,
+      "loss": 0.5723,
+      "step": 1426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704396952440778e-06,
+      "loss": 0.5852,
+      "step": 1427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.691442656915122e-06,
+      "loss": 0.5698,
+      "step": 1428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.678492452265713e-06,
+      "loss": 0.6188,
+      "step": 1429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66554636144095e-06,
+      "loss": 0.5528,
+      "step": 1430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.652604407381927e-06,
+      "loss": 0.4962,
+      "step": 1431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63966661302243e-06,
+      "loss": 0.5354,
+      "step": 1432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626733001288852e-06,
+      "loss": 0.6037,
+      "step": 1433
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613803595100191e-06,
+      "loss": 0.5642,
+      "step": 1434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600878417367986e-06,
+      "loss": 0.5552,
+      "step": 1435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.587957490996276e-06,
+      "loss": 0.5917,
+      "step": 1436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575040838881578e-06,
+      "loss": 0.4955,
+      "step": 1437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56212848391283e-06,
+      "loss": 0.4655,
+      "step": 1438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5492204489713496e-06,
+      "loss": 0.5805,
+      "step": 1439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.536316756930811e-06,
+      "loss": 0.5269,
+      "step": 1440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523417430657186e-06,
+      "loss": 0.5041,
+      "step": 1441
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510522493008703e-06,
+      "loss": 0.557,
+      "step": 1442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497631966835828e-06,
+      "loss": 0.5283,
+      "step": 1443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484745874981196e-06,
+      "loss": 0.5773,
+      "step": 1444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471864240279598e-06,
+      "loss": 0.5015,
+      "step": 1445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.458987085557916e-06,
+      "loss": 0.6004,
+      "step": 1446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.446114433635094e-06,
+      "loss": 0.6328,
+      "step": 1447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.433246307322099e-06,
+      "loss": 0.4885,
+      "step": 1448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.420382729421883e-06,
+      "loss": 0.4946,
+      "step": 1449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4075237227293285e-06,
+      "loss": 0.5858,
+      "step": 1450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3946693100312305e-06,
+      "loss": 0.5818,
+      "step": 1451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3818195141062286e-06,
+      "loss": 0.5666,
+      "step": 1452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368974357724789e-06,
+      "loss": 0.51,
+      "step": 1453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.356133863649155e-06,
+      "loss": 0.6107,
+      "step": 1454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343298054633314e-06,
+      "loss": 0.5256,
+      "step": 1455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3304669534229424e-06,
+      "loss": 0.6266,
+      "step": 1456
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317640582755373e-06,
+      "loss": 0.4227,
+      "step": 1457
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304818965359567e-06,
+      "loss": 0.5468,
+      "step": 1458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292002123956052e-06,
+      "loss": 0.547,
+      "step": 1459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2791900812569e-06,
+      "loss": 0.6453,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266382859965673e-06,
+      "loss": 0.5326,
+      "step": 1461
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.253580482777395e-06,
+      "loss": 0.4997,
+      "step": 1462
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2407829723784965e-06,
+      "loss": 0.5204,
+      "step": 1463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.227990351446797e-06,
+      "loss": 0.4646,
+      "step": 1464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2152026426514395e-06,
+      "loss": 0.4634,
+      "step": 1465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202419868652875e-06,
+      "loss": 0.5461,
+      "step": 1466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189642052102799e-06,
+      "loss": 0.5295,
+      "step": 1467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1768692156441225e-06,
+      "loss": 0.5827,
+      "step": 1468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.164101381910939e-06,
+      "loss": 0.505,
+      "step": 1469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151338573528471e-06,
+      "loss": 0.5948,
+      "step": 1470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.138580813113038e-06,
+      "loss": 0.5359,
+      "step": 1471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.125828123272016e-06,
+      "loss": 0.5049,
+      "step": 1472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113080526603793e-06,
+      "loss": 0.5161,
+      "step": 1473
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100338045697727e-06,
+      "loss": 0.5361,
+      "step": 1474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.087600703134123e-06,
+      "loss": 0.575,
+      "step": 1475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074868521484167e-06,
+      "loss": 0.5481,
+      "step": 1476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.062141523309918e-06,
+      "loss": 0.5742,
+      "step": 1477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049419731164221e-06,
+      "loss": 0.6022,
+      "step": 1478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036703167590724e-06,
+      "loss": 0.5573,
+      "step": 1479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.023991855123793e-06,
+      "loss": 0.4864,
+      "step": 1480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011285816288496e-06,
+      "loss": 0.5354,
+      "step": 1481
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998585073600552e-06,
+      "loss": 0.5219,
+      "step": 1482
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9858896495663046e-06,
+      "loss": 0.5856,
+      "step": 1483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973199566682653e-06,
+      "loss": 0.4696,
+      "step": 1484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.96051484743705e-06,
+      "loss": 0.5239,
+      "step": 1485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947835514307433e-06,
+      "loss": 0.5953,
+      "step": 1486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935161589762204e-06,
+      "loss": 0.4928,
+      "step": 1487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922493096260174e-06,
+      "loss": 0.5384,
+      "step": 1488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.5605,
+      "step": 1489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897172492172793e-06,
+      "loss": 0.5389,
+      "step": 1490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.88452042645679e-06,
+      "loss": 0.4586,
+      "step": 1491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8718738815226e-06,
+      "loss": 0.5247,
+      "step": 1492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859232879780515e-06,
+      "loss": 0.5356,
+      "step": 1493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846597443631005e-06,
+      "loss": 0.5678,
+      "step": 1494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833967595464679e-06,
+      "loss": 0.5726,
+      "step": 1495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821343357662249e-06,
+      "loss": 0.5192,
+      "step": 1496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8087247525944745e-06,
+      "loss": 0.5449,
+      "step": 1497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796111802622148e-06,
+      "loss": 0.5564,
+      "step": 1498
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783504530096023e-06,
+      "loss": 0.562,
+      "step": 1499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770902957356802e-06,
+      "loss": 0.5878,
+      "step": 1500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758307106735094e-06,
+      "loss": 0.4945,
+      "step": 1501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745717000551356e-06,
+      "loss": 0.6107,
+      "step": 1502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73313266111587e-06,
+      "loss": 0.5483,
+      "step": 1503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720554110728703e-06,
+      "loss": 0.602,
+      "step": 1504
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707981371679657e-06,
+      "loss": 0.5029,
+      "step": 1505
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6954144662482375e-06,
+      "loss": 0.6,
+      "step": 1506
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682853416703619e-06,
+      "loss": 0.5179,
+      "step": 1507
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670298245304587e-06,
+      "loss": 0.5245,
+      "step": 1508
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657748974299529e-06,
+      "loss": 0.5461,
+      "step": 1509
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645205625926354e-06,
+      "loss": 0.564,
+      "step": 1510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6326682224124925e-06,
+      "loss": 0.4821,
+      "step": 1511
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.620136785974834e-06,
+      "loss": 0.5846,
+      "step": 1512
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.607611338819697e-06,
+      "loss": 0.512,
+      "step": 1513
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5950919031427874e-06,
+      "loss": 0.4794,
+      "step": 1514
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582578501129147e-06,
+      "loss": 0.4321,
+      "step": 1515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5700711549531435e-06,
+      "loss": 0.4924,
+      "step": 1516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557569886778401e-06,
+      "loss": 0.5929,
+      "step": 1517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5450747187577745e-06,
+      "loss": 0.5654,
+      "step": 1518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.532585673033317e-06,
+      "loss": 0.5074,
+      "step": 1519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520102771736225e-06,
+      "loss": 0.5781,
+      "step": 1520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.507626036986804e-06,
+      "loss": 0.5409,
+      "step": 1521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495155490894442e-06,
+      "loss": 0.433,
+      "step": 1522
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48269115555755e-06,
+      "loss": 0.5725,
+      "step": 1523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.470233053063546e-06,
+      "loss": 0.5355,
+      "step": 1524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.457781205488791e-06,
+      "loss": 0.5495,
+      "step": 1525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445335634898567e-06,
+      "loss": 0.5961,
+      "step": 1526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432896363347028e-06,
+      "loss": 0.5973,
+      "step": 1527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420463412877176e-06,
+      "loss": 0.5531,
+      "step": 1528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.408036805520801e-06,
+      "loss": 0.4906,
+      "step": 1529
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395616563298466e-06,
+      "loss": 0.527,
+      "step": 1530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3832027082194385e-06,
+      "loss": 0.5394,
+      "step": 1531
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370795262281675e-06,
+      "loss": 0.6113,
+      "step": 1532
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358394247471779e-06,
+      "loss": 0.4755,
+      "step": 1533
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3459996857649516e-06,
+      "loss": 0.4681,
+      "step": 1534
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333611599124966e-06,
+      "loss": 0.573,
+      "step": 1535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321230009504107e-06,
+      "loss": 0.5013,
+      "step": 1536
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.308854938843161e-06,
+      "loss": 0.6231,
+      "step": 1537
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296486409071354e-06,
+      "loss": 0.5352,
+      "step": 1538
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28412444210633e-06,
+      "loss": 0.5859,
+      "step": 1539
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271769059854092e-06,
+      "loss": 0.453,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259420284208987e-06,
+      "loss": 0.5412,
+      "step": 1541
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.247078137053637e-06,
+      "loss": 0.533,
+      "step": 1542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234742640258938e-06,
+      "loss": 0.6573,
+      "step": 1543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222413815683987e-06,
+      "loss": 0.5209,
+      "step": 1544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.210091685176067e-06,
+      "loss": 0.5536,
+      "step": 1545
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197776270570594e-06,
+      "loss": 0.5907,
+      "step": 1546
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185467593691081e-06,
+      "loss": 0.6496,
+      "step": 1547
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.571,
+      "step": 1548
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160870540344261e-06,
+      "loss": 0.575,
+      "step": 1549
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.148582207464134e-06,
+      "loss": 0.4739,
+      "step": 1550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13630069948425e-06,
+      "loss": 0.5767,
+      "step": 1551
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124026038168039e-06,
+      "loss": 0.4838,
+      "step": 1552
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.111758245266795e-06,
+      "loss": 0.532,
+      "step": 1553
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.099497342519651e-06,
+      "loss": 0.5599,
+      "step": 1554
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0872433516535225e-06,
+      "loss": 0.5179,
+      "step": 1555
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0749962943830865e-06,
+      "loss": 0.5736,
+      "step": 1556
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0627561924107145e-06,
+      "loss": 0.4923,
+      "step": 1557
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05052306742647e-06,
+      "loss": 0.5896,
+      "step": 1558
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038296941108046e-06,
+      "loss": 0.4628,
+      "step": 1559
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0260778351207386e-06,
+      "loss": 0.5315,
+      "step": 1560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.013865771117394e-06,
+      "loss": 0.584,
+      "step": 1561
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001660770738394e-06,
+      "loss": 0.5021,
+      "step": 1562
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989462855611585e-06,
+      "loss": 0.5782,
+      "step": 1563
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977272047352274e-06,
+      "loss": 0.3732,
+      "step": 1564
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965088367563162e-06,
+      "loss": 0.5936,
+      "step": 1565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952911837834332e-06,
+      "loss": 0.5809,
+      "step": 1566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940742479743186e-06,
+      "loss": 0.5276,
+      "step": 1567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9285803148544155e-06,
+      "loss": 0.5872,
+      "step": 1568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916425364719975e-06,
+      "loss": 0.6352,
+      "step": 1569
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904277650879027e-06,
+      "loss": 0.6446,
+      "step": 1570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892137194857914e-06,
+      "loss": 0.5808,
+      "step": 1571
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.88000401817012e-06,
+      "loss": 0.539,
+      "step": 1572
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867878142316221e-06,
+      "loss": 0.5012,
+      "step": 1573
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.855759588783861e-06,
+      "loss": 0.5604,
+      "step": 1574
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843648379047708e-06,
+      "loss": 0.4624,
+      "step": 1575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8315445345694196e-06,
+      "loss": 0.4802,
+      "step": 1576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8194480767976e-06,
+      "loss": 0.6033,
+      "step": 1577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.807359027167753e-06,
+      "loss": 0.4447,
+      "step": 1578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795277407102273e-06,
+      "loss": 0.5931,
+      "step": 1579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.783203238010382e-06,
+      "loss": 0.5294,
+      "step": 1580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7711365412880895e-06,
+      "loss": 0.5798,
+      "step": 1581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759077338318173e-06,
+      "loss": 0.5927,
+      "step": 1582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747025650470135e-06,
+      "loss": 0.5056,
+      "step": 1583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.734981499100145e-06,
+      "loss": 0.5459,
+      "step": 1584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7229449055510335e-06,
+      "loss": 0.5095,
+      "step": 1585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.710915891152227e-06,
+      "loss": 0.6088,
+      "step": 1586
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698894477219726e-06,
+      "loss": 0.5857,
+      "step": 1587
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68688068505607e-06,
+      "loss": 0.5074,
+      "step": 1588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674874535950279e-06,
+      "loss": 0.5006,
+      "step": 1589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662876051177831e-06,
+      "loss": 0.5483,
+      "step": 1590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.650885252000631e-06,
+      "loss": 0.6472,
+      "step": 1591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638902159666962e-06,
+      "loss": 0.5702,
+      "step": 1592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626926795411447e-06,
+      "loss": 0.5984,
+      "step": 1593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.614959180455016e-06,
+      "loss": 0.477,
+      "step": 1594
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.602999336004862e-06,
+      "loss": 0.5572,
+      "step": 1595
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.591047283254417e-06,
+      "loss": 0.5436,
+      "step": 1596
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579103043383305e-06,
+      "loss": 0.4919,
+      "step": 1597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567166637557293e-06,
+      "loss": 0.5874,
+      "step": 1598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5552380869282855e-06,
+      "loss": 0.5041,
+      "step": 1599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543317412634244e-06,
+      "loss": 0.5455,
+      "step": 1600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531404635799191e-06,
+      "loss": 0.5052,
+      "step": 1601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.519499777533154e-06,
+      "loss": 0.4949,
+      "step": 1602
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.507602858932113e-06,
+      "loss": 0.5435,
+      "step": 1603
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495713901077995e-06,
+      "loss": 0.5928,
+      "step": 1604
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4838329250386076e-06,
+      "loss": 0.6195,
+      "step": 1605
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471959951867627e-06,
+      "loss": 0.5406,
+      "step": 1606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.4779,
+      "step": 1607
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4482380982745985e-06,
+      "loss": 0.5761,
+      "step": 1608
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436389259888841e-06,
+      "loss": 0.518,
+      "step": 1609
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.424548508443972e-06,
+      "loss": 0.5765,
+      "step": 1610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412715864922389e-06,
+      "loss": 0.5839,
+      "step": 1611
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400891350292105e-06,
+      "loss": 0.5213,
+      "step": 1612
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.38907498550674e-06,
+      "loss": 0.5741,
+      "step": 1613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.377266791505476e-06,
+      "loss": 0.5752,
+      "step": 1614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365466789213001e-06,
+      "loss": 0.5054,
+      "step": 1615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3536749995394945e-06,
+      "loss": 0.4276,
+      "step": 1616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341891443380585e-06,
+      "loss": 0.608,
+      "step": 1617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.330116141617308e-06,
+      "loss": 0.5137,
+      "step": 1618
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318349115116079e-06,
+      "loss": 0.4922,
+      "step": 1619
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.306590384728638e-06,
+      "loss": 0.5918,
+      "step": 1620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294839971292026e-06,
+      "loss": 0.6453,
+      "step": 1621
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283097895628552e-06,
+      "loss": 0.5998,
+      "step": 1622
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2713641785457504e-06,
+      "loss": 0.4878,
+      "step": 1623
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.259638840836332e-06,
+      "loss": 0.2957,
+      "step": 1624
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.247921903278177e-06,
+      "loss": 0.2726,
+      "step": 1625
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2362133866342625e-06,
+      "loss": 0.2722,
+      "step": 1626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.224513311652654e-06,
+      "loss": 0.2436,
+      "step": 1627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212821699066459e-06,
+      "loss": 0.2151,
+      "step": 1628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20113856959378e-06,
+      "loss": 0.3148,
+      "step": 1629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189463943937697e-06,
+      "loss": 0.2805,
+      "step": 1630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1777978427862094e-06,
+      "loss": 0.2377,
+      "step": 1631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166140286812227e-06,
+      "loss": 0.1958,
+      "step": 1632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1544912966735e-06,
+      "loss": 0.2196,
+      "step": 1633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.142850893012609e-06,
+      "loss": 0.2997,
+      "step": 1634
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131219096456921e-06,
+      "loss": 0.2391,
+      "step": 1635
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.119595927618546e-06,
+      "loss": 0.261,
+      "step": 1636
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1079814070943e-06,
+      "loss": 0.2806,
+      "step": 1637
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0963755554656856e-06,
+      "loss": 0.2571,
+      "step": 1638
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.084778393298837e-06,
+      "loss": 0.3336,
+      "step": 1639
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073189941144495e-06,
+      "loss": 0.2405,
+      "step": 1640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.06161021953796e-06,
+      "loss": 0.2939,
+      "step": 1641
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.050039248999057e-06,
+      "loss": 0.2555,
+      "step": 1642
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0384770500321175e-06,
+      "loss": 0.2215,
+      "step": 1643
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.026923643125924e-06,
+      "loss": 0.2832,
+      "step": 1644
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015379048753669e-06,
+      "loss": 0.2368,
+      "step": 1645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.003843287372947e-06,
+      "loss": 0.2394,
+      "step": 1646
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9923163794256805e-06,
+      "loss": 0.2788,
+      "step": 1647
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.980798345338117e-06,
+      "loss": 0.2493,
+      "step": 1648
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9692892055207784e-06,
+      "loss": 0.2366,
+      "step": 1649
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957788980368416e-06,
+      "loss": 0.2715,
+      "step": 1650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9462976902599945e-06,
+      "loss": 0.2447,
+      "step": 1651
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.934815355558636e-06,
+      "loss": 0.255,
+      "step": 1652
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923341996611604e-06,
+      "loss": 0.2384,
+      "step": 1653
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9118776337502425e-06,
+      "loss": 0.2678,
+      "step": 1654
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.900422287289966e-06,
+      "loss": 0.2786,
+      "step": 1655
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.888975977530212e-06,
+      "loss": 0.2506,
+      "step": 1656
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.877538724754392e-06,
+      "loss": 0.217,
+      "step": 1657
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866110549229881e-06,
+      "loss": 0.2486,
+      "step": 1658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85469147120796e-06,
+      "loss": 0.2525,
+      "step": 1659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.843281510923793e-06,
+      "loss": 0.2954,
+      "step": 1660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.831880688596392e-06,
+      "loss": 0.1827,
+      "step": 1661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.820489024428566e-06,
+      "loss": 0.213,
+      "step": 1662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809106538606896e-06,
+      "loss": 0.2314,
+      "step": 1663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797733251301705e-06,
+      "loss": 0.2488,
+      "step": 1664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7863691826670146e-06,
+      "loss": 0.2319,
+      "step": 1665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.775014352840512e-06,
+      "loss": 0.2566,
+      "step": 1666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7636687819435066e-06,
+      "loss": 0.2962,
+      "step": 1667
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7523324900808986e-06,
+      "loss": 0.257,
+      "step": 1668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741005497341154e-06,
+      "loss": 0.29,
+      "step": 1669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.729687823796262e-06,
+      "loss": 0.2551,
+      "step": 1670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.718379489501682e-06,
+      "loss": 0.256,
+      "step": 1671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.707080514496345e-06,
+      "loss": 0.2352,
+      "step": 1672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695790918802577e-06,
+      "loss": 0.1649,
+      "step": 1673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.684510722426094e-06,
+      "loss": 0.1883,
+      "step": 1674
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673239945355962e-06,
+      "loss": 0.2846,
+      "step": 1675
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.661978607564538e-06,
+      "loss": 0.2612,
+      "step": 1676
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.650726729007465e-06,
+      "loss": 0.2801,
+      "step": 1677
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639484329623627e-06,
+      "loss": 0.2749,
+      "step": 1678
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628251429335099e-06,
+      "loss": 0.2424,
+      "step": 1679
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.617028048047124e-06,
+      "loss": 0.2574,
+      "step": 1680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.605814205648087e-06,
+      "loss": 0.2772,
+      "step": 1681
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594609922009462e-06,
+      "loss": 0.2159,
+      "step": 1682
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.583415216985791e-06,
+      "loss": 0.2786,
+      "step": 1683
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572230110414633e-06,
+      "loss": 0.255,
+      "step": 1684
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.56105462211654e-06,
+      "loss": 0.211,
+      "step": 1685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5498887718950244e-06,
+      "loss": 0.2462,
+      "step": 1686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.538732579536523e-06,
+      "loss": 0.286,
+      "step": 1687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5275860648103496e-06,
+      "loss": 0.246,
+      "step": 1688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516449247468666e-06,
+      "loss": 0.3446,
+      "step": 1689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505322147246463e-06,
+      "loss": 0.3438,
+      "step": 1690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.494204783861502e-06,
+      "loss": 0.2451,
+      "step": 1691
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4830971770142985e-06,
+      "loss": 0.2593,
+      "step": 1692
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4719993463880695e-06,
+      "loss": 0.2224,
+      "step": 1693
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460911311648709e-06,
+      "loss": 0.2472,
+      "step": 1694
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4498330924447596e-06,
+      "loss": 0.2201,
+      "step": 1695
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4387647084073695e-06,
+      "loss": 0.2325,
+      "step": 1696
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.427706179150247e-06,
+      "loss": 0.3582,
+      "step": 1697
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.416657524269652e-06,
+      "loss": 0.2418,
+      "step": 1698
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.40561876334434e-06,
+      "loss": 0.2938,
+      "step": 1699
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394589915935533e-06,
+      "loss": 0.2141,
+      "step": 1700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.383571001586883e-06,
+      "loss": 0.216,
+      "step": 1701
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3725620398244454e-06,
+      "loss": 0.2751,
+      "step": 1702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.361563050156639e-06,
+      "loss": 0.2232,
+      "step": 1703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3505740520742134e-06,
+      "loss": 0.2805,
+      "step": 1704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339595065050206e-06,
+      "loss": 0.2571,
+      "step": 1705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328626108539914e-06,
+      "loss": 0.3001,
+      "step": 1706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.317667201980868e-06,
+      "loss": 0.1833,
+      "step": 1707
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3067183647927855e-06,
+      "loss": 0.2197,
+      "step": 1708
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29577961637754e-06,
+      "loss": 0.274,
+      "step": 1709
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.284850976119121e-06,
+      "loss": 0.2418,
+      "step": 1710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273932463383619e-06,
+      "loss": 0.2115,
+      "step": 1711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2630240975191695e-06,
+      "loss": 0.2552,
+      "step": 1712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2521258978559324e-06,
+      "loss": 0.2395,
+      "step": 1713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2412378837060465e-06,
+      "loss": 0.2751,
+      "step": 1714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2303600743636e-06,
+      "loss": 0.2147,
+      "step": 1715
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219492489104604e-06,
+      "loss": 0.2296,
+      "step": 1716
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208635147186956e-06,
+      "loss": 0.249,
+      "step": 1717
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197788067850388e-06,
+      "loss": 0.203,
+      "step": 1718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186951270316455e-06,
+      "loss": 0.2284,
+      "step": 1719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176124773788497e-06,
+      "loss": 0.2475,
+      "step": 1720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165308597451586e-06,
+      "loss": 0.2101,
+      "step": 1721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154502760472522e-06,
+      "loss": 0.2283,
+      "step": 1722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.143707281999767e-06,
+      "loss": 0.2718,
+      "step": 1723
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13292218116344e-06,
+      "loss": 0.3606,
+      "step": 1724
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.2531,
+      "step": 1725
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111383188828553e-06,
+      "loss": 0.2605,
+      "step": 1726
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100629335498131e-06,
+      "loss": 0.2688,
+      "step": 1727
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0898859361403595e-06,
+      "loss": 0.2588,
+      "step": 1728
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079153009793068e-06,
+      "loss": 0.2346,
+      "step": 1729
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068430575475526e-06,
+      "loss": 0.2185,
+      "step": 1730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057718652188409e-06,
+      "loss": 0.21,
+      "step": 1731
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.047017258913765e-06,
+      "loss": 0.1869,
+      "step": 1732
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.036326414614985e-06,
+      "loss": 0.2041,
+      "step": 1733
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025646138236774e-06,
+      "loss": 0.2325,
+      "step": 1734
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014976448705095e-06,
+      "loss": 0.2509,
+      "step": 1735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004317364927164e-06,
+      "loss": 0.2113,
+      "step": 1736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99366890579139e-06,
+      "loss": 0.2511,
+      "step": 1737
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983031090167368e-06,
+      "loss": 0.2316,
+      "step": 1738
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97240393690583e-06,
+      "loss": 0.2598,
+      "step": 1739
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9617874648386e-06,
+      "loss": 0.242,
+      "step": 1740
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951181692778594e-06,
+      "loss": 0.2533,
+      "step": 1741
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94058663951975e-06,
+      "loss": 0.2956,
+      "step": 1742
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930002323837026e-06,
+      "loss": 0.1934,
+      "step": 1743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.919428764486338e-06,
+      "loss": 0.2516,
+      "step": 1744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908865980204555e-06,
+      "loss": 0.2403,
+      "step": 1745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898313989709447e-06,
+      "loss": 0.2412,
+      "step": 1746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88777281169965e-06,
+      "loss": 0.2444,
+      "step": 1747
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877242464854654e-06,
+      "loss": 0.2781,
+      "step": 1748
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86672296783474e-06,
+      "loss": 0.2056,
+      "step": 1749
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85621433928097e-06,
+      "loss": 0.2467,
+      "step": 1750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845716597815154e-06,
+      "loss": 0.2356,
+      "step": 1751
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835229762039798e-06,
+      "loss": 0.2284,
+      "step": 1752
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.824753850538082e-06,
+      "loss": 0.2827,
+      "step": 1753
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8142888818738367e-06,
+      "loss": 0.2489,
+      "step": 1754
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8038348745914966e-06,
+      "loss": 0.2418,
+      "step": 1755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7933918472160757e-06,
+      "loss": 0.2556,
+      "step": 1756
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.782959818253126e-06,
+      "loss": 0.2066,
+      "step": 1757
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7725388061887056e-06,
+      "loss": 0.2399,
+      "step": 1758
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7621288294893634e-06,
+      "loss": 0.2123,
+      "step": 1759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7517299066020874e-06,
+      "loss": 0.2127,
+      "step": 1760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741342055954269e-06,
+      "loss": 0.2767,
+      "step": 1761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7309652959536947e-06,
+      "loss": 0.1938,
+      "step": 1762
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.720599644988482e-06,
+      "loss": 0.3144,
+      "step": 1763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7102451214270776e-06,
+      "loss": 0.1957,
+      "step": 1764
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.699901743618194e-06,
+      "loss": 0.2044,
+      "step": 1765
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.689569529890805e-06,
+      "loss": 0.214,
+      "step": 1766
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6792484985541034e-06,
+      "loss": 0.2696,
+      "step": 1767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6689386678974504e-06,
+      "loss": 0.25,
+      "step": 1768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.658640056190378e-06,
+      "loss": 0.255,
+      "step": 1769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.64835268168252e-06,
+      "loss": 0.198,
+      "step": 1770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6380765626036095e-06,
+      "loss": 0.2871,
+      "step": 1771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6278117171634366e-06,
+      "loss": 0.2597,
+      "step": 1772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.617558163551802e-06,
+      "loss": 0.2177,
+      "step": 1773
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607315919938501e-06,
+      "loss": 0.2662,
+      "step": 1774
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.597085004473293e-06,
+      "loss": 0.2458,
+      "step": 1775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.586865435285858e-06,
+      "loss": 0.296,
+      "step": 1776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.576657230485775e-06,
+      "loss": 0.25,
+      "step": 1777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5664604081624787e-06,
+      "loss": 0.2613,
+      "step": 1778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.556274986385231e-06,
+      "loss": 0.2094,
+      "step": 1779
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.546100983203099e-06,
+      "loss": 0.1816,
+      "step": 1780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5359384166449185e-06,
+      "loss": 0.2269,
+      "step": 1781
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5257873047192448e-06,
+      "loss": 0.2249,
+      "step": 1782
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.51564766541435e-06,
+      "loss": 0.237,
+      "step": 1783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.505519516698165e-06,
+      "loss": 0.3057,
+      "step": 1784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4954028765182633e-06,
+      "loss": 0.3133,
+      "step": 1785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4852977628018323e-06,
+      "loss": 0.3499,
+      "step": 1786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.475204193455618e-06,
+      "loss": 0.2414,
+      "step": 1787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4651221863659236e-06,
+      "loss": 0.302,
+      "step": 1788
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4550517593985512e-06,
+      "loss": 0.2449,
+      "step": 1789
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4449929303987963e-06,
+      "loss": 0.1947,
+      "step": 1790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434945717191388e-06,
+      "loss": 0.2294,
+      "step": 1791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4249101375804804e-06,
+      "loss": 0.2958,
+      "step": 1792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414886209349615e-06,
+      "loss": 0.2597,
+      "step": 1793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4048739502616747e-06,
+      "loss": 0.2176,
+      "step": 1794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.394873378058876e-06,
+      "loss": 0.2219,
+      "step": 1795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.384884510462717e-06,
+      "loss": 0.2022,
+      "step": 1796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3749073651739594e-06,
+      "loss": 0.2298,
+      "step": 1797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3649419598725964e-06,
+      "loss": 0.2568,
+      "step": 1798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3549883122178086e-06,
+      "loss": 0.3431,
+      "step": 1799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.345046439847941e-06,
+      "loss": 0.2336,
+      "step": 1800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3351163603804805e-06,
+      "loss": 0.2458,
+      "step": 1801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.325198091412013e-06,
+      "loss": 0.2538,
+      "step": 1802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3152916505181976e-06,
+      "loss": 0.2347,
+      "step": 1803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3053970552537285e-06,
+      "loss": 0.2661,
+      "step": 1804
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2955143231523067e-06,
+      "loss": 0.236,
+      "step": 1805
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2856434717266193e-06,
+      "loss": 0.2925,
+      "step": 1806
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2757845184683e-06,
+      "loss": 0.3279,
+      "step": 1807
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2659374808478892e-06,
+      "loss": 0.1701,
+      "step": 1808
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2561023763148237e-06,
+      "loss": 0.2239,
+      "step": 1809
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2462792222973826e-06,
+      "loss": 0.2376,
+      "step": 1810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2364680362026767e-06,
+      "loss": 0.2611,
+      "step": 1811
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2266688354166107e-06,
+      "loss": 0.2332,
+      "step": 1812
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.216881637303839e-06,
+      "loss": 0.2581,
+      "step": 1813
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207106459207758e-06,
+      "loss": 0.3197,
+      "step": 1814
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1973433184504632e-06,
+      "loss": 0.204,
+      "step": 1815
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1875922323327137e-06,
+      "loss": 0.2729,
+      "step": 1816
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.177853218133905e-06,
+      "loss": 0.2573,
+      "step": 1817
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1681262931120504e-06,
+      "loss": 0.2006,
+      "step": 1818
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.158411474503735e-06,
+      "loss": 0.2914,
+      "step": 1819
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1487087795240976e-06,
+      "loss": 0.222,
+      "step": 1820
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1390182253667745e-06,
+      "loss": 0.2342,
+      "step": 1821
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1293398292039077e-06,
+      "loss": 0.3162,
+      "step": 1822
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1196736081860855e-06,
+      "loss": 0.2711,
+      "step": 1823
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110019579442328e-06,
+      "loss": 0.2345,
+      "step": 1824
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.100377760080041e-06,
+      "loss": 0.2454,
+      "step": 1825
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.090748167184997e-06,
+      "loss": 0.2103,
+      "step": 1826
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0811308178213063e-06,
+      "loss": 0.1984,
+      "step": 1827
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0715257290313836e-06,
+      "loss": 0.2253,
+      "step": 1828
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0619329178359103e-06,
+      "loss": 0.2281,
+      "step": 1829
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0523524012338224e-06,
+      "loss": 0.2742,
+      "step": 1830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.042784196202255e-06,
+      "loss": 0.2047,
+      "step": 1831
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0332283196965384e-06,
+      "loss": 0.2053,
+      "step": 1832
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.023684788650154e-06,
+      "loss": 0.2267,
+      "step": 1833
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0141536199747e-06,
+      "loss": 0.2476,
+      "step": 1834
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004634830559874e-06,
+      "loss": 0.2302,
+      "step": 1835
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9951284372734392e-06,
+      "loss": 0.2569,
+      "step": 1836
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.985634456961184e-06,
+      "loss": 0.2088,
+      "step": 1837
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.976152906446903e-06,
+      "loss": 0.1999,
+      "step": 1838
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9666838025323685e-06,
+      "loss": 0.1999,
+      "step": 1839
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9572271619972957e-06,
+      "loss": 0.2858,
+      "step": 1840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.947783001599315e-06,
+      "loss": 0.2001,
+      "step": 1841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.938351338073937e-06,
+      "loss": 0.2247,
+      "step": 1842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.2162,
+      "step": 1843
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9195255684722778e-06,
+      "loss": 0.2433,
+      "step": 1844
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9101314957561864e-06,
+      "loss": 0.2246,
+      "step": 1845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9007499866330037e-06,
+      "loss": 0.3401,
+      "step": 1846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.891381057727216e-06,
+      "loss": 0.2007,
+      "step": 1847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8820247256410272e-06,
+      "loss": 0.2161,
+      "step": 1848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8726810069543156e-06,
+      "loss": 0.3225,
+      "step": 1849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.863349918224607e-06,
+      "loss": 0.1747,
+      "step": 1850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8540314759870446e-06,
+      "loss": 0.2043,
+      "step": 1851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.844725696754359e-06,
+      "loss": 0.2779,
+      "step": 1852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8354325970168483e-06,
+      "loss": 0.2291,
+      "step": 1853
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826152193242342e-06,
+      "loss": 0.2322,
+      "step": 1854
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8168845018761616e-06,
+      "loss": 0.2788,
+      "step": 1855
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8076295393411126e-06,
+      "loss": 0.2279,
+      "step": 1856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7983873220374415e-06,
+      "loss": 0.2413,
+      "step": 1857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7891578663428033e-06,
+      "loss": 0.2602,
+      "step": 1858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7799411886122496e-06,
+      "loss": 0.238,
+      "step": 1859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.770737305178176e-06,
+      "loss": 0.2843,
+      "step": 1860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7615462323503186e-06,
+      "loss": 0.2834,
+      "step": 1861
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7523679864157083e-06,
+      "loss": 0.2222,
+      "step": 1862
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7432025836386412e-06,
+      "loss": 0.2896,
+      "step": 1863
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.734050040260655e-06,
+      "loss": 0.2222,
+      "step": 1864
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.724910372500508e-06,
+      "loss": 0.3001,
+      "step": 1865
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715783596554136e-06,
+      "loss": 0.1785,
+      "step": 1866
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7066697285946376e-06,
+      "loss": 0.2327,
+      "step": 1867
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6975687847722197e-06,
+      "loss": 0.2744,
+      "step": 1868
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6884807812142043e-06,
+      "loss": 0.2375,
+      "step": 1869
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.679405734024977e-06,
+      "loss": 0.2049,
+      "step": 1870
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670343659285968e-06,
+      "loss": 0.243,
+      "step": 1871
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6612945730556115e-06,
+      "loss": 0.2648,
+      "step": 1872
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6522584913693295e-06,
+      "loss": 0.2579,
+      "step": 1873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.643235430239499e-06,
+      "loss": 0.2406,
+      "step": 1874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6342254056554306e-06,
+      "loss": 0.2848,
+      "step": 1875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62522843358332e-06,
+      "loss": 0.2269,
+      "step": 1876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.616244529966244e-06,
+      "loss": 0.2311,
+      "step": 1877
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.607273710724121e-06,
+      "loss": 0.1959,
+      "step": 1878
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.598315991753675e-06,
+      "loss": 0.2513,
+      "step": 1879
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5893713889284257e-06,
+      "loss": 0.2659,
+      "step": 1880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5804399180986417e-06,
+      "loss": 0.2592,
+      "step": 1881
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5715215950913253e-06,
+      "loss": 0.1922,
+      "step": 1882
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5626164357101857e-06,
+      "loss": 0.2214,
+      "step": 1883
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5537244557355965e-06,
+      "loss": 0.2318,
+      "step": 1884
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.544845670924575e-06,
+      "loss": 0.2836,
+      "step": 1885
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5359800970107663e-06,
+      "loss": 0.2222,
+      "step": 1886
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527127749704399e-06,
+      "loss": 0.2877,
+      "step": 1887
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5182886446922673e-06,
+      "loss": 0.215,
+      "step": 1888
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.509462797637693e-06,
+      "loss": 0.1795,
+      "step": 1889
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5006502241805064e-06,
+      "loss": 0.3077,
+      "step": 1890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4918509399370194e-06,
+      "loss": 0.3146,
+      "step": 1891
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4830649604999967e-06,
+      "loss": 0.2538,
+      "step": 1892
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4742923014386154e-06,
+      "loss": 0.3021,
+      "step": 1893
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4655329782984617e-06,
+      "loss": 0.2217,
+      "step": 1894
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4567870066014785e-06,
+      "loss": 0.2357,
+      "step": 1895
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448054401845957e-06,
+      "loss": 0.2525,
+      "step": 1896
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4393351795065023e-06,
+      "loss": 0.1937,
+      "step": 1897
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4306293550339943e-06,
+      "loss": 0.3099,
+      "step": 1898
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.421936943855586e-06,
+      "loss": 0.2315,
+      "step": 1899
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4132579613746475e-06,
+      "loss": 0.2648,
+      "step": 1900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4045924229707663e-06,
+      "loss": 0.2625,
+      "step": 1901
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.395940343999691e-06,
+      "loss": 0.2323,
+      "step": 1902
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387301739793333e-06,
+      "loss": 0.256,
+      "step": 1903
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3786766256597226e-06,
+      "loss": 0.2832,
+      "step": 1904
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3700650168829765e-06,
+      "loss": 0.2305,
+      "step": 1905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.361466928723293e-06,
+      "loss": 0.2059,
+      "step": 1906
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.352882376416895e-06,
+      "loss": 0.2269,
+      "step": 1907
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.344311375176034e-06,
+      "loss": 0.1862,
+      "step": 1908
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3357539401889438e-06,
+      "loss": 0.1913,
+      "step": 1909
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3272100866198133e-06,
+      "loss": 0.3016,
+      "step": 1910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3186798296087663e-06,
+      "loss": 0.2021,
+      "step": 1911
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3101631842718376e-06,
+      "loss": 0.2962,
+      "step": 1912
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3016601657009364e-06,
+      "loss": 0.2406,
+      "step": 1913
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.293170788963831e-06,
+      "loss": 0.2052,
+      "step": 1914
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284695069104107e-06,
+      "loss": 0.2693,
+      "step": 1915
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2762330211411523e-06,
+      "loss": 0.2062,
+      "step": 1916
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2677846600701305e-06,
+      "loss": 0.2078,
+      "step": 1917
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.259350000861952e-06,
+      "loss": 0.2344,
+      "step": 1918
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2509290584632394e-06,
+      "loss": 0.2208,
+      "step": 1919
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2425218477963197e-06,
+      "loss": 0.2057,
+      "step": 1920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234128383759174e-06,
+      "loss": 0.1667,
+      "step": 1921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2257486812254336e-06,
+      "loss": 0.2648,
+      "step": 1922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2173827550443417e-06,
+      "loss": 0.3612,
+      "step": 1923
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.209030620040723e-06,
+      "loss": 0.2257,
+      "step": 1924
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2006922910149743e-06,
+      "loss": 0.1899,
+      "step": 1925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.192367782743016e-06,
+      "loss": 0.2461,
+      "step": 1926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1840571099762865e-06,
+      "loss": 0.1491,
+      "step": 1927
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1757602874416993e-06,
+      "loss": 0.2768,
+      "step": 1928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.167477329841633e-06,
+      "loss": 0.2358,
+      "step": 1929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1592082518538926e-06,
+      "loss": 0.252,
+      "step": 1930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.150953068131686e-06,
+      "loss": 0.229,
+      "step": 1931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.142711793303599e-06,
+      "loss": 0.2405,
+      "step": 1932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1344844419735757e-06,
+      "loss": 0.2616,
+      "step": 1933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1262710287208833e-06,
+      "loss": 0.2776,
+      "step": 1934
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.118071568100094e-06,
+      "loss": 0.2241,
+      "step": 1935
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1098860746410498e-06,
+      "loss": 0.1682,
+      "step": 1936
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.101714562848841e-06,
+      "loss": 0.254,
+      "step": 1937
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0935570472037892e-06,
+      "loss": 0.2709,
+      "step": 1938
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0854135421614108e-06,
+      "loss": 0.2177,
+      "step": 1939
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0772840621523905e-06,
+      "loss": 0.2648,
+      "step": 1940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069168621582567e-06,
+      "loss": 0.2566,
+      "step": 1941
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061067234832893e-06,
+      "loss": 0.2936,
+      "step": 1942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0529799162594242e-06,
+      "loss": 0.2977,
+      "step": 1943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.044906680193285e-06,
+      "loss": 0.2156,
+      "step": 1944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0368475409406396e-06,
+      "loss": 0.2732,
+      "step": 1945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0288025127826806e-06,
+      "loss": 0.2869,
+      "step": 1946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0207716099755838e-06,
+      "loss": 0.2369,
+      "step": 1947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012754846750509e-06,
+      "loss": 0.217,
+      "step": 1948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.004752237313544e-06,
+      "loss": 0.216,
+      "step": 1949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9967637958457066e-06,
+      "loss": 0.2754,
+      "step": 1950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9887895365029077e-06,
+      "loss": 0.2766,
+      "step": 1951
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9808294734159197e-06,
+      "loss": 0.24,
+      "step": 1952
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972883620690366e-06,
+      "loss": 0.3273,
+      "step": 1953
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9649519924066797e-06,
+      "loss": 0.2091,
+      "step": 1954
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957034602620098e-06,
+      "loss": 0.1846,
+      "step": 1955
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949131465360624e-06,
+      "loss": 0.2212,
+      "step": 1956
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9412425946329994e-06,
+      "loss": 0.2188,
+      "step": 1957
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9333680044166847e-06,
+      "loss": 0.2731,
+      "step": 1958
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.925507708665841e-06,
+      "loss": 0.2514,
+      "step": 1959
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9176617213092973e-06,
+      "loss": 0.2029,
+      "step": 1960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.25,
+      "step": 1961
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9020127273676204e-06,
+      "loss": 0.2291,
+      "step": 1962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8942097485132626e-06,
+      "loss": 0.2498,
+      "step": 1963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8864211335147165e-06,
+      "loss": 0.2556,
+      "step": 1964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8786468961737902e-06,
+      "loss": 0.2859,
+      "step": 1965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8708870502668075e-06,
+      "loss": 0.2185,
+      "step": 1966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8631416095445965e-06,
+      "loss": 0.222,
+      "step": 1967
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8554105877324525e-06,
+      "loss": 0.252,
+      "step": 1968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8476939985301257e-06,
+      "loss": 0.24,
+      "step": 1969
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8399918556117913e-06,
+      "loss": 0.2471,
+      "step": 1970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8323041726260172e-06,
+      "loss": 0.2188,
+      "step": 1971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824630963195756e-06,
+      "loss": 0.2426,
+      "step": 1972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.81697224091831e-06,
+      "loss": 0.1975,
+      "step": 1973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8093280193653074e-06,
+      "loss": 0.2941,
+      "step": 1974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8016983120826792e-06,
+      "loss": 0.2624,
+      "step": 1975
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7940831325906417e-06,
+      "loss": 0.2283,
+      "step": 1976
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7864824943836633e-06,
+      "loss": 0.2163,
+      "step": 1977
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7788964109304495e-06,
+      "loss": 0.2199,
+      "step": 1978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7713248956739082e-06,
+      "loss": 0.2581,
+      "step": 1979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7637679620311287e-06,
+      "loss": 0.2087,
+      "step": 1980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7562256233933717e-06,
+      "loss": 0.3089,
+      "step": 1981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7486978931260313e-06,
+      "loss": 0.2645,
+      "step": 1982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7411847845686082e-06,
+      "loss": 0.2824,
+      "step": 1983
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7336863110346968e-06,
+      "loss": 0.2296,
+      "step": 1984
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7262024858119597e-06,
+      "loss": 0.2342,
+      "step": 1985
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7187333221621006e-06,
+      "loss": 0.2437,
+      "step": 1986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.711278833320844e-06,
+      "loss": 0.3142,
+      "step": 1987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.703839032497906e-06,
+      "loss": 0.2982,
+      "step": 1988
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6964139328769736e-06,
+      "loss": 0.1708,
+      "step": 1989
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6890035476156884e-06,
+      "loss": 0.2036,
+      "step": 1990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6816078898456178e-06,
+      "loss": 0.2379,
+      "step": 1991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6742269726722217e-06,
+      "loss": 0.1803,
+      "step": 1992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6668608091748495e-06,
+      "loss": 0.2623,
+      "step": 1993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6595094124067035e-06,
+      "loss": 0.1986,
+      "step": 1994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6521727953948164e-06,
+      "loss": 0.2596,
+      "step": 1995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6448509711400273e-06,
+      "loss": 0.3034,
+      "step": 1996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.637543952616969e-06,
+      "loss": 0.2521,
+      "step": 1997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6302517527740358e-06,
+      "loss": 0.2619,
+      "step": 1998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6229743845333635e-06,
+      "loss": 0.1621,
+      "step": 1999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6157118607908006e-06,
+      "loss": 0.2861,
+      "step": 2000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6084641944158918e-06,
+      "loss": 0.2541,
+      "step": 2001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.601231398251859e-06,
+      "loss": 0.226,
+      "step": 2002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5940134851155698e-06,
+      "loss": 0.2685,
+      "step": 2003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5868104677975183e-06,
+      "loss": 0.2974,
+      "step": 2004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5796223590617987e-06,
+      "loss": 0.2989,
+      "step": 2005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5724491716460932e-06,
+      "loss": 0.267,
+      "step": 2006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5652909182616404e-06,
+      "loss": 0.235,
+      "step": 2007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.558147611593216e-06,
+      "loss": 0.2001,
+      "step": 2008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5510192642991073e-06,
+      "loss": 0.2893,
+      "step": 2009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5439058890110892e-06,
+      "loss": 0.2471,
+      "step": 2010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5368074983344128e-06,
+      "loss": 0.2439,
+      "step": 2011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.529724104847775e-06,
+      "loss": 0.2386,
+      "step": 2012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.522655721103291e-06,
+      "loss": 0.2367,
+      "step": 2013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5156023596264835e-06,
+      "loss": 0.2005,
+      "step": 2014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5085640329162544e-06,
+      "loss": 0.2074,
+      "step": 2015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5015407534448577e-06,
+      "loss": 0.2724,
+      "step": 2016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.494532533657893e-06,
+      "loss": 0.1903,
+      "step": 2017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4875393859742626e-06,
+      "loss": 0.2404,
+      "step": 2018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.480561322786167e-06,
+      "loss": 0.266,
+      "step": 2019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4735983564590784e-06,
+      "loss": 0.2215,
+      "step": 2020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4666504993317089e-06,
+      "loss": 0.186,
+      "step": 2021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4597177637159998e-06,
+      "loss": 0.198,
+      "step": 2022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4528001618970966e-06,
+      "loss": 0.2493,
+      "step": 2023
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4458977061333301e-06,
+      "loss": 0.1668,
+      "step": 2024
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4390104086561886e-06,
+      "loss": 0.2039,
+      "step": 2025
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.432138281670299e-06,
+      "loss": 0.1788,
+      "step": 2026
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4252813373534013e-06,
+      "loss": 0.2506,
+      "step": 2027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4184395878563395e-06,
+      "loss": 0.2352,
+      "step": 2028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4116130453030296e-06,
+      "loss": 0.2365,
+      "step": 2029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.404801721790432e-06,
+      "loss": 0.2634,
+      "step": 2030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3980056293885503e-06,
+      "loss": 0.2073,
+      "step": 2031
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3912247801403856e-06,
+      "loss": 0.2326,
+      "step": 2032
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3844591860619382e-06,
+      "loss": 0.2217,
+      "step": 2033
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3777088591421717e-06,
+      "loss": 0.2594,
+      "step": 2034
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3709738113429904e-06,
+      "loss": 0.2254,
+      "step": 2035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3642540545992332e-06,
+      "loss": 0.2858,
+      "step": 2036
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3575496008186307e-06,
+      "loss": 0.2264,
+      "step": 2037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3508604618818067e-06,
+      "loss": 0.2521,
+      "step": 2038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3441866496422385e-06,
+      "loss": 0.2468,
+      "step": 2039
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3375281759262493e-06,
+      "loss": 0.2193,
+      "step": 2040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.330885052532981e-06,
+      "loss": 0.2577,
+      "step": 2041
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3242572912343665e-06,
+      "loss": 0.2577,
+      "step": 2042
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3176449037751294e-06,
+      "loss": 0.2481,
+      "step": 2043
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3110479018727373e-06,
+      "loss": 0.1974,
+      "step": 2044
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3044662972174005e-06,
+      "loss": 0.2659,
+      "step": 2045
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2979001014720472e-06,
+      "loss": 0.2755,
+      "step": 2046
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2913493262722942e-06,
+      "loss": 0.2274,
+      "step": 2047
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2848139832264328e-06,
+      "loss": 0.2358,
+      "step": 2048
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2782940839154113e-06,
+      "loss": 0.2672,
+      "step": 2049
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2717896398928088e-06,
+      "loss": 0.3015,
+      "step": 2050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2653006626848207e-06,
+      "loss": 0.2115,
+      "step": 2051
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2588271637902293e-06,
+      "loss": 0.2509,
+      "step": 2052
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2523691546803872e-06,
+      "loss": 0.2017,
+      "step": 2053
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.245926646799205e-06,
+      "loss": 0.3118,
+      "step": 2054
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.239499651563123e-06,
+      "loss": 0.2645,
+      "step": 2055
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.233088180361085e-06,
+      "loss": 0.2518,
+      "step": 2056
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2266922445545348e-06,
+      "loss": 0.2335,
+      "step": 2057
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2203118554773807e-06,
+      "loss": 0.3091,
+      "step": 2058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2139470244359853e-06,
+      "loss": 0.271,
+      "step": 2059
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2075977627091373e-06,
+      "loss": 0.2185,
+      "step": 2060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.201264081548038e-06,
+      "loss": 0.2242,
+      "step": 2061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.194945992176285e-06,
+      "loss": 0.2025,
+      "step": 2062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1886435057898338e-06,
+      "loss": 0.2404,
+      "step": 2063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1823566335570036e-06,
+      "loss": 0.1756,
+      "step": 2064
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.176085386618434e-06,
+      "loss": 0.2871,
+      "step": 2065
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1698297760870824e-06,
+      "loss": 0.2438,
+      "step": 2066
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1635898130481983e-06,
+      "loss": 0.348,
+      "step": 2067
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1573655085592983e-06,
+      "loss": 0.2508,
+      "step": 2068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151156873650151e-06,
+      "loss": 0.1951,
+      "step": 2069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1449639193227646e-06,
+      "loss": 0.3475,
+      "step": 2070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.138786656551354e-06,
+      "loss": 0.2809,
+      "step": 2071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1326250962823338e-06,
+      "loss": 0.2241,
+      "step": 2072
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1264792494342858e-06,
+      "loss": 0.2003,
+      "step": 2073
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1203491268979504e-06,
+      "loss": 0.2514,
+      "step": 2074
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.114234739536204e-06,
+      "loss": 0.196,
+      "step": 2075
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1081360981840428e-06,
+      "loss": 0.2682,
+      "step": 2076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1020532136485517e-06,
+      "loss": 0.2454,
+      "step": 2077
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0959860967089042e-06,
+      "loss": 0.2635,
+      "step": 2078
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.2678,
+      "step": 2079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0838992085940748e-06,
+      "loss": 0.2553,
+      "step": 2080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0778794588374542e-06,
+      "loss": 0.2494,
+      "step": 2081
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0718755195137442e-06,
+      "loss": 0.2119,
+      "step": 2082
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0658874012622244e-06,
+      "loss": 0.2503,
+      "step": 2083
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0599151146941268e-06,
+      "loss": 0.242,
+      "step": 2084
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0539586703926396e-06,
+      "loss": 0.2633,
+      "step": 2085
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0480180789128691e-06,
+      "loss": 0.3036,
+      "step": 2086
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0420933507818332e-06,
+      "loss": 0.2964,
+      "step": 2087
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0361844964984435e-06,
+      "loss": 0.2758,
+      "step": 2088
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0302915265334722e-06,
+      "loss": 0.2671,
+      "step": 2089
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.024414451329555e-06,
+      "loss": 0.2473,
+      "step": 2090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0185532813011523e-06,
+      "loss": 0.2337,
+      "step": 2091
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0127080268345434e-06,
+      "loss": 0.1679,
+      "step": 2092
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0068786982878087e-06,
+      "loss": 0.1924,
+      "step": 2093
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0010653059907982e-06,
+      "loss": 0.2145,
+      "step": 2094
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.952678602451272e-07,
+      "loss": 0.2173,
+      "step": 2095
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.894863713241532e-07,
+      "loss": 0.2373,
+      "step": 2096
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.837208494729567e-07,
+      "loss": 0.2922,
+      "step": 2097
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77971304908325e-07,
+      "loss": 0.278,
+      "step": 2098
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.722377478187317e-07,
+      "loss": 0.2084,
+      "step": 2099
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66520188364316e-07,
+      "loss": 0.2269,
+      "step": 2100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.608186366768746e-07,
+      "loss": 0.2494,
+      "step": 2101
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.551331028598365e-07,
+      "loss": 0.2525,
+      "step": 2102
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.494635969882426e-07,
+      "loss": 0.3632,
+      "step": 2103
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.438101291087364e-07,
+      "loss": 0.291,
+      "step": 2104
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.381727092395365e-07,
+      "loss": 0.225,
+      "step": 2105
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.325513473704273e-07,
+      "loss": 0.2551,
+      "step": 2106
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.269460534627372e-07,
+      "loss": 0.2597,
+      "step": 2107
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.213568374493176e-07,
+      "loss": 0.2773,
+      "step": 2108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.157837092345334e-07,
+      "loss": 0.2498,
+      "step": 2109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.102266786942426e-07,
+      "loss": 0.2011,
+      "step": 2110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.046857556757704e-07,
+      "loss": 0.2137,
+      "step": 2111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.991609499979037e-07,
+      "loss": 0.1704,
+      "step": 2112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.936522714508678e-07,
+      "loss": 0.203,
+      "step": 2113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.881597297963107e-07,
+      "loss": 0.234,
+      "step": 2114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.82683334767287e-07,
+      "loss": 0.1806,
+      "step": 2115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.772230960682282e-07,
+      "loss": 0.214,
+      "step": 2116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.71779023374949e-07,
+      "loss": 0.2731,
+      "step": 2117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.663511263346114e-07,
+      "loss": 0.2804,
+      "step": 2118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.609394145657146e-07,
+      "loss": 0.2296,
+      "step": 2119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.555438976580743e-07,
+      "loss": 0.2954,
+      "step": 2120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.501645851728091e-07,
+      "loss": 0.2174,
+      "step": 2121
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.448014866423238e-07,
+      "loss": 0.276,
+      "step": 2122
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.394546115702928e-07,
+      "loss": 0.2582,
+      "step": 2123
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.34123969431635e-07,
+      "loss": 0.1809,
+      "step": 2124
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.28809569672514e-07,
+      "loss": 0.2305,
+      "step": 2125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.235114217103012e-07,
+      "loss": 0.2551,
+      "step": 2126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.182295349335734e-07,
+      "loss": 0.2188,
+      "step": 2127
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.129639187020954e-07,
+      "loss": 0.1986,
+      "step": 2128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.077145823467924e-07,
+      "loss": 0.2755,
+      "step": 2129
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.024815351697457e-07,
+      "loss": 0.292,
+      "step": 2130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.972647864441718e-07,
+      "loss": 0.2003,
+      "step": 2131
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.920643454144017e-07,
+      "loss": 0.3067,
+      "step": 2132
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.868802212958704e-07,
+      "loss": 0.2304,
+      "step": 2133
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.817124232751006e-07,
+      "loss": 0.3047,
+      "step": 2134
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.765609605096802e-07,
+      "loss": 0.2629,
+      "step": 2135
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.714258421282572e-07,
+      "loss": 0.2724,
+      "step": 2136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.663070772305081e-07,
+      "loss": 0.2988,
+      "step": 2137
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.274,
+      "step": 2138
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.561186441398393e-07,
+      "loss": 0.1997,
+      "step": 2139
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.510489940013244e-07,
+      "loss": 0.2337,
+      "step": 2140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.459957334552526e-07,
+      "loss": 0.2126,
+      "step": 2141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.409588714562477e-07,
+      "loss": 0.3267,
+      "step": 2142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.359384169298744e-07,
+      "loss": 0.2654,
+      "step": 2143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.309343787726264e-07,
+      "loss": 0.3001,
+      "step": 2144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.259467658519026e-07,
+      "loss": 0.2441,
+      "step": 2145
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.209755870059953e-07,
+      "loss": 0.2682,
+      "step": 2146
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.160208510440747e-07,
+      "loss": 0.2441,
+      "step": 2147
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110825667461762e-07,
+      "loss": 0.2707,
+      "step": 2148
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.061607428631823e-07,
+      "loss": 0.2631,
+      "step": 2149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.012553881168016e-07,
+      "loss": 0.2033,
+      "step": 2150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963665111995633e-07,
+      "loss": 0.2261,
+      "step": 2151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.914941207747972e-07,
+      "loss": 0.2395,
+      "step": 2152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.866382254766158e-07,
+      "loss": 0.2414,
+      "step": 2153
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.817988339099035e-07,
+      "loss": 0.226,
+      "step": 2154
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.769759546502952e-07,
+      "loss": 0.2646,
+      "step": 2155
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72169596244171e-07,
+      "loss": 0.2601,
+      "step": 2156
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673797672086335e-07,
+      "loss": 0.2303,
+      "step": 2157
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.62606476031491e-07,
+      "loss": 0.2032,
+      "step": 2158
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.578497311712484e-07,
+      "loss": 0.2647,
+      "step": 2159
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.531095410570898e-07,
+      "loss": 0.2645,
+      "step": 2160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.483859140888648e-07,
+      "loss": 0.1766,
+      "step": 2161
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.436788586370724e-07,
+      "loss": 0.206,
+      "step": 2162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.3898838304284e-07,
+      "loss": 0.3012,
+      "step": 2163
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.343144956179203e-07,
+      "loss": 0.2631,
+      "step": 2164
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.296572046446725e-07,
+      "loss": 0.3241,
+      "step": 2165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.250165183760426e-07,
+      "loss": 0.2626,
+      "step": 2166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.203924450355514e-07,
+      "loss": 0.1967,
+      "step": 2167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.157849928172832e-07,
+      "loss": 0.2223,
+      "step": 2168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.111941698858681e-07,
+      "loss": 0.2853,
+      "step": 2169
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.066199843764697e-07,
+      "loss": 0.2684,
+      "step": 2170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.020624443947664e-07,
+      "loss": 0.2473,
+      "step": 2171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.975215580169402e-07,
+      "loss": 0.371,
+      "step": 2172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.929973332896677e-07,
+      "loss": 0.1879,
+      "step": 2173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.884897782300914e-07,
+      "loss": 0.269,
+      "step": 2174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.839989008258217e-07,
+      "loss": 0.227,
+      "step": 2175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.795247090349099e-07,
+      "loss": 0.2617,
+      "step": 2176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.750672107858435e-07,
+      "loss": 0.2608,
+      "step": 2177
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.706264139775286e-07,
+      "loss": 0.2457,
+      "step": 2178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.662023264792715e-07,
+      "loss": 0.2025,
+      "step": 2179
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.617949561307701e-07,
+      "loss": 0.2343,
+      "step": 2180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.574043107421023e-07,
+      "loss": 0.2582,
+      "step": 2181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.530303980937046e-07,
+      "loss": 0.259,
+      "step": 2182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.486732259363647e-07,
+      "loss": 0.2183,
+      "step": 2183
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.443328019912042e-07,
+      "loss": 0.2274,
+      "step": 2184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.400091339496638e-07,
+      "loss": 0.2688,
+      "step": 2185
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.357022294734959e-07,
+      "loss": 0.1865,
+      "step": 2186
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.314120961947467e-07,
+      "loss": 0.254,
+      "step": 2187
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.271387417157392e-07,
+      "loss": 0.1648,
+      "step": 2188
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.228821736090684e-07,
+      "loss": 0.215,
+      "step": 2189
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.186423994175771e-07,
+      "loss": 0.2337,
+      "step": 2190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.144194266543557e-07,
+      "loss": 0.2859,
+      "step": 2191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.102132628027168e-07,
+      "loss": 0.2417,
+      "step": 2192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.060239153161872e-07,
+      "loss": 0.2095,
+      "step": 2193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.018513916184963e-07,
+      "loss": 0.2244,
+      "step": 2194
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.976956991035587e-07,
+      "loss": 0.2451,
+      "step": 2195
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.935568451354645e-07,
+      "loss": 0.1786,
+      "step": 2196
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.2102,
+      "step": 2197
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.853296821469589e-07,
+      "loss": 0.2391,
+      "step": 2198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.812413877054833e-07,
+      "loss": 0.2503,
+      "step": 2199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.771699609686919e-07,
+      "loss": 0.2435,
+      "step": 2200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.731154091513546e-07,
+      "loss": 0.2387,
+      "step": 2201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.690777394383339e-07,
+      "loss": 0.2238,
+      "step": 2202
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.650569589845766e-07,
+      "loss": 0.241,
+      "step": 2203
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.610530749151032e-07,
+      "loss": 0.233,
+      "step": 2204
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.570660943249927e-07,
+      "loss": 0.2031,
+      "step": 2205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5309602427936584e-07,
+      "loss": 0.2503,
+      "step": 2206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.491428718133817e-07,
+      "loss": 0.2533,
+      "step": 2207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4520664393222e-07,
+      "loss": 0.2275,
+      "step": 2208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.412873476110702e-07,
+      "loss": 0.2279,
+      "step": 2209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3738498979511545e-07,
+      "loss": 0.2656,
+      "step": 2210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.334995773995221e-07,
+      "loss": 0.2178,
+      "step": 2211
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.296311173094314e-07,
+      "loss": 0.2437,
+      "step": 2212
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2577961637994544e-07,
+      "loss": 0.1933,
+      "step": 2213
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2194508143610925e-07,
+      "loss": 0.2381,
+      "step": 2214
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.181275192729084e-07,
+      "loss": 0.2387,
+      "step": 2215
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1432693665524715e-07,
+      "loss": 0.1865,
+      "step": 2216
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1054334031794373e-07,
+      "loss": 0.2403,
+      "step": 2217
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.067767369657161e-07,
+      "loss": 0.2528,
+      "step": 2218
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0302713327316834e-07,
+      "loss": 0.24,
+      "step": 2219
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.992945358847833e-07,
+      "loss": 0.2574,
+      "step": 2220
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.955789514149022e-07,
+      "loss": 0.2809,
+      "step": 2221
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9188038644772495e-07,
+      "loss": 0.3305,
+      "step": 2222
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8819884753728665e-07,
+      "loss": 0.2554,
+      "step": 2223
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8453434120745535e-07,
+      "loss": 0.2926,
+      "step": 2224
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.808868739519167e-07,
+      "loss": 0.2142,
+      "step": 2225
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.772564522341582e-07,
+      "loss": 0.2036,
+      "step": 2226
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.736430824874637e-07,
+      "loss": 0.2249,
+      "step": 2227
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.700467711149025e-07,
+      "loss": 0.1955,
+      "step": 2228
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6646752448931345e-07,
+      "loss": 0.2294,
+      "step": 2229
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629053489532963e-07,
+      "loss": 0.1743,
+      "step": 2230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5936025081919957e-07,
+      "loss": 0.2289,
+      "step": 2231
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5583223636911027e-07,
+      "loss": 0.2036,
+      "step": 2232
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5232131185484075e-07,
+      "loss": 0.2003,
+      "step": 2233
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.488274834979233e-07,
+      "loss": 0.2985,
+      "step": 2234
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.453507574895898e-07,
+      "loss": 0.253,
+      "step": 2235
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4189113999076983e-07,
+      "loss": 0.2217,
+      "step": 2236
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3844863713207276e-07,
+      "loss": 0.2433,
+      "step": 2237
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.350232550137833e-07,
+      "loss": 0.2084,
+      "step": 2238
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3161499970584597e-07,
+      "loss": 0.315,
+      "step": 2239
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.282238772478541e-07,
+      "loss": 0.1964,
+      "step": 2240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2484989364904295e-07,
+      "loss": 0.2923,
+      "step": 2241
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2149305488827553e-07,
+      "loss": 0.1943,
+      "step": 2242
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1815336691403464e-07,
+      "loss": 0.3159,
+      "step": 2243
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.148308356444085e-07,
+      "loss": 0.2466,
+      "step": 2244
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.115254669670864e-07,
+      "loss": 0.2346,
+      "step": 2245
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.082372667393441e-07,
+      "loss": 0.2641,
+      "step": 2246
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.049662407880294e-07,
+      "loss": 0.2195,
+      "step": 2247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0171239490956237e-07,
+      "loss": 0.2371,
+      "step": 2248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.984757348699152e-07,
+      "loss": 0.2278,
+      "step": 2249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.952562664046088e-07,
+      "loss": 0.2492,
+      "step": 2250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9205399521869847e-07,
+      "loss": 0.2457,
+      "step": 2251
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8886892698676394e-07,
+      "loss": 0.2315,
+      "step": 2252
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.857010673529015e-07,
+      "loss": 0.2349,
+      "step": 2253
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825504219307118e-07,
+      "loss": 0.2202,
+      "step": 2254
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7941699630329556e-07,
+      "loss": 0.2405,
+      "step": 2255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7630079602323447e-07,
+      "loss": 0.2697,
+      "step": 2256
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7320182661258687e-07,
+      "loss": 0.2286,
+      "step": 2257
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.701200935628767e-07,
+      "loss": 0.2114,
+      "step": 2258
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6705560233508787e-07,
+      "loss": 0.1924,
+      "step": 2259
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6400835835964645e-07,
+      "loss": 0.1452,
+      "step": 2260
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6097836703641856e-07,
+      "loss": 0.1814,
+      "step": 2261
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5796563373469585e-07,
+      "loss": 0.2495,
+      "step": 2262
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5497016379318894e-07,
+      "loss": 0.2362,
+      "step": 2263
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5199196252001623e-07,
+      "loss": 0.2737,
+      "step": 2264
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4903103519269724e-07,
+      "loss": 0.3056,
+      "step": 2265
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4608738705813706e-07,
+      "loss": 0.2802,
+      "step": 2266
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4316102333262647e-07,
+      "loss": 0.2011,
+      "step": 2267
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4025194920182405e-07,
+      "loss": 0.2253,
+      "step": 2268
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3736016982075172e-07,
+      "loss": 0.2237,
+      "step": 2269
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3448569031378043e-07,
+      "loss": 0.2563,
+      "step": 2270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.316285157746312e-07,
+      "loss": 0.2887,
+      "step": 2271
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2878865126635618e-07,
+      "loss": 0.1987,
+      "step": 2272
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2596610182133328e-07,
+      "loss": 0.2352,
+      "step": 2273
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2316087244125928e-07,
+      "loss": 0.2707,
+      "step": 2274
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2037296809713448e-07,
+      "loss": 0.2557,
+      "step": 2275
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1760239372926372e-07,
+      "loss": 0.2394,
+      "step": 2276
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1484915424723973e-07,
+      "loss": 0.2192,
+      "step": 2277
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.121132545299376e-07,
+      "loss": 0.191,
+      "step": 2278
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.093946994255036e-07,
+      "loss": 0.2468,
+      "step": 2279
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0669349375135094e-07,
+      "loss": 0.2919,
+      "step": 2280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0400964229414732e-07,
+      "loss": 0.2645,
+      "step": 2281
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0134314980980952e-07,
+      "loss": 0.2294,
+      "step": 2282
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.986940210234922e-07,
+      "loss": 0.2362,
+      "step": 2283
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9606226062957922e-07,
+      "loss": 0.2462,
+      "step": 2284
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9344787329168002e-07,
+      "loss": 0.2253,
+      "step": 2285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908508636426176e-07,
+      "loss": 0.2706,
+      "step": 2286
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8827123628441634e-07,
+      "loss": 0.2477,
+      "step": 2287
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8570899578830293e-07,
+      "loss": 0.1967,
+      "step": 2288
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.831641466946954e-07,
+      "loss": 0.2665,
+      "step": 2289
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8063669351318757e-07,
+      "loss": 0.2725,
+      "step": 2290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7812664072255014e-07,
+      "loss": 0.1859,
+      "step": 2291
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.756339927707196e-07,
+      "loss": 0.2347,
+      "step": 2292
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.731587540747903e-07,
+      "loss": 0.2401,
+      "step": 2293
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.70700929021006e-07,
+      "loss": 0.2434,
+      "step": 2294
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.682605219647515e-07,
+      "loss": 0.2799,
+      "step": 2295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.658375372305465e-07,
+      "loss": 0.2425,
+      "step": 2296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6343197911203978e-07,
+      "loss": 0.2486,
+      "step": 2297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6104385187199812e-07,
+      "loss": 0.1662,
+      "step": 2298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5867315974229968e-07,
+      "loss": 0.2624,
+      "step": 2299
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5631990692392296e-07,
+      "loss": 0.1657,
+      "step": 2300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5398409758695e-07,
+      "loss": 0.2167,
+      "step": 2301
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5166573587054867e-07,
+      "loss": 0.2774,
+      "step": 2302
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4936482588296942e-07,
+      "loss": 0.2339,
+      "step": 2303
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4708137170153626e-07,
+      "loss": 0.2423,
+      "step": 2304
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.448153773726402e-07,
+      "loss": 0.2248,
+      "step": 2305
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4256684691173584e-07,
+      "loss": 0.2958,
+      "step": 2306
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4033578430332706e-07,
+      "loss": 0.1968,
+      "step": 2307
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.381221935009669e-07,
+      "loss": 0.2272,
+      "step": 2308
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3592607842724648e-07,
+      "loss": 0.2463,
+      "step": 2309
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3374744297378839e-07,
+      "loss": 0.2404,
+      "step": 2310
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3158629100124e-07,
+      "loss": 0.2423,
+      "step": 2311
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2944262633927007e-07,
+      "loss": 0.2341,
+      "step": 2312
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2731645278655448e-07,
+      "loss": 0.2754,
+      "step": 2313
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2520777411077822e-07,
+      "loss": 0.2293,
+      "step": 2314
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.2041,
+      "step": 2315
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2104291630576136e-07,
+      "loss": 0.2103,
+      "step": 2316
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1898674455685045e-07,
+      "loss": 0.2272,
+      "step": 2317
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1694808244552824e-07,
+      "loss": 0.2036,
+      "step": 2318
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1492693358440276e-07,
+      "loss": 0.2583,
+      "step": 2319
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1292330155505016e-07,
+      "loss": 0.2214,
+      "step": 2320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.109371899080025e-07,
+      "loss": 0.2656,
+      "step": 2321
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0896860216274563e-07,
+      "loss": 0.2102,
+      "step": 2322
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0701754180771462e-07,
+      "loss": 0.2487,
+      "step": 2323
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0508401230028387e-07,
+      "loss": 0.2849,
+      "step": 2324
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0316801706676038e-07,
+      "loss": 0.1942,
+      "step": 2325
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0126955950238271e-07,
+      "loss": 0.2409,
+      "step": 2326
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.93886429713098e-08,
+      "loss": 0.2169,
+      "step": 2327
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.752527080661655e-08,
+      "loss": 0.2442,
+      "step": 2328
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.56794463102917e-08,
+      "loss": 0.2263,
+      "step": 2329
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.38511727532232e-08,
+      "loss": 0.2038,
+      "step": 2330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.204045337520395e-08,
+      "loss": 0.3241,
+      "step": 2331
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.024729138491506e-08,
+      "loss": 0.1977,
+      "step": 2332
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.847168995992916e-08,
+      "loss": 0.2846,
+      "step": 2333
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.671365224669492e-08,
+      "loss": 0.2333,
+      "step": 2334
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.497318136054477e-08,
+      "loss": 0.2191,
+      "step": 2335
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.325028038567606e-08,
+      "loss": 0.2557,
+      "step": 2336
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.154495237515436e-08,
+      "loss": 0.1953,
+      "step": 2337
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.985720035090239e-08,
+      "loss": 0.2103,
+      "step": 2338
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.818702730370109e-08,
+      "loss": 0.2481,
+      "step": 2339
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.653443619317747e-08,
+      "loss": 0.2608,
+      "step": 2340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.489942994780452e-08,
+      "loss": 0.1832,
+      "step": 2341
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.328201146489244e-08,
+      "loss": 0.251,
+      "step": 2342
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.16821836105841e-08,
+      "loss": 0.2721,
+      "step": 2343
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.009994921985508e-08,
+      "loss": 0.2149,
+      "step": 2344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.853531109650147e-08,
+      "loss": 0.1865,
+      "step": 2345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.698827201313762e-08,
+      "loss": 0.1614,
+      "step": 2346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.545883471119174e-08,
+      "loss": 0.2759,
+      "step": 2347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.394700190090252e-08,
+      "loss": 0.235,
+      "step": 2348
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.245277626131142e-08,
+      "loss": 0.2051,
+      "step": 2349
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.097616044025922e-08,
+      "loss": 0.2769,
+      "step": 2350
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.951715705437955e-08,
+      "loss": 0.2696,
+      "step": 2351
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.807576868909981e-08,
+      "loss": 0.2615,
+      "step": 2352
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.665199789862907e-08,
+      "loss": 0.3078,
+      "step": 2353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5245847205959156e-08,
+      "loss": 0.2206,
+      "step": 2354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3857319102857967e-08,
+      "loss": 0.2809,
+      "step": 2355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.248641604986393e-08,
+      "loss": 0.2189,
+      "step": 2356
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.113314047628493e-08,
+      "loss": 0.1716,
+      "step": 2357
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.979749478019158e-08,
+      "loss": 0.2309,
+      "step": 2358
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8479481328413955e-08,
+      "loss": 0.2271,
+      "step": 2359
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7179102456533786e-08,
+      "loss": 0.1993,
+      "step": 2360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.589636046888779e-08,
+      "loss": 0.2607,
+      "step": 2361
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.463125763855769e-08,
+      "loss": 0.2393,
+      "step": 2362
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.338379620736577e-08,
+      "loss": 0.2625,
+      "step": 2363
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2153978385875985e-08,
+      "loss": 0.2392,
+      "step": 2364
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.094180635338396e-08,
+      "loss": 0.2291,
+      "step": 2365
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.974728225791924e-08,
+      "loss": 0.2086,
+      "step": 2366
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8570408216236366e-08,
+      "loss": 0.2187,
+      "step": 2367
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.741118631381269e-08,
+      "loss": 0.2433,
+      "step": 2368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.626961860484723e-08,
+      "loss": 0.236,
+      "step": 2369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.514570711225296e-08,
+      "loss": 0.2713,
+      "step": 2370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.403945382765561e-08,
+      "loss": 0.2253,
+      "step": 2371
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.295086071139153e-08,
+      "loss": 0.2635,
+      "step": 2372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.187992969249876e-08,
+      "loss": 0.2139,
+      "step": 2373
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.082666266872036e-08,
+      "loss": 0.2821,
+      "step": 2374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9791061506496686e-08,
+      "loss": 0.2881,
+      "step": 2375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8773128040964214e-08,
+      "loss": 0.2413,
+      "step": 2376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7772864075950036e-08,
+      "loss": 0.2229,
+      "step": 2377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6790271383970723e-08,
+      "loss": 0.2572,
+      "step": 2378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5825351706227908e-08,
+      "loss": 0.2878,
+      "step": 2379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4878106752607157e-08,
+      "loss": 0.2954,
+      "step": 2380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3948538201672423e-08,
+      "loss": 0.3295,
+      "step": 2381
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.303664770066494e-08,
+      "loss": 0.2305,
+      "step": 2382
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2142436865499884e-08,
+      "loss": 0.2351,
+      "step": 2383
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1265907280759725e-08,
+      "loss": 0.2028,
+      "step": 2384
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.040706049970087e-08,
+      "loss": 0.2631,
+      "step": 2385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9565898044239252e-08,
+      "loss": 0.2132,
+      "step": 2386
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8742421404956968e-08,
+      "loss": 0.2103,
+      "step": 2387
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7936632041094527e-08,
+      "loss": 0.2305,
+      "step": 2388
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7148531380550836e-08,
+      "loss": 0.3106,
+      "step": 2389
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6378120819877665e-08,
+      "loss": 0.1841,
+      "step": 2390
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.562540172427962e-08,
+      "loss": 0.2225,
+      "step": 2391
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4890375427613069e-08,
+      "loss": 0.2387,
+      "step": 2392
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4173043232380557e-08,
+      "loss": 0.2072,
+      "step": 2393
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3473406409728607e-08,
+      "loss": 0.2792,
+      "step": 2394
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2791466199447711e-08,
+      "loss": 0.2875,
+      "step": 2395
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2127223809970112e-08,
+      "loss": 0.2378,
+      "step": 2396
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1480680418365364e-08,
+      "loss": 0.2001,
+      "step": 2397
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0851837170340329e-08,
+      "loss": 0.2492,
+      "step": 2398
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0240695180234739e-08,
+      "loss": 0.3008,
+      "step": 2399
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.647255531023415e-09,
+      "loss": 0.2488,
+      "step": 2400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.071519274308494e-09,
+      "loss": 0.2591,
+      "step": 2401
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.513487430324985e-09,
+      "loss": 0.2423,
+      "step": 2402
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.973160987931883e-09,
+      "loss": 0.2322,
+      "step": 2403
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.450540904612169e-09,
+      "loss": 0.2481,
+      "step": 2404
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.945628106477254e-09,
+      "loss": 0.2194,
+      "step": 2405
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4584234882547616e-09,
+      "loss": 0.2717,
+      "step": 2406
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.988927913295195e-09,
+      "loss": 0.2822,
+      "step": 2407
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.537142213569713e-09,
+      "loss": 0.246,
+      "step": 2408
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1030671896623585e-09,
+      "loss": 0.2922,
+      "step": 2409
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6867036107767215e-09,
+      "loss": 0.2088,
+      "step": 2410
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.288052214727057e-09,
+      "loss": 0.2855,
+      "step": 2411
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.907113707946053e-09,
+      "loss": 0.1692,
+      "step": 2412
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5438887654737355e-09,
+      "loss": 0.2182,
+      "step": 2413
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.198378030963012e-09,
+      "loss": 0.208,
+      "step": 2414
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.870582116676346e-09,
+      "loss": 0.2157,
+      "step": 2415
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5605016034813134e-09,
+      "loss": 0.2076,
+      "step": 2416
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.268137040859486e-09,
+      "loss": 0.2028,
+      "step": 2417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.993488946891997e-09,
+      "loss": 0.307,
+      "step": 2418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7365578082706447e-09,
+      "loss": 0.3018,
+      "step": 2419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4973440802890094e-09,
+      "loss": 0.2639,
+      "step": 2420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.275848186845785e-09,
+      "loss": 0.2743,
+      "step": 2421
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0720705204414483e-09,
+      "loss": 0.2223,
+      "step": 2422
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.860114421826993e-10,
+      "loss": 0.2473,
+      "step": 2423
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.176712817724696e-10,
+      "loss": 0.3147,
+      "step": 2424
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.670503375188041e-10,
+      "loss": 0.2396,
+      "step": 2425
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3414887633042023e-10,
+      "loss": 0.263,
+      "step": 2426
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1896713371337706e-10,
+      "loss": 0.3078,
+      "step": 2427
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2150531377551633e-10,
+      "loss": 0.2524,
+      "step": 2428
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4176358922535216e-10,
+      "loss": 0.2462,
+      "step": 2429
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.97421013687405e-11,
+      "loss": 0.2789,
+      "step": 2430
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.544096010998921e-11,
+      "loss": 0.2202,
+      "step": 2431
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.860243952968361e-12,
+      "loss": 0.2295,
+      "step": 2432
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.294,
+      "step": 2433
+    },
+    {
+      "epoch": 3.0,
+      "step": 2433,
+      "total_flos": 1.9269628383880806e+17,
+      "train_loss": 0.606949764915175,
+      "train_runtime": 9820.3372,
+      "train_samples_per_second": 15.868,
+      "train_steps_per_second": 0.248
+    }
+  ],
+  "max_steps": 2433,
+  "num_train_epochs": 3,
+  "total_flos": 1.9269628383880806e+17,
+  "trial_name": null,
+  "trial_params": null
+}