diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,444 +2,26364 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 10.0,
-  "global_step": 70,
+  "global_step": 4390,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.14,
-      "learning_rate": 4.928571428571429e-05,
-      "loss": 3.0215,
+      "epoch": 0.0,
+      "learning_rate": 4.998861047835991e-05,
+      "loss": 11.7092,
       "step": 1
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.8571428571428576e-05,
-      "loss": 3.396,
+      "epoch": 0.0,
+      "learning_rate": 4.9977220956719826e-05,
+      "loss": 9.6548,
       "step": 2
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.785714285714286e-05,
-      "loss": 2.7081,
+      "epoch": 0.01,
+      "learning_rate": 4.996583143507973e-05,
+      "loss": 8.4051,
       "step": 3
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 4.714285714285714e-05,
-      "loss": 2.81,
+      "epoch": 0.01,
+      "learning_rate": 4.9954441913439635e-05,
+      "loss": 9.5686,
       "step": 4
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.642857142857143e-05,
-      "loss": 2.7994,
+      "epoch": 0.01,
+      "learning_rate": 4.9943052391799543e-05,
+      "loss": 8.9303,
       "step": 5
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 4.5714285714285716e-05,
-      "loss": 2.5904,
+      "epoch": 0.01,
+      "learning_rate": 4.993166287015946e-05,
+      "loss": 7.7487,
       "step": 6
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 4.5e-05,
-      "loss": 3.1253,
+      "epoch": 0.02,
+      "learning_rate": 4.992027334851937e-05,
+      "loss": 6.5083,
       "step": 7
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 4.428571428571428e-05,
-      "loss": 2.7843,
+      "epoch": 0.02,
+      "learning_rate": 4.990888382687927e-05,
+      "loss": 8.1923,
       "step": 8
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 4.3571428571428576e-05,
-      "loss": 2.6066,
+      "epoch": 0.02,
+      "learning_rate": 4.989749430523918e-05,
+      "loss": 8.0454,
       "step": 9
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 4.2857142857142856e-05,
-      "loss": 2.6174,
+      "epoch": 0.02,
+      "learning_rate": 4.988610478359909e-05,
+      "loss": 7.6701,
       "step": 10
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 4.214285714285714e-05,
-      "loss": 2.619,
+      "epoch": 0.03,
+      "learning_rate": 4.9874715261959e-05,
+      "loss": 7.1795,
       "step": 11
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 4.1428571428571437e-05,
-      "loss": 2.7048,
+      "epoch": 0.03,
+      "learning_rate": 4.9863325740318914e-05,
+      "loss": 7.6715,
       "step": 12
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 4.0714285714285717e-05,
-      "loss": 2.5045,
+      "epoch": 0.03,
+      "learning_rate": 4.9851936218678816e-05,
+      "loss": 7.9991,
       "step": 13
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 4e-05,
-      "loss": 2.6195,
+      "epoch": 0.03,
+      "learning_rate": 4.9840546697038724e-05,
+      "loss": 6.5221,
       "step": 14
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 3.928571428571429e-05,
-      "loss": 2.276,
+      "epoch": 0.03,
+      "learning_rate": 4.982915717539863e-05,
+      "loss": 7.3411,
       "step": 15
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 3.857142857142858e-05,
-      "loss": 2.4503,
+      "epoch": 0.04,
+      "learning_rate": 4.981776765375855e-05,
+      "loss": 5.8229,
       "step": 16
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 3.785714285714286e-05,
-      "loss": 2.5767,
+      "epoch": 0.04,
+      "learning_rate": 4.9806378132118455e-05,
+      "loss": 6.8379,
       "step": 17
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 3.7142857142857143e-05,
-      "loss": 2.4256,
+      "epoch": 0.04,
+      "learning_rate": 4.9794988610478364e-05,
+      "loss": 7.0664,
       "step": 18
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 3.642857142857143e-05,
-      "loss": 2.6956,
+      "epoch": 0.04,
+      "learning_rate": 4.978359908883827e-05,
+      "loss": 7.0716,
       "step": 19
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 3.571428571428572e-05,
-      "loss": 2.5666,
+      "epoch": 0.05,
+      "learning_rate": 4.977220956719818e-05,
+      "loss": 8.7517,
       "step": 20
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 3.5e-05,
-      "loss": 2.5238,
+      "epoch": 0.05,
+      "learning_rate": 4.976082004555809e-05,
+      "loss": 6.6135,
       "step": 21
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 3.428571428571429e-05,
-      "loss": 2.4623,
+      "epoch": 0.05,
+      "learning_rate": 4.9749430523918e-05,
+      "loss": 6.1742,
       "step": 22
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 3.357142857142857e-05,
-      "loss": 2.5452,
+      "epoch": 0.05,
+      "learning_rate": 4.9738041002277905e-05,
+      "loss": 8.0588,
       "step": 23
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 3.285714285714286e-05,
-      "loss": 2.4905,
+      "epoch": 0.05,
+      "learning_rate": 4.972665148063781e-05,
+      "loss": 7.9152,
       "step": 24
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 3.2142857142857144e-05,
-      "loss": 2.5106,
+      "epoch": 0.06,
+      "learning_rate": 4.971526195899772e-05,
+      "loss": 5.1515,
       "step": 25
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 3.142857142857143e-05,
-      "loss": 2.2134,
+      "epoch": 0.06,
+      "learning_rate": 4.9703872437357636e-05,
+      "loss": 5.6302,
       "step": 26
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 3.071428571428572e-05,
-      "loss": 2.3214,
+      "epoch": 0.06,
+      "learning_rate": 4.9692482915717544e-05,
+      "loss": 9.6288,
       "step": 27
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 3e-05,
-      "loss": 1.8621,
+      "epoch": 0.06,
+      "learning_rate": 4.968109339407745e-05,
+      "loss": 5.228,
       "step": 28
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 2.9285714285714288e-05,
-      "loss": 2.4828,
+      "epoch": 0.07,
+      "learning_rate": 4.966970387243736e-05,
+      "loss": 6.7267,
       "step": 29
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 2.857142857142857e-05,
-      "loss": 2.3785,
+      "epoch": 0.07,
+      "learning_rate": 4.965831435079727e-05,
+      "loss": 5.9025,
       "step": 30
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 2.785714285714286e-05,
-      "loss": 2.2715,
+      "epoch": 0.07,
+      "learning_rate": 4.964692482915718e-05,
+      "loss": 6.6566,
       "step": 31
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 2.714285714285714e-05,
-      "loss": 2.1797,
+      "epoch": 0.07,
+      "learning_rate": 4.963553530751709e-05,
+      "loss": 6.449,
       "step": 32
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 2.642857142857143e-05,
-      "loss": 2.22,
+      "epoch": 0.08,
+      "learning_rate": 4.9624145785876994e-05,
+      "loss": 5.6756,
       "step": 33
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 2.5714285714285714e-05,
-      "loss": 2.4549,
+      "epoch": 0.08,
+      "learning_rate": 4.96127562642369e-05,
+      "loss": 6.2954,
       "step": 34
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.5e-05,
-      "loss": 2.207,
+      "epoch": 0.08,
+      "learning_rate": 4.960136674259681e-05,
+      "loss": 6.7187,
       "step": 35
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 2.4285714285714288e-05,
-      "loss": 2.3845,
+      "epoch": 0.08,
+      "learning_rate": 4.9589977220956725e-05,
+      "loss": 7.1454,
       "step": 36
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 2.357142857142857e-05,
-      "loss": 2.2862,
+      "epoch": 0.08,
+      "learning_rate": 4.957858769931663e-05,
+      "loss": 5.7555,
       "step": 37
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 2.2857142857142858e-05,
-      "loss": 2.2408,
+      "epoch": 0.09,
+      "learning_rate": 4.956719817767654e-05,
+      "loss": 6.1015,
       "step": 38
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 2.214285714285714e-05,
-      "loss": 2.4339,
+      "epoch": 0.09,
+      "learning_rate": 4.955580865603645e-05,
+      "loss": 6.5185,
       "step": 39
     },
     {
-      "epoch": 5.71,
-      "learning_rate": 2.1428571428571428e-05,
-      "loss": 1.9903,
+      "epoch": 0.09,
+      "learning_rate": 4.954441913439636e-05,
+      "loss": 4.8291,
       "step": 40
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 2.0714285714285718e-05,
-      "loss": 2.2583,
+      "epoch": 0.09,
+      "learning_rate": 4.9533029612756266e-05,
+      "loss": 5.2706,
       "step": 41
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 2e-05,
-      "loss": 2.6031,
+      "epoch": 0.1,
+      "learning_rate": 4.952164009111618e-05,
+      "loss": 7.173,
       "step": 42
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 1.928571428571429e-05,
-      "loss": 2.1461,
+      "epoch": 0.1,
+      "learning_rate": 4.951025056947608e-05,
+      "loss": 8.1812,
       "step": 43
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 1.8571428571428572e-05,
-      "loss": 2.1812,
+      "epoch": 0.1,
+      "learning_rate": 4.949886104783599e-05,
+      "loss": 7.6324,
       "step": 44
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 1.785714285714286e-05,
-      "loss": 2.1179,
+      "epoch": 0.1,
+      "learning_rate": 4.94874715261959e-05,
+      "loss": 7.6471,
       "step": 45
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 1.7142857142857145e-05,
-      "loss": 2.3383,
+      "epoch": 0.1,
+      "learning_rate": 4.9476082004555814e-05,
+      "loss": 6.6633,
       "step": 46
     },
     {
-      "epoch": 6.71,
-      "learning_rate": 1.642857142857143e-05,
-      "loss": 2.3931,
+      "epoch": 0.11,
+      "learning_rate": 4.946469248291572e-05,
+      "loss": 6.1763,
       "step": 47
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 1.5714285714285715e-05,
-      "loss": 2.1934,
+      "epoch": 0.11,
+      "learning_rate": 4.945330296127563e-05,
+      "loss": 4.8749,
       "step": 48
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 1.5e-05,
-      "loss": 2.189,
+      "epoch": 0.11,
+      "learning_rate": 4.944191343963554e-05,
+      "loss": 7.3018,
       "step": 49
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 1.4285714285714285e-05,
-      "loss": 2.0834,
+      "epoch": 0.11,
+      "learning_rate": 4.9430523917995447e-05,
+      "loss": 6.4854,
       "step": 50
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 1.357142857142857e-05,
-      "loss": 2.1371,
+      "epoch": 0.12,
+      "learning_rate": 4.9419134396355355e-05,
+      "loss": 5.4195,
       "step": 51
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 1.2857142857142857e-05,
-      "loss": 2.0772,
+      "epoch": 0.12,
+      "learning_rate": 4.940774487471526e-05,
+      "loss": 6.822,
       "step": 52
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 1.2142857142857144e-05,
-      "loss": 2.3063,
+      "epoch": 0.12,
+      "learning_rate": 4.939635535307517e-05,
+      "loss": 7.2982,
       "step": 53
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 1.1428571428571429e-05,
-      "loss": 2.2691,
+      "epoch": 0.12,
+      "learning_rate": 4.938496583143508e-05,
+      "loss": 6.6605,
       "step": 54
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 1.0714285714285714e-05,
-      "loss": 2.2412,
+      "epoch": 0.13,
+      "learning_rate": 4.937357630979499e-05,
+      "loss": 4.8833,
       "step": 55
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 1e-05,
-      "loss": 2.5235,
+      "epoch": 0.13,
+      "learning_rate": 4.93621867881549e-05,
+      "loss": 6.4155,
       "step": 56
     },
     {
-      "epoch": 8.14,
-      "learning_rate": 9.285714285714286e-06,
-      "loss": 2.239,
+      "epoch": 0.13,
+      "learning_rate": 4.935079726651481e-05,
+      "loss": 4.1882,
       "step": 57
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 8.571428571428573e-06,
-      "loss": 2.1908,
+      "epoch": 0.13,
+      "learning_rate": 4.933940774487472e-05,
+      "loss": 6.7254,
       "step": 58
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 7.857142857142858e-06,
-      "loss": 2.1594,
+      "epoch": 0.13,
+      "learning_rate": 4.932801822323463e-05,
+      "loss": 6.5771,
       "step": 59
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 7.142857142857143e-06,
-      "loss": 2.2055,
+      "epoch": 0.14,
+      "learning_rate": 4.9316628701594535e-05,
+      "loss": 5.0929,
       "step": 60
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 6.428571428571429e-06,
-      "loss": 1.975,
+      "epoch": 0.14,
+      "learning_rate": 4.9305239179954444e-05,
+      "loss": 8.5807,
       "step": 61
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 5.7142857142857145e-06,
-      "loss": 2.2883,
+      "epoch": 0.14,
+      "learning_rate": 4.929384965831435e-05,
+      "loss": 4.8718,
       "step": 62
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 5e-06,
-      "loss": 1.9735,
+      "epoch": 0.14,
+      "learning_rate": 4.928246013667426e-05,
+      "loss": 6.4467,
       "step": 63
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 4.285714285714286e-06,
-      "loss": 2.189,
+      "epoch": 0.15,
+      "learning_rate": 4.927107061503417e-05,
+      "loss": 5.6606,
       "step": 64
     },
     {
-      "epoch": 9.29,
-      "learning_rate": 3.5714285714285714e-06,
-      "loss": 2.3047,
+      "epoch": 0.15,
+      "learning_rate": 4.9259681093394076e-05,
+      "loss": 6.4872,
       "step": 65
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 1.908,
+      "epoch": 0.15,
+      "learning_rate": 4.924829157175399e-05,
+      "loss": 8.0231,
       "step": 66
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 2.142857142857143e-06,
-      "loss": 2.0436,
+      "epoch": 0.15,
+      "learning_rate": 4.92369020501139e-05,
+      "loss": 6.0967,
       "step": 67
     },
     {
-      "epoch": 9.71,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 2.2566,
+      "epoch": 0.15,
+      "learning_rate": 4.922551252847381e-05,
+      "loss": 6.8669,
       "step": 68
     },
     {
-      "epoch": 9.86,
-      "learning_rate": 7.142857142857143e-07,
-      "loss": 2.2008,
+      "epoch": 0.16,
+      "learning_rate": 4.9214123006833716e-05,
+      "loss": 6.5913,
       "step": 69
     },
     {
-      "epoch": 10.0,
-      "learning_rate": 0.0,
-      "loss": 2.3549,
+      "epoch": 0.16,
+      "learning_rate": 4.9202733485193624e-05,
+      "loss": 5.0972,
       "step": 70
     },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.919134396355353e-05,
+      "loss": 5.1875,
+      "step": 71
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.917995444191344e-05,
+      "loss": 5.143,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9168564920273356e-05,
+      "loss": 4.976,
+      "step": 73
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915717539863326e-05,
+      "loss": 6.2076,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9145785876993165e-05,
+      "loss": 6.3557,
+      "step": 75
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.913439635535308e-05,
+      "loss": 3.9665,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912300683371299e-05,
+      "loss": 6.9845,
+      "step": 77
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91116173120729e-05,
+      "loss": 5.5932,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9100227790432805e-05,
+      "loss": 8.17,
+      "step": 79
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908883826879271e-05,
+      "loss": 6.9583,
+      "step": 80
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907744874715262e-05,
+      "loss": 5.9453,
+      "step": 81
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.906605922551253e-05,
+      "loss": 6.1392,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9054669703872444e-05,
+      "loss": 5.7155,
+      "step": 83
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9043280182232346e-05,
+      "loss": 6.3088,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9031890660592254e-05,
+      "loss": 7.3777,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902050113895217e-05,
+      "loss": 6.2157,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.900911161731208e-05,
+      "loss": 5.0033,
+      "step": 87
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8997722095671985e-05,
+      "loss": 4.8096,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8986332574031894e-05,
+      "loss": 5.0981,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.89749430523918e-05,
+      "loss": 7.8127,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.896355353075171e-05,
+      "loss": 5.5104,
+      "step": 91
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.895216400911162e-05,
+      "loss": 4.2981,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.894077448747153e-05,
+      "loss": 5.3174,
+      "step": 93
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8929384965831435e-05,
+      "loss": 5.5535,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.891799544419134e-05,
+      "loss": 4.1204,
+      "step": 95
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.890660592255126e-05,
+      "loss": 5.5789,
+      "step": 96
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8895216400911166e-05,
+      "loss": 6.7247,
+      "step": 97
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8883826879271074e-05,
+      "loss": 7.0594,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8872437357630976e-05,
+      "loss": 5.1639,
+      "step": 99
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.886104783599089e-05,
+      "loss": 6.2181,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.88496583143508e-05,
+      "loss": 4.1885,
+      "step": 101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.883826879271071e-05,
+      "loss": 5.3262,
+      "step": 102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.882687927107062e-05,
+      "loss": 7.0901,
+      "step": 103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8815489749430524e-05,
+      "loss": 6.3874,
+      "step": 104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.880410022779043e-05,
+      "loss": 5.7896,
+      "step": 105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.879271070615035e-05,
+      "loss": 5.2289,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8781321184510255e-05,
+      "loss": 4.9152,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.876993166287016e-05,
+      "loss": 4.0898,
+      "step": 108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.875854214123007e-05,
+      "loss": 5.399,
+      "step": 109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.874715261958998e-05,
+      "loss": 5.256,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.873576309794989e-05,
+      "loss": 7.3301,
+      "step": 111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8724373576309796e-05,
+      "loss": 5.971,
+      "step": 112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.871298405466971e-05,
+      "loss": 5.5653,
+      "step": 113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.870159453302961e-05,
+      "loss": 6.7108,
+      "step": 114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.869020501138952e-05,
+      "loss": 5.104,
+      "step": 115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8678815489749436e-05,
+      "loss": 3.1726,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8667425968109344e-05,
+      "loss": 6.0224,
+      "step": 117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.865603644646925e-05,
+      "loss": 6.4886,
+      "step": 118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.864464692482916e-05,
+      "loss": 5.3235,
+      "step": 119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.863325740318907e-05,
+      "loss": 4.7159,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8621867881548977e-05,
+      "loss": 6.0486,
+      "step": 121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8610478359908885e-05,
+      "loss": 4.8432,
+      "step": 122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.85990888382688e-05,
+      "loss": 3.8479,
+      "step": 123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.85876993166287e-05,
+      "loss": 5.1927,
+      "step": 124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.857630979498861e-05,
+      "loss": 4.9575,
+      "step": 125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8564920273348524e-05,
+      "loss": 6.6181,
+      "step": 126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.855353075170843e-05,
+      "loss": 3.5296,
+      "step": 127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.854214123006834e-05,
+      "loss": 5.9652,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.853075170842825e-05,
+      "loss": 3.7124,
+      "step": 129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.851936218678816e-05,
+      "loss": 4.6301,
+      "step": 130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8507972665148065e-05,
+      "loss": 4.6318,
+      "step": 131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8496583143507974e-05,
+      "loss": 4.9169,
+      "step": 132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.848519362186789e-05,
+      "loss": 5.6412,
+      "step": 133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847380410022779e-05,
+      "loss": 5.703,
+      "step": 134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.84624145785877e-05,
+      "loss": 5.5157,
+      "step": 135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.845102505694761e-05,
+      "loss": 5.302,
+      "step": 136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.843963553530752e-05,
+      "loss": 8.4144,
+      "step": 137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.842824601366743e-05,
+      "loss": 4.4277,
+      "step": 138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.841685649202734e-05,
+      "loss": 3.3021,
+      "step": 139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8405466970387246e-05,
+      "loss": 5.3697,
+      "step": 140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8394077448747154e-05,
+      "loss": 6.1515,
+      "step": 141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.838268792710706e-05,
+      "loss": 5.0783,
+      "step": 142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837129840546698e-05,
+      "loss": 5.6476,
+      "step": 143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.835990888382688e-05,
+      "loss": 6.0066,
+      "step": 144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.834851936218679e-05,
+      "loss": 5.6522,
+      "step": 145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83371298405467e-05,
+      "loss": 6.2684,
+      "step": 146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.832574031890661e-05,
+      "loss": 6.2464,
+      "step": 147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.831435079726652e-05,
+      "loss": 8.025,
+      "step": 148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.830296127562643e-05,
+      "loss": 5.7292,
+      "step": 149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8291571753986335e-05,
+      "loss": 6.6286,
+      "step": 150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.828018223234624e-05,
+      "loss": 5.2295,
+      "step": 151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.826879271070615e-05,
+      "loss": 3.3221,
+      "step": 152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8257403189066066e-05,
+      "loss": 5.561,
+      "step": 153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.824601366742597e-05,
+      "loss": 4.0204,
+      "step": 154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8234624145785876e-05,
+      "loss": 5.223,
+      "step": 155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8223234624145784e-05,
+      "loss": 5.8434,
+      "step": 156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82118451025057e-05,
+      "loss": 4.7687,
+      "step": 157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820045558086561e-05,
+      "loss": 5.6896,
+      "step": 158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8189066059225516e-05,
+      "loss": 4.4732,
+      "step": 159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8177676537585424e-05,
+      "loss": 5.0196,
+      "step": 160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.816628701594533e-05,
+      "loss": 5.2394,
+      "step": 161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815489749430524e-05,
+      "loss": 4.4934,
+      "step": 162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8143507972665155e-05,
+      "loss": 3.838,
+      "step": 163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813211845102506e-05,
+      "loss": 4.5055,
+      "step": 164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8120728929384965e-05,
+      "loss": 6.3589,
+      "step": 165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.810933940774487e-05,
+      "loss": 4.5191,
+      "step": 166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.809794988610479e-05,
+      "loss": 5.4652,
+      "step": 167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8086560364464696e-05,
+      "loss": 5.786,
+      "step": 168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8075170842824604e-05,
+      "loss": 5.3748,
+      "step": 169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.806378132118451e-05,
+      "loss": 5.2441,
+      "step": 170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.805239179954442e-05,
+      "loss": 8.0993,
+      "step": 171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.804100227790433e-05,
+      "loss": 6.6524,
+      "step": 172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8029612756264244e-05,
+      "loss": 5.0897,
+      "step": 173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.801822323462415e-05,
+      "loss": 5.6209,
+      "step": 174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8006833712984054e-05,
+      "loss": 6.0429,
+      "step": 175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.799544419134396e-05,
+      "loss": 5.6825,
+      "step": 176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.798405466970388e-05,
+      "loss": 4.1911,
+      "step": 177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7972665148063785e-05,
+      "loss": 6.3678,
+      "step": 178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796127562642369e-05,
+      "loss": 5.4064,
+      "step": 179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.79498861047836e-05,
+      "loss": 6.096,
+      "step": 180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.793849658314351e-05,
+      "loss": 6.7477,
+      "step": 181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.792710706150342e-05,
+      "loss": 5.9499,
+      "step": 182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.791571753986333e-05,
+      "loss": 5.4382,
+      "step": 183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.790432801822324e-05,
+      "loss": 6.2821,
+      "step": 184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.789293849658314e-05,
+      "loss": 4.3839,
+      "step": 185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.788154897494305e-05,
+      "loss": 6.8907,
+      "step": 186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7870159453302966e-05,
+      "loss": 4.5461,
+      "step": 187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7858769931662874e-05,
+      "loss": 6.7334,
+      "step": 188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.784738041002278e-05,
+      "loss": 4.8625,
+      "step": 189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.783599088838269e-05,
+      "loss": 5.446,
+      "step": 190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78246013667426e-05,
+      "loss": 4.1151,
+      "step": 191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.781321184510251e-05,
+      "loss": 3.6497,
+      "step": 192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.780182232346242e-05,
+      "loss": 5.479,
+      "step": 193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.779043280182233e-05,
+      "loss": 5.2026,
+      "step": 194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.777904328018223e-05,
+      "loss": 5.0487,
+      "step": 195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.776765375854214e-05,
+      "loss": 5.3173,
+      "step": 196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7756264236902054e-05,
+      "loss": 5.2775,
+      "step": 197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.774487471526196e-05,
+      "loss": 5.9495,
+      "step": 198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.773348519362187e-05,
+      "loss": 5.4733,
+      "step": 199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772209567198178e-05,
+      "loss": 5.0203,
+      "step": 200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.771070615034169e-05,
+      "loss": 5.6938,
+      "step": 201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7699316628701595e-05,
+      "loss": 4.6525,
+      "step": 202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7687927107061504e-05,
+      "loss": 4.2312,
+      "step": 203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767653758542142e-05,
+      "loss": 5.7577,
+      "step": 204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.766514806378132e-05,
+      "loss": 5.9478,
+      "step": 205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.765375854214123e-05,
+      "loss": 4.3839,
+      "step": 206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.764236902050114e-05,
+      "loss": 4.9267,
+      "step": 207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.763097949886105e-05,
+      "loss": 4.1385,
+      "step": 208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.761958997722096e-05,
+      "loss": 2.5577,
+      "step": 209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.760820045558087e-05,
+      "loss": 4.1588,
+      "step": 210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7596810933940776e-05,
+      "loss": 3.5204,
+      "step": 211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7585421412300684e-05,
+      "loss": 3.791,
+      "step": 212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757403189066059e-05,
+      "loss": 6.5942,
+      "step": 213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.756264236902051e-05,
+      "loss": 5.7704,
+      "step": 214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.755125284738041e-05,
+      "loss": 6.9098,
+      "step": 215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.753986332574032e-05,
+      "loss": 3.6208,
+      "step": 216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.752847380410023e-05,
+      "loss": 4.7234,
+      "step": 217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.751708428246014e-05,
+      "loss": 3.2192,
+      "step": 218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.750569476082005e-05,
+      "loss": 4.6995,
+      "step": 219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.749430523917996e-05,
+      "loss": 5.2337,
+      "step": 220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7482915717539865e-05,
+      "loss": 5.6331,
+      "step": 221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747152619589977e-05,
+      "loss": 4.4731,
+      "step": 222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.746013667425968e-05,
+      "loss": 5.0083,
+      "step": 223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7448747152619596e-05,
+      "loss": 4.3566,
+      "step": 224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.74373576309795e-05,
+      "loss": 5.6376,
+      "step": 225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7425968109339406e-05,
+      "loss": 4.5897,
+      "step": 226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.741457858769932e-05,
+      "loss": 6.7104,
+      "step": 227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.740318906605923e-05,
+      "loss": 2.9332,
+      "step": 228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.739179954441914e-05,
+      "loss": 5.1857,
+      "step": 229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7380410022779046e-05,
+      "loss": 4.8685,
+      "step": 230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7369020501138954e-05,
+      "loss": 5.3684,
+      "step": 231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735763097949886e-05,
+      "loss": 4.9177,
+      "step": 232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734624145785877e-05,
+      "loss": 5.3389,
+      "step": 233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7334851936218685e-05,
+      "loss": 6.4771,
+      "step": 234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7323462414578587e-05,
+      "loss": 5.9667,
+      "step": 235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7312072892938495e-05,
+      "loss": 3.7441,
+      "step": 236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.730068337129841e-05,
+      "loss": 4.7473,
+      "step": 237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.728929384965832e-05,
+      "loss": 6.4161,
+      "step": 238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7277904328018226e-05,
+      "loss": 6.6409,
+      "step": 239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7266514806378134e-05,
+      "loss": 3.5049,
+      "step": 240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.725512528473804e-05,
+      "loss": 5.2361,
+      "step": 241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724373576309795e-05,
+      "loss": 5.0304,
+      "step": 242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723234624145786e-05,
+      "loss": 8.1464,
+      "step": 243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7220956719817774e-05,
+      "loss": 4.506,
+      "step": 244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.720956719817768e-05,
+      "loss": 4.4911,
+      "step": 245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7198177676537584e-05,
+      "loss": 5.01,
+      "step": 246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.71867881548975e-05,
+      "loss": 5.0284,
+      "step": 247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.717539863325741e-05,
+      "loss": 4.7329,
+      "step": 248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7164009111617315e-05,
+      "loss": 4.8994,
+      "step": 249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.715261958997722e-05,
+      "loss": 8.3168,
+      "step": 250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.714123006833713e-05,
+      "loss": 5.016,
+      "step": 251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.712984054669704e-05,
+      "loss": 2.2203,
+      "step": 252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.711845102505695e-05,
+      "loss": 5.1958,
+      "step": 253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.710706150341686e-05,
+      "loss": 4.0275,
+      "step": 254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.709567198177677e-05,
+      "loss": 3.1182,
+      "step": 255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.708428246013667e-05,
+      "loss": 4.6726,
+      "step": 256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707289293849659e-05,
+      "loss": 6.3335,
+      "step": 257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7061503416856496e-05,
+      "loss": 3.8147,
+      "step": 258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7050113895216404e-05,
+      "loss": 6.8395,
+      "step": 259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.703872437357631e-05,
+      "loss": 3.974,
+      "step": 260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.702733485193622e-05,
+      "loss": 4.0756,
+      "step": 261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.701594533029613e-05,
+      "loss": 3.5896,
+      "step": 262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.700455580865604e-05,
+      "loss": 4.5518,
+      "step": 263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.699316628701595e-05,
+      "loss": 5.341,
+      "step": 264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.698177676537586e-05,
+      "loss": 5.4774,
+      "step": 265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697038724373576e-05,
+      "loss": 5.5283,
+      "step": 266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6958997722095676e-05,
+      "loss": 5.1392,
+      "step": 267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6947608200455584e-05,
+      "loss": 4.9303,
+      "step": 268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.693621867881549e-05,
+      "loss": 6.1368,
+      "step": 269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69248291571754e-05,
+      "loss": 6.696,
+      "step": 270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.691343963553531e-05,
+      "loss": 6.4681,
+      "step": 271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690205011389522e-05,
+      "loss": 4.2989,
+      "step": 272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6890660592255125e-05,
+      "loss": 5.5542,
+      "step": 273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.687927107061504e-05,
+      "loss": 5.3746,
+      "step": 274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686788154897495e-05,
+      "loss": 4.993,
+      "step": 275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685649202733485e-05,
+      "loss": 4.3661,
+      "step": 276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6845102505694765e-05,
+      "loss": 4.5313,
+      "step": 277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683371298405467e-05,
+      "loss": 5.2945,
+      "step": 278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682232346241458e-05,
+      "loss": 3.9437,
+      "step": 279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.681093394077449e-05,
+      "loss": 5.0809,
+      "step": 280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.67995444191344e-05,
+      "loss": 6.2962,
+      "step": 281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6788154897494306e-05,
+      "loss": 5.2766,
+      "step": 282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6776765375854214e-05,
+      "loss": 5.0507,
+      "step": 283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.676537585421413e-05,
+      "loss": 4.5777,
+      "step": 284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.675398633257404e-05,
+      "loss": 3.7641,
+      "step": 285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.674259681093394e-05,
+      "loss": 5.3443,
+      "step": 286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6731207289293854e-05,
+      "loss": 5.9211,
+      "step": 287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.671981776765376e-05,
+      "loss": 3.986,
+      "step": 288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.670842824601367e-05,
+      "loss": 4.3643,
+      "step": 289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.669703872437358e-05,
+      "loss": 4.5729,
+      "step": 290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.668564920273349e-05,
+      "loss": 4.6944,
+      "step": 291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6674259681093395e-05,
+      "loss": 5.0941,
+      "step": 292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.66628701594533e-05,
+      "loss": 4.0779,
+      "step": 293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665148063781322e-05,
+      "loss": 3.944,
+      "step": 294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6640091116173126e-05,
+      "loss": 5.3061,
+      "step": 295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.662870159453303e-05,
+      "loss": 6.3,
+      "step": 296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6617312072892936e-05,
+      "loss": 5.2882,
+      "step": 297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660592255125285e-05,
+      "loss": 5.5267,
+      "step": 298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659453302961276e-05,
+      "loss": 5.719,
+      "step": 299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658314350797267e-05,
+      "loss": 5.5993,
+      "step": 300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6571753986332576e-05,
+      "loss": 4.954,
+      "step": 301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6560364464692484e-05,
+      "loss": 5.3487,
+      "step": 302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654897494305239e-05,
+      "loss": 5.3259,
+      "step": 303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653758542141231e-05,
+      "loss": 3.8235,
+      "step": 304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6526195899772215e-05,
+      "loss": 4.7128,
+      "step": 305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.651480637813212e-05,
+      "loss": 4.762,
+      "step": 306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6503416856492025e-05,
+      "loss": 3.7675,
+      "step": 307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.649202733485194e-05,
+      "loss": 5.7067,
+      "step": 308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.648063781321185e-05,
+      "loss": 4.1396,
+      "step": 309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6469248291571756e-05,
+      "loss": 6.6756,
+      "step": 310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6457858769931664e-05,
+      "loss": 3.1153,
+      "step": 311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.644646924829157e-05,
+      "loss": 5.5039,
+      "step": 312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.643507972665148e-05,
+      "loss": 5.1996,
+      "step": 313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6423690205011396e-05,
+      "loss": 3.0499,
+      "step": 314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6412300683371304e-05,
+      "loss": 4.1734,
+      "step": 315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6400911161731205e-05,
+      "loss": 5.7884,
+      "step": 316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6389521640091114e-05,
+      "loss": 2.9146,
+      "step": 317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.637813211845103e-05,
+      "loss": 5.2242,
+      "step": 318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.636674259681094e-05,
+      "loss": 3.9209,
+      "step": 319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6355353075170845e-05,
+      "loss": 4.5516,
+      "step": 320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634396355353075e-05,
+      "loss": 3.429,
+      "step": 321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.633257403189066e-05,
+      "loss": 5.0716,
+      "step": 322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632118451025057e-05,
+      "loss": 6.1228,
+      "step": 323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6309794988610485e-05,
+      "loss": 5.1168,
+      "step": 324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.629840546697039e-05,
+      "loss": 3.8001,
+      "step": 325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6287015945330294e-05,
+      "loss": 4.5676,
+      "step": 326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.62756264236902e-05,
+      "loss": 4.7286,
+      "step": 327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.626423690205012e-05,
+      "loss": 6.5656,
+      "step": 328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6252847380410026e-05,
+      "loss": 5.1669,
+      "step": 329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6241457858769934e-05,
+      "loss": 5.0266,
+      "step": 330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623006833712984e-05,
+      "loss": 4.6504,
+      "step": 331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.621867881548975e-05,
+      "loss": 3.7962,
+      "step": 332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.620728929384966e-05,
+      "loss": 3.4521,
+      "step": 333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6195899772209573e-05,
+      "loss": 3.905,
+      "step": 334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.618451025056948e-05,
+      "loss": 5.7641,
+      "step": 335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617312072892939e-05,
+      "loss": 3.3831,
+      "step": 336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616173120728929e-05,
+      "loss": 3.9597,
+      "step": 337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6150341685649206e-05,
+      "loss": 4.5485,
+      "step": 338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6138952164009115e-05,
+      "loss": 3.9907,
+      "step": 339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.612756264236902e-05,
+      "loss": 4.9508,
+      "step": 340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.611617312072893e-05,
+      "loss": 4.3987,
+      "step": 341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.610478359908884e-05,
+      "loss": 5.4862,
+      "step": 342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.609339407744875e-05,
+      "loss": 5.694,
+      "step": 343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6082004555808656e-05,
+      "loss": 4.0943,
+      "step": 344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607061503416857e-05,
+      "loss": 3.9401,
+      "step": 345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.605922551252848e-05,
+      "loss": 5.3028,
+      "step": 346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.604783599088838e-05,
+      "loss": 4.9781,
+      "step": 347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6036446469248295e-05,
+      "loss": 2.1853,
+      "step": 348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60250569476082e-05,
+      "loss": 5.7778,
+      "step": 349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.601366742596811e-05,
+      "loss": 5.1681,
+      "step": 350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600227790432802e-05,
+      "loss": 3.6947,
+      "step": 351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.599088838268793e-05,
+      "loss": 4.6537,
+      "step": 352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5979498861047836e-05,
+      "loss": 3.3695,
+      "step": 353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5968109339407744e-05,
+      "loss": 4.1743,
+      "step": 354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.595671981776766e-05,
+      "loss": 6.6035,
+      "step": 355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.594533029612757e-05,
+      "loss": 5.0347,
+      "step": 356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.593394077448747e-05,
+      "loss": 4.1491,
+      "step": 357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5922551252847384e-05,
+      "loss": 5.2064,
+      "step": 358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.591116173120729e-05,
+      "loss": 4.3117,
+      "step": 359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.58997722095672e-05,
+      "loss": 5.4789,
+      "step": 360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.588838268792711e-05,
+      "loss": 4.3252,
+      "step": 361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.587699316628702e-05,
+      "loss": 4.9125,
+      "step": 362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5865603644646925e-05,
+      "loss": 4.6831,
+      "step": 363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.585421412300683e-05,
+      "loss": 4.8532,
+      "step": 364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.584282460136675e-05,
+      "loss": 4.0289,
+      "step": 365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5831435079726656e-05,
+      "loss": 3.8643,
+      "step": 366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582004555808656e-05,
+      "loss": 2.8357,
+      "step": 367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.580865603644647e-05,
+      "loss": 4.1606,
+      "step": 368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.579726651480638e-05,
+      "loss": 5.1973,
+      "step": 369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.578587699316629e-05,
+      "loss": 4.1285,
+      "step": 370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.57744874715262e-05,
+      "loss": 3.9759,
+      "step": 371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5763097949886106e-05,
+      "loss": 3.7831,
+      "step": 372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5751708428246014e-05,
+      "loss": 6.37,
+      "step": 373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.574031890660592e-05,
+      "loss": 4.3329,
+      "step": 374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.572892938496584e-05,
+      "loss": 5.6949,
+      "step": 375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5717539863325745e-05,
+      "loss": 4.5326,
+      "step": 376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.570615034168565e-05,
+      "loss": 3.7581,
+      "step": 377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.569476082004556e-05,
+      "loss": 4.9291,
+      "step": 378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.568337129840547e-05,
+      "loss": 4.6771,
+      "step": 379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567198177676538e-05,
+      "loss": 4.4971,
+      "step": 380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5660592255125286e-05,
+      "loss": 4.8761,
+      "step": 381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5649202733485194e-05,
+      "loss": 4.3404,
+      "step": 382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56378132118451e-05,
+      "loss": 3.8512,
+      "step": 383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.562642369020501e-05,
+      "loss": 4.0373,
+      "step": 384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5615034168564926e-05,
+      "loss": 4.3845,
+      "step": 385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5603644646924834e-05,
+      "loss": 6.5267,
+      "step": 386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5592255125284735e-05,
+      "loss": 4.4642,
+      "step": 387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.558086560364465e-05,
+      "loss": 6.0062,
+      "step": 388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556947608200456e-05,
+      "loss": 5.3915,
+      "step": 389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.555808656036447e-05,
+      "loss": 4.9747,
+      "step": 390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5546697038724375e-05,
+      "loss": 3.8527,
+      "step": 391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.553530751708428e-05,
+      "loss": 6.1012,
+      "step": 392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552391799544419e-05,
+      "loss": 4.8889,
+      "step": 393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.55125284738041e-05,
+      "loss": 4.271,
+      "step": 394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5501138952164015e-05,
+      "loss": 4.3964,
+      "step": 395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.548974943052392e-05,
+      "loss": 3.1894,
+      "step": 396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5478359908883824e-05,
+      "loss": 5.1568,
+      "step": 397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546697038724374e-05,
+      "loss": 5.0407,
+      "step": 398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545558086560365e-05,
+      "loss": 5.1178,
+      "step": 399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5444191343963556e-05,
+      "loss": 4.9971,
+      "step": 400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5432801822323464e-05,
+      "loss": 2.9578,
+      "step": 401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542141230068337e-05,
+      "loss": 6.015,
+      "step": 402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.541002277904328e-05,
+      "loss": 4.5975,
+      "step": 403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.539863325740319e-05,
+      "loss": 4.2815,
+      "step": 404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5387243735763104e-05,
+      "loss": 6.1726,
+      "step": 405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.537585421412301e-05,
+      "loss": 4.0198,
+      "step": 406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.536446469248291e-05,
+      "loss": 4.3409,
+      "step": 407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.535307517084283e-05,
+      "loss": 2.9408,
+      "step": 408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5341685649202736e-05,
+      "loss": 5.8869,
+      "step": 409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5330296127562645e-05,
+      "loss": 3.237,
+      "step": 410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.531890660592255e-05,
+      "loss": 2.3125,
+      "step": 411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.530751708428246e-05,
+      "loss": 4.2522,
+      "step": 412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.529612756264237e-05,
+      "loss": 4.4033,
+      "step": 413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.528473804100228e-05,
+      "loss": 5.02,
+      "step": 414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527334851936219e-05,
+      "loss": 4.2451,
+      "step": 415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52619589977221e-05,
+      "loss": 5.5409,
+      "step": 416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5250569476082e-05,
+      "loss": 5.8935,
+      "step": 417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.523917995444192e-05,
+      "loss": 2.8071,
+      "step": 418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5227790432801825e-05,
+      "loss": 4.5095,
+      "step": 419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.521640091116173e-05,
+      "loss": 5.0845,
+      "step": 420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.520501138952164e-05,
+      "loss": 5.1359,
+      "step": 421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.519362186788155e-05,
+      "loss": 5.2268,
+      "step": 422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.518223234624146e-05,
+      "loss": 3.017,
+      "step": 423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5170842824601366e-05,
+      "loss": 4.3239,
+      "step": 424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.515945330296128e-05,
+      "loss": 3.8762,
+      "step": 425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.514806378132119e-05,
+      "loss": 3.0631,
+      "step": 426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.51366742596811e-05,
+      "loss": 4.144,
+      "step": 427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5125284738041006e-05,
+      "loss": 4.7862,
+      "step": 428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5113895216400914e-05,
+      "loss": 5.1028,
+      "step": 429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.510250569476082e-05,
+      "loss": 3.6585,
+      "step": 430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.509111617312073e-05,
+      "loss": 5.9563,
+      "step": 431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.507972665148064e-05,
+      "loss": 4.8223,
+      "step": 432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506833712984055e-05,
+      "loss": 4.1669,
+      "step": 433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5056947608200455e-05,
+      "loss": 3.9688,
+      "step": 434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.504555808656037e-05,
+      "loss": 4.9243,
+      "step": 435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503416856492028e-05,
+      "loss": 3.8482,
+      "step": 436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5022779043280186e-05,
+      "loss": 3.6732,
+      "step": 437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5011389521640095e-05,
+      "loss": 3.3671,
+      "step": 438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5e-05,
+      "loss": 5.5247,
+      "step": 439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.498861047835991e-05,
+      "loss": 5.1439,
+      "step": 440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.497722095671982e-05,
+      "loss": 4.0039,
+      "step": 441
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.496583143507973e-05,
+      "loss": 3.36,
+      "step": 442
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4954441913439636e-05,
+      "loss": 3.9027,
+      "step": 443
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4943052391799544e-05,
+      "loss": 3.8234,
+      "step": 444
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.493166287015946e-05,
+      "loss": 3.5136,
+      "step": 445
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492027334851937e-05,
+      "loss": 3.5471,
+      "step": 446
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4908883826879275e-05,
+      "loss": 3.3956,
+      "step": 447
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.489749430523918e-05,
+      "loss": 4.0521,
+      "step": 448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.488610478359909e-05,
+      "loss": 4.6453,
+      "step": 449
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4874715261959e-05,
+      "loss": 5.4109,
+      "step": 450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.486332574031891e-05,
+      "loss": 6.1848,
+      "step": 451
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4851936218678816e-05,
+      "loss": 2.4494,
+      "step": 452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4840546697038724e-05,
+      "loss": 4.802,
+      "step": 453
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.482915717539863e-05,
+      "loss": 4.0729,
+      "step": 454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481776765375855e-05,
+      "loss": 5.1832,
+      "step": 455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4806378132118456e-05,
+      "loss": 3.9975,
+      "step": 456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4794988610478364e-05,
+      "loss": 4.267,
+      "step": 457
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4783599088838266e-05,
+      "loss": 4.43,
+      "step": 458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477220956719818e-05,
+      "loss": 4.7333,
+      "step": 459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476082004555809e-05,
+      "loss": 3.5931,
+      "step": 460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4749430523918e-05,
+      "loss": 3.643,
+      "step": 461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4738041002277905e-05,
+      "loss": 5.3564,
+      "step": 462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.472665148063781e-05,
+      "loss": 3.2053,
+      "step": 463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.471526195899772e-05,
+      "loss": 4.8187,
+      "step": 464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4703872437357637e-05,
+      "loss": 2.2085,
+      "step": 465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4692482915717545e-05,
+      "loss": 3.9911,
+      "step": 466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.468109339407745e-05,
+      "loss": 4.0765,
+      "step": 467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4669703872437354e-05,
+      "loss": 1.1664,
+      "step": 468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.465831435079727e-05,
+      "loss": 4.8137,
+      "step": 469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.464692482915718e-05,
+      "loss": 4.8991,
+      "step": 470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4635535307517086e-05,
+      "loss": 3.9248,
+      "step": 471
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4624145785877e-05,
+      "loss": 3.2483,
+      "step": 472
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.46127562642369e-05,
+      "loss": 4.4666,
+      "step": 473
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.460136674259681e-05,
+      "loss": 2.3851,
+      "step": 474
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4589977220956725e-05,
+      "loss": 6.2086,
+      "step": 475
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4578587699316634e-05,
+      "loss": 5.0022,
+      "step": 476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456719817767654e-05,
+      "loss": 4.9041,
+      "step": 477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.455580865603644e-05,
+      "loss": 3.0523,
+      "step": 478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.454441913439636e-05,
+      "loss": 5.3083,
+      "step": 479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4533029612756266e-05,
+      "loss": 3.7251,
+      "step": 480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4521640091116175e-05,
+      "loss": 3.28,
+      "step": 481
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451025056947609e-05,
+      "loss": 4.2009,
+      "step": 482
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.449886104783599e-05,
+      "loss": 3.8922,
+      "step": 483
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.44874715261959e-05,
+      "loss": 3.2768,
+      "step": 484
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4476082004555814e-05,
+      "loss": 4.8212,
+      "step": 485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.446469248291572e-05,
+      "loss": 2.6052,
+      "step": 486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.445330296127563e-05,
+      "loss": 4.7922,
+      "step": 487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.444191343963553e-05,
+      "loss": 2.8647,
+      "step": 488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.443052391799545e-05,
+      "loss": 3.3726,
+      "step": 489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4419134396355355e-05,
+      "loss": 4.1974,
+      "step": 490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4407744874715263e-05,
+      "loss": 4.2478,
+      "step": 491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.439635535307518e-05,
+      "loss": 4.0767,
+      "step": 492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.438496583143508e-05,
+      "loss": 4.7489,
+      "step": 493
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437357630979499e-05,
+      "loss": 3.7689,
+      "step": 494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4362186788154896e-05,
+      "loss": 3.2168,
+      "step": 495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.435079726651481e-05,
+      "loss": 4.8521,
+      "step": 496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.433940774487472e-05,
+      "loss": 4.4044,
+      "step": 497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.432801822323462e-05,
+      "loss": 3.1875,
+      "step": 498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4316628701594536e-05,
+      "loss": 3.2144,
+      "step": 499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4305239179954444e-05,
+      "loss": 3.7766,
+      "step": 500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.429384965831435e-05,
+      "loss": 3.634,
+      "step": 501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.428246013667427e-05,
+      "loss": 2.8294,
+      "step": 502
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427107061503417e-05,
+      "loss": 4.9571,
+      "step": 503
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.425968109339408e-05,
+      "loss": 3.9025,
+      "step": 504
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4248291571753985e-05,
+      "loss": 4.996,
+      "step": 505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.42369020501139e-05,
+      "loss": 4.9177,
+      "step": 506
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.422551252847381e-05,
+      "loss": 5.3733,
+      "step": 507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4214123006833716e-05,
+      "loss": 3.8932,
+      "step": 508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4202733485193625e-05,
+      "loss": 4.3604,
+      "step": 509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.419134396355353e-05,
+      "loss": 4.1412,
+      "step": 510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.417995444191344e-05,
+      "loss": 4.9158,
+      "step": 511
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4168564920273356e-05,
+      "loss": 5.385,
+      "step": 512
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.415717539863326e-05,
+      "loss": 4.8995,
+      "step": 513
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4145785876993166e-05,
+      "loss": 4.788,
+      "step": 514
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4134396355353074e-05,
+      "loss": 4.6207,
+      "step": 515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412300683371299e-05,
+      "loss": 4.9936,
+      "step": 516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41116173120729e-05,
+      "loss": 3.3529,
+      "step": 517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4100227790432805e-05,
+      "loss": 4.1422,
+      "step": 518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4088838268792713e-05,
+      "loss": 1.8616,
+      "step": 519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.407744874715262e-05,
+      "loss": 3.8323,
+      "step": 520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406605922551253e-05,
+      "loss": 4.1239,
+      "step": 521
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4054669703872445e-05,
+      "loss": 3.2862,
+      "step": 522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4043280182232346e-05,
+      "loss": 6.2788,
+      "step": 523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4031890660592255e-05,
+      "loss": 4.6367,
+      "step": 524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402050113895216e-05,
+      "loss": 5.112,
+      "step": 525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.400911161731208e-05,
+      "loss": 4.0429,
+      "step": 526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3997722095671986e-05,
+      "loss": 5.0633,
+      "step": 527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3986332574031894e-05,
+      "loss": 5.081,
+      "step": 528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.39749430523918e-05,
+      "loss": 3.661,
+      "step": 529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396355353075171e-05,
+      "loss": 3.4382,
+      "step": 530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.395216400911162e-05,
+      "loss": 4.2483,
+      "step": 531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3940774487471534e-05,
+      "loss": 5.288,
+      "step": 532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3929384965831435e-05,
+      "loss": 3.6802,
+      "step": 533
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.391799544419134e-05,
+      "loss": 4.2897,
+      "step": 534
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.390660592255125e-05,
+      "loss": 4.2861,
+      "step": 535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3895216400911167e-05,
+      "loss": 4.3842,
+      "step": 536
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3883826879271075e-05,
+      "loss": 3.3977,
+      "step": 537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387243735763098e-05,
+      "loss": 3.791,
+      "step": 538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.386104783599089e-05,
+      "loss": 5.8116,
+      "step": 539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.38496583143508e-05,
+      "loss": 4.6096,
+      "step": 540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.383826879271071e-05,
+      "loss": 4.2682,
+      "step": 541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3826879271070616e-05,
+      "loss": 4.451,
+      "step": 542
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3815489749430524e-05,
+      "loss": 3.5013,
+      "step": 543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.380410022779043e-05,
+      "loss": 4.2927,
+      "step": 544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.379271070615034e-05,
+      "loss": 2.5419,
+      "step": 545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3781321184510255e-05,
+      "loss": 4.0713,
+      "step": 546
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3769931662870164e-05,
+      "loss": 3.0441,
+      "step": 547
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.375854214123007e-05,
+      "loss": 3.8489,
+      "step": 548
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.374715261958998e-05,
+      "loss": 2.8462,
+      "step": 549
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.373576309794989e-05,
+      "loss": 4.1786,
+      "step": 550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3724373576309796e-05,
+      "loss": 5.0482,
+      "step": 551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3712984054669705e-05,
+      "loss": 4.782,
+      "step": 552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.370159453302961e-05,
+      "loss": 5.4005,
+      "step": 553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.369020501138952e-05,
+      "loss": 3.2131,
+      "step": 554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.367881548974943e-05,
+      "loss": 4.6117,
+      "step": 555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3667425968109344e-05,
+      "loss": 5.1029,
+      "step": 556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365603644646925e-05,
+      "loss": 3.9967,
+      "step": 557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.364464692482916e-05,
+      "loss": 3.7775,
+      "step": 558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.363325740318907e-05,
+      "loss": 4.1765,
+      "step": 559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362186788154898e-05,
+      "loss": 4.7995,
+      "step": 560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3610478359908885e-05,
+      "loss": 3.7265,
+      "step": 561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3599088838268793e-05,
+      "loss": 4.182,
+      "step": 562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358769931662871e-05,
+      "loss": 5.1092,
+      "step": 563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357630979498861e-05,
+      "loss": 3.6195,
+      "step": 564
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.356492027334852e-05,
+      "loss": 3.8401,
+      "step": 565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.355353075170843e-05,
+      "loss": 5.0265,
+      "step": 566
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.354214123006834e-05,
+      "loss": 5.7561,
+      "step": 567
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353075170842825e-05,
+      "loss": 4.3377,
+      "step": 568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351936218678816e-05,
+      "loss": 3.6901,
+      "step": 569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3507972665148066e-05,
+      "loss": 6.0479,
+      "step": 570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3496583143507974e-05,
+      "loss": 3.4372,
+      "step": 571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.348519362186788e-05,
+      "loss": 4.4532,
+      "step": 572
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34738041002278e-05,
+      "loss": 4.1838,
+      "step": 573
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34624145785877e-05,
+      "loss": 2.836,
+      "step": 574
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.345102505694761e-05,
+      "loss": 4.7012,
+      "step": 575
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.343963553530752e-05,
+      "loss": 4.4546,
+      "step": 576
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.342824601366743e-05,
+      "loss": 4.2986,
+      "step": 577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341685649202734e-05,
+      "loss": 3.2922,
+      "step": 578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3405466970387246e-05,
+      "loss": 5.6031,
+      "step": 579
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3394077448747155e-05,
+      "loss": 3.6583,
+      "step": 580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.338268792710706e-05,
+      "loss": 3.8802,
+      "step": 581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337129840546697e-05,
+      "loss": 3.8014,
+      "step": 582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3359908883826886e-05,
+      "loss": 4.7704,
+      "step": 583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.334851936218679e-05,
+      "loss": 5.2258,
+      "step": 584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3337129840546696e-05,
+      "loss": 4.9469,
+      "step": 585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.332574031890661e-05,
+      "loss": 4.6626,
+      "step": 586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.331435079726652e-05,
+      "loss": 3.7777,
+      "step": 587
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.330296127562643e-05,
+      "loss": 4.3624,
+      "step": 588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.329157175398633e-05,
+      "loss": 5.3928,
+      "step": 589
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3280182232346244e-05,
+      "loss": 2.6942,
+      "step": 590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326879271070615e-05,
+      "loss": 3.6714,
+      "step": 591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.325740318906606e-05,
+      "loss": 4.5303,
+      "step": 592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3246013667425975e-05,
+      "loss": 3.404,
+      "step": 593
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3234624145785876e-05,
+      "loss": 5.0388,
+      "step": 594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3223234624145785e-05,
+      "loss": 3.8656,
+      "step": 595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32118451025057e-05,
+      "loss": 2.8938,
+      "step": 596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.320045558086561e-05,
+      "loss": 5.5352,
+      "step": 597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3189066059225516e-05,
+      "loss": 3.7565,
+      "step": 598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3177676537585424e-05,
+      "loss": 3.2011,
+      "step": 599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316628701594533e-05,
+      "loss": 5.0238,
+      "step": 600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.315489749430524e-05,
+      "loss": 6.4132,
+      "step": 601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.314350797266515e-05,
+      "loss": 5.5027,
+      "step": 602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3132118451025064e-05,
+      "loss": 4.7896,
+      "step": 603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3120728929384965e-05,
+      "loss": 3.9138,
+      "step": 604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3109339407744873e-05,
+      "loss": 3.5225,
+      "step": 605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.309794988610479e-05,
+      "loss": 5.1113,
+      "step": 606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3086560364464697e-05,
+      "loss": 3.6958,
+      "step": 607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3075170842824605e-05,
+      "loss": 6.2519,
+      "step": 608
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306378132118451e-05,
+      "loss": 5.9003,
+      "step": 609
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.305239179954442e-05,
+      "loss": 4.0551,
+      "step": 610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.304100227790433e-05,
+      "loss": 3.8801,
+      "step": 611
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.302961275626424e-05,
+      "loss": 4.0641,
+      "step": 612
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301822323462415e-05,
+      "loss": 4.9948,
+      "step": 613
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3006833712984054e-05,
+      "loss": 4.145,
+      "step": 614
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299544419134396e-05,
+      "loss": 6.1493,
+      "step": 615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298405466970388e-05,
+      "loss": 2.7276,
+      "step": 616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2972665148063785e-05,
+      "loss": 5.3351,
+      "step": 617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2961275626423694e-05,
+      "loss": 4.7405,
+      "step": 618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.29498861047836e-05,
+      "loss": 3.8263,
+      "step": 619
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293849658314351e-05,
+      "loss": 4.6732,
+      "step": 620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.292710706150342e-05,
+      "loss": 4.9215,
+      "step": 621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2915717539863326e-05,
+      "loss": 3.7222,
+      "step": 622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.290432801822324e-05,
+      "loss": 4.0339,
+      "step": 623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.289293849658314e-05,
+      "loss": 4.6049,
+      "step": 624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.288154897494305e-05,
+      "loss": 4.3873,
+      "step": 625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2870159453302966e-05,
+      "loss": 4.2938,
+      "step": 626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2858769931662874e-05,
+      "loss": 4.3877,
+      "step": 627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.284738041002278e-05,
+      "loss": 3.8315,
+      "step": 628
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.283599088838269e-05,
+      "loss": 4.154,
+      "step": 629
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28246013667426e-05,
+      "loss": 4.8419,
+      "step": 630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281321184510251e-05,
+      "loss": 4.4535,
+      "step": 631
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2801822323462415e-05,
+      "loss": 3.9572,
+      "step": 632
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.279043280182233e-05,
+      "loss": 4.7239,
+      "step": 633
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.277904328018223e-05,
+      "loss": 3.9885,
+      "step": 634
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276765375854214e-05,
+      "loss": 4.8701,
+      "step": 635
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.275626423690205e-05,
+      "loss": 4.4303,
+      "step": 636
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.274487471526196e-05,
+      "loss": 3.0225,
+      "step": 637
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.273348519362187e-05,
+      "loss": 5.866,
+      "step": 638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272209567198178e-05,
+      "loss": 5.2628,
+      "step": 639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271070615034169e-05,
+      "loss": 4.2152,
+      "step": 640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2699316628701596e-05,
+      "loss": 3.8122,
+      "step": 641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2687927107061504e-05,
+      "loss": 2.9753,
+      "step": 642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.267653758542142e-05,
+      "loss": 3.4472,
+      "step": 643
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.266514806378132e-05,
+      "loss": 3.6087,
+      "step": 644
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265375854214123e-05,
+      "loss": 4.543,
+      "step": 645
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.264236902050114e-05,
+      "loss": 3.9913,
+      "step": 646
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.263097949886105e-05,
+      "loss": 4.0886,
+      "step": 647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261958997722096e-05,
+      "loss": 4.8276,
+      "step": 648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.260820045558087e-05,
+      "loss": 3.3445,
+      "step": 649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2596810933940777e-05,
+      "loss": 4.0064,
+      "step": 650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2585421412300685e-05,
+      "loss": 3.9325,
+      "step": 651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257403189066059e-05,
+      "loss": 5.3082,
+      "step": 652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256264236902051e-05,
+      "loss": 2.038,
+      "step": 653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2551252847380416e-05,
+      "loss": 4.7109,
+      "step": 654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.253986332574032e-05,
+      "loss": 4.3733,
+      "step": 655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2528473804100226e-05,
+      "loss": 4.1101,
+      "step": 656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251708428246014e-05,
+      "loss": 5.4305,
+      "step": 657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.250569476082005e-05,
+      "loss": 4.8375,
+      "step": 658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.249430523917996e-05,
+      "loss": 3.8022,
+      "step": 659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2482915717539865e-05,
+      "loss": 4.5994,
+      "step": 660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2471526195899774e-05,
+      "loss": 4.9205,
+      "step": 661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246013667425968e-05,
+      "loss": 6.1354,
+      "step": 662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.24487471526196e-05,
+      "loss": 1.2473,
+      "step": 663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2437357630979505e-05,
+      "loss": 2.2868,
+      "step": 664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2425968109339406e-05,
+      "loss": 5.4158,
+      "step": 665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2414578587699315e-05,
+      "loss": 4.4452,
+      "step": 666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.240318906605923e-05,
+      "loss": 3.9093,
+      "step": 667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.239179954441914e-05,
+      "loss": 3.8827,
+      "step": 668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2380410022779046e-05,
+      "loss": 3.9207,
+      "step": 669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2369020501138954e-05,
+      "loss": 4.4915,
+      "step": 670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.235763097949886e-05,
+      "loss": 3.7661,
+      "step": 671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.234624145785877e-05,
+      "loss": 2.5415,
+      "step": 672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2334851936218686e-05,
+      "loss": 3.5786,
+      "step": 673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2323462414578594e-05,
+      "loss": 3.4722,
+      "step": 674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2312072892938495e-05,
+      "loss": 3.0377,
+      "step": 675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2300683371298403e-05,
+      "loss": 3.9187,
+      "step": 676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.228929384965832e-05,
+      "loss": 4.6948,
+      "step": 677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.227790432801823e-05,
+      "loss": 6.5729,
+      "step": 678
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2266514806378135e-05,
+      "loss": 3.9694,
+      "step": 679
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225512528473804e-05,
+      "loss": 3.2774,
+      "step": 680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.224373576309795e-05,
+      "loss": 3.4559,
+      "step": 681
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.223234624145786e-05,
+      "loss": 4.6314,
+      "step": 682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222095671981777e-05,
+      "loss": 3.0336,
+      "step": 683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.220956719817768e-05,
+      "loss": 3.2656,
+      "step": 684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2198177676537584e-05,
+      "loss": 4.076,
+      "step": 685
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.218678815489749e-05,
+      "loss": 4.4816,
+      "step": 686
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.217539863325741e-05,
+      "loss": 4.7967,
+      "step": 687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2164009111617315e-05,
+      "loss": 4.5978,
+      "step": 688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2152619589977224e-05,
+      "loss": 4.3149,
+      "step": 689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.214123006833713e-05,
+      "loss": 5.722,
+      "step": 690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.212984054669704e-05,
+      "loss": 3.9776,
+      "step": 691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211845102505695e-05,
+      "loss": 3.543,
+      "step": 692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2107061503416856e-05,
+      "loss": 4.0105,
+      "step": 693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.209567198177677e-05,
+      "loss": 3.0615,
+      "step": 694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.208428246013667e-05,
+      "loss": 2.49,
+      "step": 695
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207289293849658e-05,
+      "loss": 4.4175,
+      "step": 696
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2061503416856496e-05,
+      "loss": 3.7772,
+      "step": 697
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2050113895216404e-05,
+      "loss": 4.4857,
+      "step": 698
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.203872437357631e-05,
+      "loss": 4.7637,
+      "step": 699
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.202733485193622e-05,
+      "loss": 3.8691,
+      "step": 700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.201594533029613e-05,
+      "loss": 3.4015,
+      "step": 701
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200455580865604e-05,
+      "loss": 5.147,
+      "step": 702
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1993166287015945e-05,
+      "loss": 3.9907,
+      "step": 703
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.198177676537586e-05,
+      "loss": 5.1503,
+      "step": 704
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197038724373576e-05,
+      "loss": 5.6523,
+      "step": 705
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.195899772209567e-05,
+      "loss": 3.0013,
+      "step": 706
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1947608200455585e-05,
+      "loss": 3.7191,
+      "step": 707
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.193621867881549e-05,
+      "loss": 5.2139,
+      "step": 708
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19248291571754e-05,
+      "loss": 3.3182,
+      "step": 709
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191343963553531e-05,
+      "loss": 4.4892,
+      "step": 710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.190205011389522e-05,
+      "loss": 4.5257,
+      "step": 711
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1890660592255126e-05,
+      "loss": 3.7721,
+      "step": 712
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1879271070615034e-05,
+      "loss": 4.1988,
+      "step": 713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186788154897495e-05,
+      "loss": 2.9514,
+      "step": 714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.185649202733485e-05,
+      "loss": 2.4962,
+      "step": 715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.184510250569476e-05,
+      "loss": 4.0976,
+      "step": 716
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1833712984054674e-05,
+      "loss": 4.1026,
+      "step": 717
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182232346241458e-05,
+      "loss": 4.6964,
+      "step": 718
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.181093394077449e-05,
+      "loss": 5.4545,
+      "step": 719
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.17995444191344e-05,
+      "loss": 4.3369,
+      "step": 720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1788154897494307e-05,
+      "loss": 4.1629,
+      "step": 721
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1776765375854215e-05,
+      "loss": 4.6848,
+      "step": 722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176537585421412e-05,
+      "loss": 5.0396,
+      "step": 723
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.175398633257404e-05,
+      "loss": 3.8777,
+      "step": 724
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.174259681093394e-05,
+      "loss": 4.0251,
+      "step": 725
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.173120728929385e-05,
+      "loss": 4.6029,
+      "step": 726
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171981776765376e-05,
+      "loss": 3.9035,
+      "step": 727
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.170842824601367e-05,
+      "loss": 4.6807,
+      "step": 728
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.169703872437358e-05,
+      "loss": 4.8788,
+      "step": 729
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.168564920273349e-05,
+      "loss": 4.9378,
+      "step": 730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1674259681093395e-05,
+      "loss": 3.7656,
+      "step": 731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1662870159453304e-05,
+      "loss": 3.7814,
+      "step": 732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.165148063781321e-05,
+      "loss": 4.2167,
+      "step": 733
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.164009111617313e-05,
+      "loss": 3.8505,
+      "step": 734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1628701594533035e-05,
+      "loss": 4.3529,
+      "step": 735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1617312072892936e-05,
+      "loss": 5.4575,
+      "step": 736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.160592255125285e-05,
+      "loss": 4.8332,
+      "step": 737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.159453302961276e-05,
+      "loss": 3.4784,
+      "step": 738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.158314350797267e-05,
+      "loss": 4.4715,
+      "step": 739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1571753986332576e-05,
+      "loss": 5.3348,
+      "step": 740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1560364464692484e-05,
+      "loss": 3.261,
+      "step": 741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.154897494305239e-05,
+      "loss": 4.0993,
+      "step": 742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.15375854214123e-05,
+      "loss": 3.0709,
+      "step": 743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1526195899772216e-05,
+      "loss": 4.048,
+      "step": 744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1514806378132124e-05,
+      "loss": 2.3342,
+      "step": 745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1503416856492025e-05,
+      "loss": 4.2506,
+      "step": 746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.149202733485194e-05,
+      "loss": 4.1301,
+      "step": 747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.148063781321185e-05,
+      "loss": 3.8864,
+      "step": 748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146924829157176e-05,
+      "loss": 2.6433,
+      "step": 749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1457858769931665e-05,
+      "loss": 4.2471,
+      "step": 750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.144646924829157e-05,
+      "loss": 4.4804,
+      "step": 751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.143507972665148e-05,
+      "loss": 3.6997,
+      "step": 752
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142369020501139e-05,
+      "loss": 4.4154,
+      "step": 753
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1412300683371304e-05,
+      "loss": 4.6541,
+      "step": 754
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.140091116173121e-05,
+      "loss": 3.2707,
+      "step": 755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1389521640091114e-05,
+      "loss": 4.2344,
+      "step": 756
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.137813211845103e-05,
+      "loss": 4.7061,
+      "step": 757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136674259681094e-05,
+      "loss": 3.5924,
+      "step": 758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1355353075170845e-05,
+      "loss": 3.8157,
+      "step": 759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1343963553530754e-05,
+      "loss": 4.9797,
+      "step": 760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.133257403189066e-05,
+      "loss": 3.5863,
+      "step": 761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132118451025057e-05,
+      "loss": 4.1521,
+      "step": 762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.130979498861048e-05,
+      "loss": 4.0992,
+      "step": 763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.129840546697039e-05,
+      "loss": 4.06,
+      "step": 764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.12870159453303e-05,
+      "loss": 5.1037,
+      "step": 765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.12756264236902e-05,
+      "loss": 4.0429,
+      "step": 766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126423690205012e-05,
+      "loss": 4.9454,
+      "step": 767
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1252847380410026e-05,
+      "loss": 4.2007,
+      "step": 768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1241457858769934e-05,
+      "loss": 3.8143,
+      "step": 769
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.123006833712984e-05,
+      "loss": 3.2415,
+      "step": 770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121867881548975e-05,
+      "loss": 5.0029,
+      "step": 771
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.120728929384966e-05,
+      "loss": 4.77,
+      "step": 772
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.119589977220957e-05,
+      "loss": 4.066,
+      "step": 773
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.118451025056948e-05,
+      "loss": 5.1107,
+      "step": 774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117312072892939e-05,
+      "loss": 3.8146,
+      "step": 775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116173120728929e-05,
+      "loss": 3.691,
+      "step": 776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.115034168564921e-05,
+      "loss": 3.796,
+      "step": 777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1138952164009115e-05,
+      "loss": 2.9657,
+      "step": 778
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.112756264236902e-05,
+      "loss": 3.8946,
+      "step": 779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111617312072893e-05,
+      "loss": 5.0799,
+      "step": 780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.110478359908884e-05,
+      "loss": 4.0241,
+      "step": 781
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.109339407744875e-05,
+      "loss": 4.7757,
+      "step": 782
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1082004555808656e-05,
+      "loss": 3.6373,
+      "step": 783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107061503416857e-05,
+      "loss": 3.4799,
+      "step": 784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.105922551252848e-05,
+      "loss": 3.1831,
+      "step": 785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.104783599088838e-05,
+      "loss": 4.461,
+      "step": 786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.103644646924829e-05,
+      "loss": 5.3129,
+      "step": 787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1025056947608204e-05,
+      "loss": 3.2919,
+      "step": 788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101366742596811e-05,
+      "loss": 4.3643,
+      "step": 789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.100227790432802e-05,
+      "loss": 5.4305,
+      "step": 790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.099088838268793e-05,
+      "loss": 4.315,
+      "step": 791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0979498861047837e-05,
+      "loss": 3.3629,
+      "step": 792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0968109339407745e-05,
+      "loss": 6.4542,
+      "step": 793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.095671981776766e-05,
+      "loss": 4.4329,
+      "step": 794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.094533029612757e-05,
+      "loss": 4.3684,
+      "step": 795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.093394077448747e-05,
+      "loss": 3.1349,
+      "step": 796
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092255125284738e-05,
+      "loss": 3.5018,
+      "step": 797
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091116173120729e-05,
+      "loss": 4.6676,
+      "step": 798
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.08997722095672e-05,
+      "loss": 3.4454,
+      "step": 799
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.088838268792711e-05,
+      "loss": 4.7973,
+      "step": 800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.087699316628702e-05,
+      "loss": 3.2544,
+      "step": 801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0865603644646925e-05,
+      "loss": 3.7521,
+      "step": 802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0854214123006834e-05,
+      "loss": 5.0676,
+      "step": 803
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.084282460136675e-05,
+      "loss": 5.13,
+      "step": 804
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.083143507972666e-05,
+      "loss": 3.1873,
+      "step": 805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082004555808656e-05,
+      "loss": 2.7409,
+      "step": 806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0808656036446466e-05,
+      "loss": 4.3346,
+      "step": 807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.079726651480638e-05,
+      "loss": 2.8558,
+      "step": 808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.078587699316629e-05,
+      "loss": 3.9207,
+      "step": 809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07744874715262e-05,
+      "loss": 4.5951,
+      "step": 810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0763097949886106e-05,
+      "loss": 3.1811,
+      "step": 811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0751708428246014e-05,
+      "loss": 3.9337,
+      "step": 812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.074031890660592e-05,
+      "loss": 4.9005,
+      "step": 813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.072892938496584e-05,
+      "loss": 3.3076,
+      "step": 814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0717539863325746e-05,
+      "loss": 2.7829,
+      "step": 815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.070615034168565e-05,
+      "loss": 4.2093,
+      "step": 816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0694760820045555e-05,
+      "loss": 4.8367,
+      "step": 817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.068337129840547e-05,
+      "loss": 3.7364,
+      "step": 818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067198177676538e-05,
+      "loss": 5.2442,
+      "step": 819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066059225512529e-05,
+      "loss": 3.5053,
+      "step": 820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0649202733485195e-05,
+      "loss": 5.6835,
+      "step": 821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06378132118451e-05,
+      "loss": 4.2826,
+      "step": 822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.062642369020501e-05,
+      "loss": 4.5851,
+      "step": 823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0615034168564926e-05,
+      "loss": 4.7558,
+      "step": 824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0603644646924834e-05,
+      "loss": 4.303,
+      "step": 825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.059225512528474e-05,
+      "loss": 3.7662,
+      "step": 826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0580865603644644e-05,
+      "loss": 3.8451,
+      "step": 827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056947608200456e-05,
+      "loss": 3.4654,
+      "step": 828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.055808656036447e-05,
+      "loss": 3.8322,
+      "step": 829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0546697038724376e-05,
+      "loss": 5.5813,
+      "step": 830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0535307517084284e-05,
+      "loss": 2.9534,
+      "step": 831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052391799544419e-05,
+      "loss": 4.5507,
+      "step": 832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05125284738041e-05,
+      "loss": 2.4864,
+      "step": 833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.050113895216401e-05,
+      "loss": 3.0065,
+      "step": 834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.048974943052392e-05,
+      "loss": 2.2133,
+      "step": 835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.047835990888383e-05,
+      "loss": 3.5685,
+      "step": 836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.046697038724373e-05,
+      "loss": 2.7335,
+      "step": 837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.045558086560365e-05,
+      "loss": 6.3468,
+      "step": 838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0444191343963556e-05,
+      "loss": 4.5155,
+      "step": 839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0432801822323464e-05,
+      "loss": 5.1838,
+      "step": 840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042141230068337e-05,
+      "loss": 4.9923,
+      "step": 841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041002277904328e-05,
+      "loss": 3.8952,
+      "step": 842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.039863325740319e-05,
+      "loss": 3.4326,
+      "step": 843
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.03872437357631e-05,
+      "loss": 4.781,
+      "step": 844
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.037585421412301e-05,
+      "loss": 4.3584,
+      "step": 845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036446469248292e-05,
+      "loss": 4.3816,
+      "step": 846
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.035307517084282e-05,
+      "loss": 3.8401,
+      "step": 847
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.034168564920274e-05,
+      "loss": 4.1301,
+      "step": 848
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0330296127562645e-05,
+      "loss": 3.9048,
+      "step": 849
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.031890660592255e-05,
+      "loss": 3.0471,
+      "step": 850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.030751708428246e-05,
+      "loss": 2.7112,
+      "step": 851
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.029612756264237e-05,
+      "loss": 4.9531,
+      "step": 852
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.028473804100228e-05,
+      "loss": 4.0546,
+      "step": 853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0273348519362186e-05,
+      "loss": 3.3567,
+      "step": 854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.02619589977221e-05,
+      "loss": 3.1925,
+      "step": 855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.025056947608201e-05,
+      "loss": 4.817,
+      "step": 856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.023917995444191e-05,
+      "loss": 4.7476,
+      "step": 857
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0227790432801826e-05,
+      "loss": 3.1915,
+      "step": 858
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0216400911161734e-05,
+      "loss": 4.2456,
+      "step": 859
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.020501138952164e-05,
+      "loss": 3.2835,
+      "step": 860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.019362186788155e-05,
+      "loss": 4.1989,
+      "step": 861
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.018223234624146e-05,
+      "loss": 4.3931,
+      "step": 862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017084282460137e-05,
+      "loss": 2.9001,
+      "step": 863
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0159453302961275e-05,
+      "loss": 2.3858,
+      "step": 864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.014806378132119e-05,
+      "loss": 2.5247,
+      "step": 865
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.01366742596811e-05,
+      "loss": 4.4906,
+      "step": 866
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0125284738041e-05,
+      "loss": 3.7757,
+      "step": 867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0113895216400914e-05,
+      "loss": 3.8259,
+      "step": 868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.010250569476082e-05,
+      "loss": 5.2635,
+      "step": 869
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.009111617312073e-05,
+      "loss": 3.1292,
+      "step": 870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.007972665148064e-05,
+      "loss": 4.6459,
+      "step": 871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006833712984055e-05,
+      "loss": 3.1908,
+      "step": 872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0056947608200455e-05,
+      "loss": 3.8403,
+      "step": 873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0045558086560364e-05,
+      "loss": 4.4295,
+      "step": 874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.003416856492028e-05,
+      "loss": 3.7787,
+      "step": 875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002277904328019e-05,
+      "loss": 3.3765,
+      "step": 876
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001138952164009e-05,
+      "loss": 3.6083,
+      "step": 877
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4e-05,
+      "loss": 1.4368,
+      "step": 878
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.998861047835991e-05,
+      "loss": 3.2122,
+      "step": 879
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.997722095671982e-05,
+      "loss": 2.6489,
+      "step": 880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996583143507973e-05,
+      "loss": 4.2684,
+      "step": 881
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9954441913439636e-05,
+      "loss": 1.9694,
+      "step": 882
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9943052391799544e-05,
+      "loss": 3.2589,
+      "step": 883
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.993166287015945e-05,
+      "loss": 5.0037,
+      "step": 884
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992027334851937e-05,
+      "loss": 3.8751,
+      "step": 885
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9908883826879276e-05,
+      "loss": 2.8458,
+      "step": 886
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.989749430523918e-05,
+      "loss": 4.9801,
+      "step": 887
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.988610478359909e-05,
+      "loss": 3.1468,
+      "step": 888
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9874715261959e-05,
+      "loss": 4.2309,
+      "step": 889
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986332574031891e-05,
+      "loss": 4.5296,
+      "step": 890
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.985193621867882e-05,
+      "loss": 4.9182,
+      "step": 891
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9840546697038725e-05,
+      "loss": 3.7365,
+      "step": 892
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.982915717539863e-05,
+      "loss": 2.9334,
+      "step": 893
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981776765375854e-05,
+      "loss": 4.4482,
+      "step": 894
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9806378132118456e-05,
+      "loss": 2.3568,
+      "step": 895
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9794988610478365e-05,
+      "loss": 3.228,
+      "step": 896
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9783599088838266e-05,
+      "loss": 5.3102,
+      "step": 897
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977220956719818e-05,
+      "loss": 4.997,
+      "step": 898
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976082004555809e-05,
+      "loss": 4.1334,
+      "step": 899
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9749430523918e-05,
+      "loss": 6.4749,
+      "step": 900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9738041002277906e-05,
+      "loss": 3.612,
+      "step": 901
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9726651480637814e-05,
+      "loss": 5.0688,
+      "step": 902
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971526195899772e-05,
+      "loss": 4.3306,
+      "step": 903
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.970387243735763e-05,
+      "loss": 2.3229,
+      "step": 904
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9692482915717545e-05,
+      "loss": 4.9858,
+      "step": 905
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.968109339407745e-05,
+      "loss": 3.3844,
+      "step": 906
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966970387243736e-05,
+      "loss": 2.4339,
+      "step": 907
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.965831435079727e-05,
+      "loss": 4.6444,
+      "step": 908
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.964692482915718e-05,
+      "loss": 3.8285,
+      "step": 909
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9635535307517086e-05,
+      "loss": 4.2341,
+      "step": 910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9624145785876994e-05,
+      "loss": 3.2534,
+      "step": 911
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96127562642369e-05,
+      "loss": 5.0014,
+      "step": 912
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.960136674259681e-05,
+      "loss": 6.1065,
+      "step": 913
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.958997722095672e-05,
+      "loss": 3.0212,
+      "step": 914
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9578587699316634e-05,
+      "loss": 4.3614,
+      "step": 915
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956719817767654e-05,
+      "loss": 4.1599,
+      "step": 916
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.955580865603645e-05,
+      "loss": 5.0953,
+      "step": 917
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.954441913439636e-05,
+      "loss": 3.3387,
+      "step": 918
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.953302961275627e-05,
+      "loss": 5.3928,
+      "step": 919
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9521640091116175e-05,
+      "loss": 4.0223,
+      "step": 920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951025056947608e-05,
+      "loss": 3.8907,
+      "step": 921
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.949886104783599e-05,
+      "loss": 5.1154,
+      "step": 922
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.94874715261959e-05,
+      "loss": 4.3147,
+      "step": 923
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.947608200455581e-05,
+      "loss": 3.8695,
+      "step": 924
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946469248291572e-05,
+      "loss": 3.179,
+      "step": 925
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.945330296127563e-05,
+      "loss": 4.1564,
+      "step": 926
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.944191343963554e-05,
+      "loss": 4.8976,
+      "step": 927
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.943052391799544e-05,
+      "loss": 3.5135,
+      "step": 928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9419134396355356e-05,
+      "loss": 4.7498,
+      "step": 929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9407744874715264e-05,
+      "loss": 3.3925,
+      "step": 930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.939635535307517e-05,
+      "loss": 4.4992,
+      "step": 931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.938496583143508e-05,
+      "loss": 4.8598,
+      "step": 932
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937357630979499e-05,
+      "loss": 4.1789,
+      "step": 933
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.93621867881549e-05,
+      "loss": 1.9236,
+      "step": 934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.935079726651481e-05,
+      "loss": 5.2944,
+      "step": 935
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933940774487472e-05,
+      "loss": 2.7277,
+      "step": 936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.932801822323463e-05,
+      "loss": 3.9701,
+      "step": 937
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931662870159453e-05,
+      "loss": 3.2435,
+      "step": 938
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9305239179954444e-05,
+      "loss": 3.2713,
+      "step": 939
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.929384965831435e-05,
+      "loss": 3.7607,
+      "step": 940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.928246013667426e-05,
+      "loss": 3.9998,
+      "step": 941
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927107061503417e-05,
+      "loss": 4.9845,
+      "step": 942
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925968109339408e-05,
+      "loss": 4.2552,
+      "step": 943
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9248291571753986e-05,
+      "loss": 4.3532,
+      "step": 944
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92369020501139e-05,
+      "loss": 4.092,
+      "step": 945
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922551252847381e-05,
+      "loss": 3.9094,
+      "step": 946
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921412300683372e-05,
+      "loss": 4.2419,
+      "step": 947
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.920273348519362e-05,
+      "loss": 2.2963,
+      "step": 948
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.919134396355353e-05,
+      "loss": 5.2482,
+      "step": 949
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.917995444191344e-05,
+      "loss": 3.1734,
+      "step": 950
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916856492027335e-05,
+      "loss": 4.0273,
+      "step": 951
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.915717539863326e-05,
+      "loss": 4.3067,
+      "step": 952
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9145785876993166e-05,
+      "loss": 5.1908,
+      "step": 953
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9134396355353074e-05,
+      "loss": 3.87,
+      "step": 954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912300683371299e-05,
+      "loss": 7.2181,
+      "step": 955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91116173120729e-05,
+      "loss": 3.1337,
+      "step": 956
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9100227790432806e-05,
+      "loss": 4.4346,
+      "step": 957
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.908883826879271e-05,
+      "loss": 3.2696,
+      "step": 958
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.907744874715262e-05,
+      "loss": 3.8024,
+      "step": 959
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906605922551253e-05,
+      "loss": 4.2184,
+      "step": 960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.905466970387244e-05,
+      "loss": 3.4126,
+      "step": 961
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9043280182232354e-05,
+      "loss": 4.5691,
+      "step": 962
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9031890660592255e-05,
+      "loss": 3.1731,
+      "step": 963
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902050113895216e-05,
+      "loss": 4.552,
+      "step": 964
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.900911161731208e-05,
+      "loss": 3.0406,
+      "step": 965
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8997722095671986e-05,
+      "loss": 3.1508,
+      "step": 966
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8986332574031895e-05,
+      "loss": 3.4265,
+      "step": 967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8974943052391796e-05,
+      "loss": 5.4161,
+      "step": 968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896355353075171e-05,
+      "loss": 3.6103,
+      "step": 969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.895216400911162e-05,
+      "loss": 4.9294,
+      "step": 970
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.894077448747153e-05,
+      "loss": 6.2261,
+      "step": 971
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.892938496583144e-05,
+      "loss": 3.3562,
+      "step": 972
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8917995444191344e-05,
+      "loss": 3.5101,
+      "step": 973
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.890660592255125e-05,
+      "loss": 3.5394,
+      "step": 974
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.889521640091117e-05,
+      "loss": 3.7131,
+      "step": 975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8883826879271075e-05,
+      "loss": 1.4657,
+      "step": 976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8872437357630983e-05,
+      "loss": 3.8771,
+      "step": 977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8861047835990885e-05,
+      "loss": 3.7374,
+      "step": 978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88496583143508e-05,
+      "loss": 5.1045,
+      "step": 979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.883826879271071e-05,
+      "loss": 5.2137,
+      "step": 980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8826879271070616e-05,
+      "loss": 3.8647,
+      "step": 981
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881548974943053e-05,
+      "loss": 3.4477,
+      "step": 982
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.880410022779043e-05,
+      "loss": 4.1642,
+      "step": 983
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.879271070615034e-05,
+      "loss": 5.148,
+      "step": 984
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.878132118451025e-05,
+      "loss": 5.7339,
+      "step": 985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8769931662870164e-05,
+      "loss": 2.7099,
+      "step": 986
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.875854214123007e-05,
+      "loss": 3.4009,
+      "step": 987
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8747152619589974e-05,
+      "loss": 4.7134,
+      "step": 988
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.873576309794989e-05,
+      "loss": 3.49,
+      "step": 989
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87243735763098e-05,
+      "loss": 4.7586,
+      "step": 990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8712984054669705e-05,
+      "loss": 3.687,
+      "step": 991
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.870159453302962e-05,
+      "loss": 2.5007,
+      "step": 992
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.869020501138952e-05,
+      "loss": 4.2222,
+      "step": 993
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.867881548974943e-05,
+      "loss": 4.3712,
+      "step": 994
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866742596810934e-05,
+      "loss": 2.6943,
+      "step": 995
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.865603644646925e-05,
+      "loss": 4.628,
+      "step": 996
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.864464692482916e-05,
+      "loss": 5.7406,
+      "step": 997
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.863325740318907e-05,
+      "loss": 3.3306,
+      "step": 998
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862186788154898e-05,
+      "loss": 2.3235,
+      "step": 999
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8610478359908886e-05,
+      "loss": 3.3689,
+      "step": 1000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8599088838268794e-05,
+      "loss": 4.5757,
+      "step": 1001
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.858769931662871e-05,
+      "loss": 3.1753,
+      "step": 1002
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.857630979498861e-05,
+      "loss": 4.8172,
+      "step": 1003
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856492027334852e-05,
+      "loss": 4.0467,
+      "step": 1004
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.855353075170843e-05,
+      "loss": 4.1106,
+      "step": 1005
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.854214123006834e-05,
+      "loss": 3.5062,
+      "step": 1006
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.853075170842825e-05,
+      "loss": 3.9264,
+      "step": 1007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851936218678816e-05,
+      "loss": 4.7678,
+      "step": 1008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8507972665148066e-05,
+      "loss": 3.7188,
+      "step": 1009
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8496583143507975e-05,
+      "loss": 4.6767,
+      "step": 1010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.848519362186788e-05,
+      "loss": 5.2841,
+      "step": 1011
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84738041002278e-05,
+      "loss": 4.5878,
+      "step": 1012
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84624145785877e-05,
+      "loss": 3.8425,
+      "step": 1013
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.845102505694761e-05,
+      "loss": 2.679,
+      "step": 1014
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8439635535307516e-05,
+      "loss": 4.9919,
+      "step": 1015
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.842824601366743e-05,
+      "loss": 3.6078,
+      "step": 1016
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841685649202734e-05,
+      "loss": 3.8711,
+      "step": 1017
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.840546697038725e-05,
+      "loss": 3.0731,
+      "step": 1018
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8394077448747155e-05,
+      "loss": 5.3156,
+      "step": 1019
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.838268792710706e-05,
+      "loss": 2.7345,
+      "step": 1020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837129840546697e-05,
+      "loss": 3.3537,
+      "step": 1021
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8359908883826887e-05,
+      "loss": 3.8499,
+      "step": 1022
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.834851936218679e-05,
+      "loss": 3.5882,
+      "step": 1023
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8337129840546696e-05,
+      "loss": 6.1484,
+      "step": 1024
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8325740318906604e-05,
+      "loss": 3.2842,
+      "step": 1025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831435079726652e-05,
+      "loss": 4.3209,
+      "step": 1026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.830296127562643e-05,
+      "loss": 3.9721,
+      "step": 1027
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8291571753986336e-05,
+      "loss": 4.7583,
+      "step": 1028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8280182232346244e-05,
+      "loss": 3.0846,
+      "step": 1029
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826879271070615e-05,
+      "loss": 4.8416,
+      "step": 1030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.825740318906606e-05,
+      "loss": 4.0177,
+      "step": 1031
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.824601366742597e-05,
+      "loss": 4.9417,
+      "step": 1032
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.823462414578588e-05,
+      "loss": 5.4386,
+      "step": 1033
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8223234624145785e-05,
+      "loss": 2.7618,
+      "step": 1034
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821184510250569e-05,
+      "loss": 4.9191,
+      "step": 1035
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.820045558086561e-05,
+      "loss": 3.9095,
+      "step": 1036
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8189066059225516e-05,
+      "loss": 2.7008,
+      "step": 1037
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8177676537585425e-05,
+      "loss": 3.1976,
+      "step": 1038
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816628701594533e-05,
+      "loss": 3.2674,
+      "step": 1039
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.815489749430524e-05,
+      "loss": 4.2568,
+      "step": 1040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.814350797266515e-05,
+      "loss": 3.9299,
+      "step": 1041
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.813211845102506e-05,
+      "loss": 4.768,
+      "step": 1042
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8120728929384966e-05,
+      "loss": 3.8306,
+      "step": 1043
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8109339407744874e-05,
+      "loss": 4.1608,
+      "step": 1044
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.809794988610478e-05,
+      "loss": 3.8067,
+      "step": 1045
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.80865603644647e-05,
+      "loss": 3.7705,
+      "step": 1046
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8075170842824605e-05,
+      "loss": 5.2807,
+      "step": 1047
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8063781321184513e-05,
+      "loss": 3.7809,
+      "step": 1048
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.805239179954442e-05,
+      "loss": 3.905,
+      "step": 1049
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.804100227790433e-05,
+      "loss": 2.7237,
+      "step": 1050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.802961275626424e-05,
+      "loss": 2.4741,
+      "step": 1051
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8018223234624146e-05,
+      "loss": 3.1944,
+      "step": 1052
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.800683371298406e-05,
+      "loss": 3.5913,
+      "step": 1053
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.799544419134396e-05,
+      "loss": 4.8227,
+      "step": 1054
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.798405466970387e-05,
+      "loss": 3.0044,
+      "step": 1055
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7972665148063786e-05,
+      "loss": 5.0847,
+      "step": 1056
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7961275626423694e-05,
+      "loss": 2.047,
+      "step": 1057
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.79498861047836e-05,
+      "loss": 2.7585,
+      "step": 1058
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.793849658314351e-05,
+      "loss": 3.7358,
+      "step": 1059
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.792710706150342e-05,
+      "loss": 3.2987,
+      "step": 1060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791571753986333e-05,
+      "loss": 3.1248,
+      "step": 1061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7904328018223235e-05,
+      "loss": 4.2381,
+      "step": 1062
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.789293849658315e-05,
+      "loss": 3.6331,
+      "step": 1063
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.788154897494305e-05,
+      "loss": 4.1034,
+      "step": 1064
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787015945330296e-05,
+      "loss": 4.5044,
+      "step": 1065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7858769931662875e-05,
+      "loss": 4.4643,
+      "step": 1066
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.784738041002278e-05,
+      "loss": 3.8745,
+      "step": 1067
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.783599088838269e-05,
+      "loss": 4.2535,
+      "step": 1068
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78246013667426e-05,
+      "loss": 3.6993,
+      "step": 1069
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781321184510251e-05,
+      "loss": 3.9537,
+      "step": 1070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7801822323462416e-05,
+      "loss": 5.2109,
+      "step": 1071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7790432801822324e-05,
+      "loss": 3.9586,
+      "step": 1072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.777904328018224e-05,
+      "loss": 4.7916,
+      "step": 1073
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776765375854214e-05,
+      "loss": 5.0228,
+      "step": 1074
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775626423690205e-05,
+      "loss": 3.7901,
+      "step": 1075
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7744874715261964e-05,
+      "loss": 3.0828,
+      "step": 1076
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.773348519362187e-05,
+      "loss": 3.7481,
+      "step": 1077
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772209567198178e-05,
+      "loss": 3.9793,
+      "step": 1078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771070615034169e-05,
+      "loss": 3.3927,
+      "step": 1079
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7699316628701596e-05,
+      "loss": 4.5633,
+      "step": 1080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7687927107061505e-05,
+      "loss": 2.5691,
+      "step": 1081
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.767653758542141e-05,
+      "loss": 4.06,
+      "step": 1082
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766514806378133e-05,
+      "loss": 3.1244,
+      "step": 1083
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.765375854214123e-05,
+      "loss": 3.9971,
+      "step": 1084
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764236902050114e-05,
+      "loss": 3.6864,
+      "step": 1085
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.763097949886105e-05,
+      "loss": 3.8497,
+      "step": 1086
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761958997722096e-05,
+      "loss": 3.2384,
+      "step": 1087
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.760820045558087e-05,
+      "loss": 4.2741,
+      "step": 1088
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.759681093394078e-05,
+      "loss": 3.569,
+      "step": 1089
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7585421412300685e-05,
+      "loss": 4.5599,
+      "step": 1090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757403189066059e-05,
+      "loss": 4.4357,
+      "step": 1091
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.75626423690205e-05,
+      "loss": 3.1993,
+      "step": 1092
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7551252847380417e-05,
+      "loss": 3.2672,
+      "step": 1093
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.753986332574032e-05,
+      "loss": 5.6199,
+      "step": 1094
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7528473804100226e-05,
+      "loss": 3.8451,
+      "step": 1095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751708428246014e-05,
+      "loss": 3.4279,
+      "step": 1096
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.750569476082005e-05,
+      "loss": 3.7235,
+      "step": 1097
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.749430523917996e-05,
+      "loss": 4.2263,
+      "step": 1098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7482915717539866e-05,
+      "loss": 4.1779,
+      "step": 1099
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7471526195899774e-05,
+      "loss": 5.2156,
+      "step": 1100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746013667425968e-05,
+      "loss": 3.9364,
+      "step": 1101
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744874715261959e-05,
+      "loss": 3.628,
+      "step": 1102
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7437357630979505e-05,
+      "loss": 2.2342,
+      "step": 1103
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.742596810933941e-05,
+      "loss": 3.1376,
+      "step": 1104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7414578587699315e-05,
+      "loss": 2.7007,
+      "step": 1105
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.740318906605923e-05,
+      "loss": 5.4253,
+      "step": 1106
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.739179954441914e-05,
+      "loss": 4.244,
+      "step": 1107
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7380410022779046e-05,
+      "loss": 4.3566,
+      "step": 1108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7369020501138955e-05,
+      "loss": 4.7238,
+      "step": 1109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.735763097949886e-05,
+      "loss": 5.299,
+      "step": 1110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.734624145785877e-05,
+      "loss": 3.2266,
+      "step": 1111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.733485193621868e-05,
+      "loss": 4.2154,
+      "step": 1112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7323462414578594e-05,
+      "loss": 1.7892,
+      "step": 1113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7312072892938496e-05,
+      "loss": 4.6587,
+      "step": 1114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7300683371298404e-05,
+      "loss": 3.644,
+      "step": 1115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.728929384965832e-05,
+      "loss": 2.5175,
+      "step": 1116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.727790432801823e-05,
+      "loss": 4.3624,
+      "step": 1117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7266514806378135e-05,
+      "loss": 3.9207,
+      "step": 1118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7255125284738043e-05,
+      "loss": 3.689,
+      "step": 1119
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.724373576309795e-05,
+      "loss": 4.5249,
+      "step": 1120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.723234624145786e-05,
+      "loss": 3.9408,
+      "step": 1121
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722095671981777e-05,
+      "loss": 3.7057,
+      "step": 1122
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.720956719817768e-05,
+      "loss": 3.3047,
+      "step": 1123
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7198177676537584e-05,
+      "loss": 3.3738,
+      "step": 1124
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.718678815489749e-05,
+      "loss": 4.4538,
+      "step": 1125
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.71753986332574e-05,
+      "loss": 5.0996,
+      "step": 1126
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7164009111617316e-05,
+      "loss": 3.2371,
+      "step": 1127
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7152619589977224e-05,
+      "loss": 3.8497,
+      "step": 1128
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.714123006833713e-05,
+      "loss": 3.1759,
+      "step": 1129
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.712984054669704e-05,
+      "loss": 4.5299,
+      "step": 1130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.711845102505695e-05,
+      "loss": 3.513,
+      "step": 1131
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.710706150341686e-05,
+      "loss": 1.977,
+      "step": 1132
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.709567198177677e-05,
+      "loss": 2.8565,
+      "step": 1133
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.708428246013668e-05,
+      "loss": 2.1834,
+      "step": 1134
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.707289293849658e-05,
+      "loss": 5.1557,
+      "step": 1135
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706150341685649e-05,
+      "loss": 4.3376,
+      "step": 1136
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7050113895216405e-05,
+      "loss": 2.8139,
+      "step": 1137
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.703872437357631e-05,
+      "loss": 2.4282,
+      "step": 1138
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.702733485193622e-05,
+      "loss": 1.5373,
+      "step": 1139
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701594533029613e-05,
+      "loss": 3.3173,
+      "step": 1140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.700455580865604e-05,
+      "loss": 2.6265,
+      "step": 1141
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6993166287015946e-05,
+      "loss": 3.8921,
+      "step": 1142
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698177676537586e-05,
+      "loss": 3.3402,
+      "step": 1143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.697038724373577e-05,
+      "loss": 2.7108,
+      "step": 1144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.695899772209567e-05,
+      "loss": 5.4658,
+      "step": 1145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.694760820045558e-05,
+      "loss": 3.7322,
+      "step": 1146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6936218678815494e-05,
+      "loss": 7.7483,
+      "step": 1147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.69248291571754e-05,
+      "loss": 2.0365,
+      "step": 1148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691343963553531e-05,
+      "loss": 4.5518,
+      "step": 1149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.690205011389522e-05,
+      "loss": 4.0917,
+      "step": 1150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6890660592255126e-05,
+      "loss": 4.677,
+      "step": 1151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6879271070615035e-05,
+      "loss": 5.4221,
+      "step": 1152
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686788154897495e-05,
+      "loss": 3.1085,
+      "step": 1153
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.685649202733486e-05,
+      "loss": 3.3841,
+      "step": 1154
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.684510250569476e-05,
+      "loss": 4.2578,
+      "step": 1155
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.683371298405467e-05,
+      "loss": 3.4327,
+      "step": 1156
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682232346241458e-05,
+      "loss": 5.1256,
+      "step": 1157
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681093394077449e-05,
+      "loss": 4.9221,
+      "step": 1158
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.67995444191344e-05,
+      "loss": 3.6421,
+      "step": 1159
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.678815489749431e-05,
+      "loss": 2.4972,
+      "step": 1160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6776765375854215e-05,
+      "loss": 4.0932,
+      "step": 1161
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6765375854214123e-05,
+      "loss": 3.3364,
+      "step": 1162
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.675398633257404e-05,
+      "loss": 4.5112,
+      "step": 1163
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6742596810933947e-05,
+      "loss": 3.1809,
+      "step": 1164
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.673120728929385e-05,
+      "loss": 4.1175,
+      "step": 1165
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6719817767653756e-05,
+      "loss": 4.2892,
+      "step": 1166
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.670842824601367e-05,
+      "loss": 3.4144,
+      "step": 1167
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.669703872437358e-05,
+      "loss": 3.9388,
+      "step": 1168
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.668564920273349e-05,
+      "loss": 4.4217,
+      "step": 1169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6674259681093396e-05,
+      "loss": 3.5351,
+      "step": 1170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6662870159453304e-05,
+      "loss": 3.0859,
+      "step": 1171
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.665148063781321e-05,
+      "loss": 3.2805,
+      "step": 1172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.664009111617312e-05,
+      "loss": 3.8536,
+      "step": 1173
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6628701594533035e-05,
+      "loss": 3.3108,
+      "step": 1174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661731207289294e-05,
+      "loss": 3.7463,
+      "step": 1175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6605922551252845e-05,
+      "loss": 1.8433,
+      "step": 1176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.659453302961276e-05,
+      "loss": 3.2689,
+      "step": 1177
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.658314350797267e-05,
+      "loss": 3.8837,
+      "step": 1178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6571753986332576e-05,
+      "loss": 3.1761,
+      "step": 1179
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6560364464692485e-05,
+      "loss": 3.0668,
+      "step": 1180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.654897494305239e-05,
+      "loss": 3.7212,
+      "step": 1181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.65375854214123e-05,
+      "loss": 2.8245,
+      "step": 1182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.652619589977221e-05,
+      "loss": 3.3913,
+      "step": 1183
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6514806378132124e-05,
+      "loss": 1.7768,
+      "step": 1184
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6503416856492026e-05,
+      "loss": 4.0475,
+      "step": 1185
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6492027334851934e-05,
+      "loss": 6.3199,
+      "step": 1186
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.648063781321185e-05,
+      "loss": 2.5516,
+      "step": 1187
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646924829157176e-05,
+      "loss": 4.6267,
+      "step": 1188
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6457858769931665e-05,
+      "loss": 4.0397,
+      "step": 1189
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6446469248291574e-05,
+      "loss": 4.0454,
+      "step": 1190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.643507972665148e-05,
+      "loss": 3.9336,
+      "step": 1191
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642369020501139e-05,
+      "loss": 6.1619,
+      "step": 1192
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.64123006833713e-05,
+      "loss": 2.9099,
+      "step": 1193
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.640091116173121e-05,
+      "loss": 4.2379,
+      "step": 1194
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6389521640091115e-05,
+      "loss": 4.1134,
+      "step": 1195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.637813211845102e-05,
+      "loss": 2.3904,
+      "step": 1196
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636674259681094e-05,
+      "loss": 4.6903,
+      "step": 1197
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6355353075170846e-05,
+      "loss": 4.0794,
+      "step": 1198
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6343963553530754e-05,
+      "loss": 2.4764,
+      "step": 1199
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.633257403189066e-05,
+      "loss": 3.6178,
+      "step": 1200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632118451025057e-05,
+      "loss": 2.7739,
+      "step": 1201
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.630979498861048e-05,
+      "loss": 3.3146,
+      "step": 1202
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.629840546697039e-05,
+      "loss": 3.5353,
+      "step": 1203
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.62870159453303e-05,
+      "loss": 2.7564,
+      "step": 1204
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.62756264236902e-05,
+      "loss": 3.1023,
+      "step": 1205
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626423690205011e-05,
+      "loss": 3.8548,
+      "step": 1206
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6252847380410027e-05,
+      "loss": 2.5454,
+      "step": 1207
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6241457858769935e-05,
+      "loss": 3.8095,
+      "step": 1208
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.623006833712984e-05,
+      "loss": 4.4206,
+      "step": 1209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621867881548975e-05,
+      "loss": 3.4695,
+      "step": 1210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.620728929384966e-05,
+      "loss": 2.6636,
+      "step": 1211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.619589977220957e-05,
+      "loss": 3.418,
+      "step": 1212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6184510250569476e-05,
+      "loss": 3.4057,
+      "step": 1213
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617312072892939e-05,
+      "loss": 3.2303,
+      "step": 1214
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616173120728929e-05,
+      "loss": 4.027,
+      "step": 1215
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.61503416856492e-05,
+      "loss": 3.7989,
+      "step": 1216
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6138952164009115e-05,
+      "loss": 4.4272,
+      "step": 1217
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6127562642369024e-05,
+      "loss": 4.0301,
+      "step": 1218
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611617312072893e-05,
+      "loss": 4.8619,
+      "step": 1219
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.610478359908884e-05,
+      "loss": 4.7324,
+      "step": 1220
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.609339407744875e-05,
+      "loss": 3.0772,
+      "step": 1221
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6082004555808656e-05,
+      "loss": 2.2499,
+      "step": 1222
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6070615034168565e-05,
+      "loss": 2.8443,
+      "step": 1223
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.605922551252848e-05,
+      "loss": 4.8003,
+      "step": 1224
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.604783599088839e-05,
+      "loss": 5.3797,
+      "step": 1225
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.603644646924829e-05,
+      "loss": 3.679,
+      "step": 1226
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6025056947608204e-05,
+      "loss": 4.4492,
+      "step": 1227
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.601366742596811e-05,
+      "loss": 4.6749,
+      "step": 1228
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.600227790432802e-05,
+      "loss": 2.5715,
+      "step": 1229
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.599088838268793e-05,
+      "loss": 3.3855,
+      "step": 1230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.597949886104784e-05,
+      "loss": 3.8342,
+      "step": 1231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5968109339407745e-05,
+      "loss": 3.2072,
+      "step": 1232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5956719817767653e-05,
+      "loss": 2.0799,
+      "step": 1233
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.594533029612757e-05,
+      "loss": 4.1076,
+      "step": 1234
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.593394077448748e-05,
+      "loss": 3.5207,
+      "step": 1235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.592255125284738e-05,
+      "loss": 1.84,
+      "step": 1236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591116173120729e-05,
+      "loss": 3.5141,
+      "step": 1237
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.58997722095672e-05,
+      "loss": 3.184,
+      "step": 1238
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.588838268792711e-05,
+      "loss": 4.3074,
+      "step": 1239
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.587699316628702e-05,
+      "loss": 5.0945,
+      "step": 1240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5865603644646926e-05,
+      "loss": 2.2768,
+      "step": 1241
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5854214123006834e-05,
+      "loss": 3.7634,
+      "step": 1242
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.584282460136674e-05,
+      "loss": 4.1213,
+      "step": 1243
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.583143507972666e-05,
+      "loss": 4.5129,
+      "step": 1244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5820045558086565e-05,
+      "loss": 4.079,
+      "step": 1245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.580865603644647e-05,
+      "loss": 4.2489,
+      "step": 1246
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.579726651480638e-05,
+      "loss": 2.6536,
+      "step": 1247
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.578587699316629e-05,
+      "loss": 4.7195,
+      "step": 1248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.57744874715262e-05,
+      "loss": 3.4295,
+      "step": 1249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5763097949886106e-05,
+      "loss": 3.522,
+      "step": 1250
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5751708428246015e-05,
+      "loss": 5.7531,
+      "step": 1251
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.574031890660592e-05,
+      "loss": 4.1642,
+      "step": 1252
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.572892938496583e-05,
+      "loss": 3.6727,
+      "step": 1253
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5717539863325746e-05,
+      "loss": 4.4349,
+      "step": 1254
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5706150341685654e-05,
+      "loss": 5.1503,
+      "step": 1255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5694760820045556e-05,
+      "loss": 5.2402,
+      "step": 1256
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.568337129840547e-05,
+      "loss": 4.3445,
+      "step": 1257
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567198177676538e-05,
+      "loss": 5.567,
+      "step": 1258
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.566059225512529e-05,
+      "loss": 4.3022,
+      "step": 1259
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5649202733485195e-05,
+      "loss": 2.4479,
+      "step": 1260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5637813211845104e-05,
+      "loss": 2.4457,
+      "step": 1261
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.562642369020501e-05,
+      "loss": 3.9943,
+      "step": 1262
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561503416856492e-05,
+      "loss": 2.9602,
+      "step": 1263
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5603644646924835e-05,
+      "loss": 4.1636,
+      "step": 1264
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.559225512528474e-05,
+      "loss": 2.6054,
+      "step": 1265
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5580865603644645e-05,
+      "loss": 4.982,
+      "step": 1266
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556947608200456e-05,
+      "loss": 3.3602,
+      "step": 1267
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.555808656036447e-05,
+      "loss": 3.4297,
+      "step": 1268
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5546697038724376e-05,
+      "loss": 4.8674,
+      "step": 1269
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5535307517084284e-05,
+      "loss": 5.367,
+      "step": 1270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.552391799544419e-05,
+      "loss": 3.1432,
+      "step": 1271
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.55125284738041e-05,
+      "loss": 4.2729,
+      "step": 1272
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.550113895216401e-05,
+      "loss": 1.8594,
+      "step": 1273
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5489749430523924e-05,
+      "loss": 4.6949,
+      "step": 1274
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.547835990888383e-05,
+      "loss": 4.8613,
+      "step": 1275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5466970387243733e-05,
+      "loss": 2.7389,
+      "step": 1276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.545558086560364e-05,
+      "loss": 2.2207,
+      "step": 1277
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5444191343963557e-05,
+      "loss": 2.5128,
+      "step": 1278
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5432801822323465e-05,
+      "loss": 3.5753,
+      "step": 1279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.542141230068337e-05,
+      "loss": 2.0318,
+      "step": 1280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541002277904328e-05,
+      "loss": 3.6463,
+      "step": 1281
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.539863325740319e-05,
+      "loss": 2.4542,
+      "step": 1282
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.53872437357631e-05,
+      "loss": 4.1335,
+      "step": 1283
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.537585421412301e-05,
+      "loss": 5.4845,
+      "step": 1284
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536446469248292e-05,
+      "loss": 4.3487,
+      "step": 1285
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.535307517084282e-05,
+      "loss": 3.6915,
+      "step": 1286
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.534168564920273e-05,
+      "loss": 0.9357,
+      "step": 1287
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5330296127562645e-05,
+      "loss": 2.8569,
+      "step": 1288
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5318906605922554e-05,
+      "loss": 4.6225,
+      "step": 1289
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.530751708428246e-05,
+      "loss": 3.3422,
+      "step": 1290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.529612756264237e-05,
+      "loss": 4.4059,
+      "step": 1291
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.528473804100228e-05,
+      "loss": 5.2657,
+      "step": 1292
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5273348519362186e-05,
+      "loss": 3.2975,
+      "step": 1293
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.52619589977221e-05,
+      "loss": 2.9637,
+      "step": 1294
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.525056947608201e-05,
+      "loss": 3.48,
+      "step": 1295
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523917995444191e-05,
+      "loss": 4.0594,
+      "step": 1296
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.522779043280182e-05,
+      "loss": 4.0241,
+      "step": 1297
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5216400911161734e-05,
+      "loss": 2.6585,
+      "step": 1298
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.520501138952164e-05,
+      "loss": 3.9501,
+      "step": 1299
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.519362186788155e-05,
+      "loss": 4.5016,
+      "step": 1300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.518223234624146e-05,
+      "loss": 2.6866,
+      "step": 1301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517084282460137e-05,
+      "loss": 3.3291,
+      "step": 1302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5159453302961275e-05,
+      "loss": 5.7273,
+      "step": 1303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.514806378132119e-05,
+      "loss": 5.3974,
+      "step": 1304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.51366742596811e-05,
+      "loss": 3.3339,
+      "step": 1305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5125284738041e-05,
+      "loss": 3.5451,
+      "step": 1306
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511389521640091e-05,
+      "loss": 3.8225,
+      "step": 1307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510250569476082e-05,
+      "loss": 3.7317,
+      "step": 1308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.509111617312073e-05,
+      "loss": 3.5054,
+      "step": 1309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.507972665148064e-05,
+      "loss": 4.464,
+      "step": 1310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506833712984055e-05,
+      "loss": 3.7213,
+      "step": 1311
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5056947608200456e-05,
+      "loss": 3.3084,
+      "step": 1312
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5045558086560364e-05,
+      "loss": 4.2152,
+      "step": 1313
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.503416856492028e-05,
+      "loss": 4.4331,
+      "step": 1314
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502277904328019e-05,
+      "loss": 2.5979,
+      "step": 1315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5011389521640096e-05,
+      "loss": 4.022,
+      "step": 1316
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5e-05,
+      "loss": 3.8312,
+      "step": 1317
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.498861047835991e-05,
+      "loss": 3.5744,
+      "step": 1318
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.497722095671982e-05,
+      "loss": 2.392,
+      "step": 1319
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496583143507973e-05,
+      "loss": 3.3766,
+      "step": 1320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4954441913439637e-05,
+      "loss": 3.0052,
+      "step": 1321
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4943052391799545e-05,
+      "loss": 4.2677,
+      "step": 1322
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.493166287015945e-05,
+      "loss": 4.5942,
+      "step": 1323
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492027334851936e-05,
+      "loss": 3.1878,
+      "step": 1324
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4908883826879276e-05,
+      "loss": 4.0483,
+      "step": 1325
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4897494305239184e-05,
+      "loss": 3.5604,
+      "step": 1326
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4886104783599086e-05,
+      "loss": 3.1879,
+      "step": 1327
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4874715261959e-05,
+      "loss": 2.8531,
+      "step": 1328
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.486332574031891e-05,
+      "loss": 2.8441,
+      "step": 1329
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485193621867882e-05,
+      "loss": 4.0258,
+      "step": 1330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4840546697038725e-05,
+      "loss": 3.0,
+      "step": 1331
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4829157175398634e-05,
+      "loss": 2.7282,
+      "step": 1332
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481776765375854e-05,
+      "loss": 3.5903,
+      "step": 1333
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.480637813211845e-05,
+      "loss": 3.6406,
+      "step": 1334
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4794988610478365e-05,
+      "loss": 3.2491,
+      "step": 1335
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.478359908883827e-05,
+      "loss": 3.7037,
+      "step": 1336
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4772209567198175e-05,
+      "loss": 3.0835,
+      "step": 1337
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476082004555809e-05,
+      "loss": 3.145,
+      "step": 1338
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4749430523918e-05,
+      "loss": 4.729,
+      "step": 1339
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4738041002277906e-05,
+      "loss": 3.0796,
+      "step": 1340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4726651480637814e-05,
+      "loss": 4.0763,
+      "step": 1341
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471526195899772e-05,
+      "loss": 3.8396,
+      "step": 1342
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.470387243735763e-05,
+      "loss": 4.1636,
+      "step": 1343
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.469248291571754e-05,
+      "loss": 2.8494,
+      "step": 1344
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4681093394077454e-05,
+      "loss": 4.0225,
+      "step": 1345
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466970387243736e-05,
+      "loss": 2.1322,
+      "step": 1346
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4658314350797263e-05,
+      "loss": 3.4794,
+      "step": 1347
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.464692482915718e-05,
+      "loss": 4.0108,
+      "step": 1348
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.463553530751709e-05,
+      "loss": 4.1043,
+      "step": 1349
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4624145785876995e-05,
+      "loss": 3.3189,
+      "step": 1350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.46127562642369e-05,
+      "loss": 3.9222,
+      "step": 1351
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.460136674259681e-05,
+      "loss": 4.5522,
+      "step": 1352
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.458997722095672e-05,
+      "loss": 1.929,
+      "step": 1353
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.457858769931663e-05,
+      "loss": 3.2023,
+      "step": 1354
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456719817767654e-05,
+      "loss": 4.4514,
+      "step": 1355
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.455580865603645e-05,
+      "loss": 5.0489,
+      "step": 1356
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.454441913439635e-05,
+      "loss": 3.6021,
+      "step": 1357
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.453302961275627e-05,
+      "loss": 2.3818,
+      "step": 1358
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4521640091116175e-05,
+      "loss": 3.003,
+      "step": 1359
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4510250569476084e-05,
+      "loss": 3.2409,
+      "step": 1360
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4498861047836e-05,
+      "loss": 4.1765,
+      "step": 1361
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.44874715261959e-05,
+      "loss": 3.6401,
+      "step": 1362
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.447608200455581e-05,
+      "loss": 3.269,
+      "step": 1363
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4464692482915716e-05,
+      "loss": 5.1317,
+      "step": 1364
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.445330296127563e-05,
+      "loss": 3.4778,
+      "step": 1365
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.444191343963554e-05,
+      "loss": 4.1865,
+      "step": 1366
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.443052391799544e-05,
+      "loss": 2.6955,
+      "step": 1367
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4419134396355356e-05,
+      "loss": 2.6172,
+      "step": 1368
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4407744874715264e-05,
+      "loss": 3.016,
+      "step": 1369
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.439635535307517e-05,
+      "loss": 3.4665,
+      "step": 1370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.438496583143508e-05,
+      "loss": 3.8491,
+      "step": 1371
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437357630979499e-05,
+      "loss": 3.692,
+      "step": 1372
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.43621867881549e-05,
+      "loss": 3.7349,
+      "step": 1373
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4350797266514805e-05,
+      "loss": 4.0286,
+      "step": 1374
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.433940774487472e-05,
+      "loss": 2.1987,
+      "step": 1375
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.432801822323463e-05,
+      "loss": 2.5429,
+      "step": 1376
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431662870159453e-05,
+      "loss": 1.8266,
+      "step": 1377
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4305239179954445e-05,
+      "loss": 4.6244,
+      "step": 1378
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.429384965831435e-05,
+      "loss": 3.6279,
+      "step": 1379
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.428246013667426e-05,
+      "loss": 4.0254,
+      "step": 1380
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427107061503417e-05,
+      "loss": 2.7581,
+      "step": 1381
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.425968109339408e-05,
+      "loss": 3.0846,
+      "step": 1382
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4248291571753986e-05,
+      "loss": 2.9704,
+      "step": 1383
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4236902050113894e-05,
+      "loss": 3.333,
+      "step": 1384
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.422551252847381e-05,
+      "loss": 3.7139,
+      "step": 1385
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421412300683372e-05,
+      "loss": 4.7406,
+      "step": 1386
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.420273348519362e-05,
+      "loss": 3.1004,
+      "step": 1387
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4191343963553534e-05,
+      "loss": 3.178,
+      "step": 1388
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.417995444191344e-05,
+      "loss": 4.852,
+      "step": 1389
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416856492027335e-05,
+      "loss": 3.5873,
+      "step": 1390
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.415717539863326e-05,
+      "loss": 3.7675,
+      "step": 1391
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4145785876993167e-05,
+      "loss": 2.9701,
+      "step": 1392
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4134396355353075e-05,
+      "loss": 2.8724,
+      "step": 1393
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412300683371298e-05,
+      "loss": 4.5524,
+      "step": 1394
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.41116173120729e-05,
+      "loss": 3.0457,
+      "step": 1395
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4100227790432806e-05,
+      "loss": 4.0933,
+      "step": 1396
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4088838268792714e-05,
+      "loss": 2.6475,
+      "step": 1397
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.407744874715262e-05,
+      "loss": 2.4124,
+      "step": 1398
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406605922551253e-05,
+      "loss": 4.2414,
+      "step": 1399
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405466970387244e-05,
+      "loss": 4.0054,
+      "step": 1400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.404328018223235e-05,
+      "loss": 2.7678,
+      "step": 1401
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4031890660592255e-05,
+      "loss": 4.3202,
+      "step": 1402
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4020501138952164e-05,
+      "loss": 4.1301,
+      "step": 1403
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.400911161731207e-05,
+      "loss": 3.7285,
+      "step": 1404
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.399772209567199e-05,
+      "loss": 3.2684,
+      "step": 1405
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3986332574031895e-05,
+      "loss": 5.2644,
+      "step": 1406
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.39749430523918e-05,
+      "loss": 4.6861,
+      "step": 1407
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396355353075171e-05,
+      "loss": 1.8129,
+      "step": 1408
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395216400911162e-05,
+      "loss": 5.4895,
+      "step": 1409
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.394077448747153e-05,
+      "loss": 4.4256,
+      "step": 1410
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3929384965831436e-05,
+      "loss": 3.4606,
+      "step": 1411
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3917995444191344e-05,
+      "loss": 4.3124,
+      "step": 1412
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.390660592255125e-05,
+      "loss": 3.6244,
+      "step": 1413
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.389521640091116e-05,
+      "loss": 4.0277,
+      "step": 1414
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3883826879271076e-05,
+      "loss": 3.6087,
+      "step": 1415
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3872437357630984e-05,
+      "loss": 4.1885,
+      "step": 1416
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386104783599089e-05,
+      "loss": 3.2077,
+      "step": 1417
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3849658314350793e-05,
+      "loss": 4.3993,
+      "step": 1418
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.383826879271071e-05,
+      "loss": 4.2382,
+      "step": 1419
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.382687927107062e-05,
+      "loss": 5.1873,
+      "step": 1420
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3815489749430525e-05,
+      "loss": 3.9697,
+      "step": 1421
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.380410022779043e-05,
+      "loss": 3.241,
+      "step": 1422
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.379271070615034e-05,
+      "loss": 1.7778,
+      "step": 1423
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.378132118451025e-05,
+      "loss": 1.4306,
+      "step": 1424
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3769931662870164e-05,
+      "loss": 3.4686,
+      "step": 1425
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.375854214123007e-05,
+      "loss": 4.014,
+      "step": 1426
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.374715261958998e-05,
+      "loss": 3.4134,
+      "step": 1427
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.373576309794988e-05,
+      "loss": 4.3382,
+      "step": 1428
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37243735763098e-05,
+      "loss": 3.772,
+      "step": 1429
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3712984054669705e-05,
+      "loss": 4.1451,
+      "step": 1430
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3701594533029614e-05,
+      "loss": 2.4503,
+      "step": 1431
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.369020501138952e-05,
+      "loss": 5.2491,
+      "step": 1432
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.367881548974943e-05,
+      "loss": 3.2613,
+      "step": 1433
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366742596810934e-05,
+      "loss": 3.7955,
+      "step": 1434
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.365603644646925e-05,
+      "loss": 2.6215,
+      "step": 1435
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.364464692482916e-05,
+      "loss": 4.4617,
+      "step": 1436
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.363325740318907e-05,
+      "loss": 2.7337,
+      "step": 1437
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.362186788154897e-05,
+      "loss": 2.9903,
+      "step": 1438
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3610478359908886e-05,
+      "loss": 3.3441,
+      "step": 1439
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3599088838268794e-05,
+      "loss": 3.6923,
+      "step": 1440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.35876993166287e-05,
+      "loss": 3.5405,
+      "step": 1441
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.357630979498861e-05,
+      "loss": 5.2684,
+      "step": 1442
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356492027334852e-05,
+      "loss": 4.1415,
+      "step": 1443
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.355353075170843e-05,
+      "loss": 4.159,
+      "step": 1444
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.354214123006834e-05,
+      "loss": 4.351,
+      "step": 1445
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.353075170842825e-05,
+      "loss": 3.685,
+      "step": 1446
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351936218678816e-05,
+      "loss": 2.8265,
+      "step": 1447
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.350797266514806e-05,
+      "loss": 3.1912,
+      "step": 1448
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3496583143507975e-05,
+      "loss": 4.4132,
+      "step": 1449
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.348519362186788e-05,
+      "loss": 4.7935,
+      "step": 1450
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347380410022779e-05,
+      "loss": 4.4637,
+      "step": 1451
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3462414578587706e-05,
+      "loss": 3.9248,
+      "step": 1452
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.345102505694761e-05,
+      "loss": 3.5972,
+      "step": 1453
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3439635535307516e-05,
+      "loss": 3.7061,
+      "step": 1454
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.342824601366743e-05,
+      "loss": 3.9629,
+      "step": 1455
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341685649202734e-05,
+      "loss": 3.6778,
+      "step": 1456
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.340546697038725e-05,
+      "loss": 4.4582,
+      "step": 1457
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.339407744874715e-05,
+      "loss": 3.0396,
+      "step": 1458
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3382687927107064e-05,
+      "loss": 3.4166,
+      "step": 1459
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337129840546697e-05,
+      "loss": 3.653,
+      "step": 1460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335990888382688e-05,
+      "loss": 2.7515,
+      "step": 1461
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3348519362186795e-05,
+      "loss": 3.5607,
+      "step": 1462
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3337129840546697e-05,
+      "loss": 4.5227,
+      "step": 1463
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3325740318906605e-05,
+      "loss": 4.014,
+      "step": 1464
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331435079726651e-05,
+      "loss": 3.9865,
+      "step": 1465
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330296127562643e-05,
+      "loss": 4.5359,
+      "step": 1466
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3291571753986336e-05,
+      "loss": 3.2667,
+      "step": 1467
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.328018223234624e-05,
+      "loss": 3.8823,
+      "step": 1468
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326879271070615e-05,
+      "loss": 3.0791,
+      "step": 1469
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.325740318906606e-05,
+      "loss": 2.3747,
+      "step": 1470
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.324601366742597e-05,
+      "loss": 4.8208,
+      "step": 1471
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3234624145785884e-05,
+      "loss": 3.8421,
+      "step": 1472
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3223234624145785e-05,
+      "loss": 3.1643,
+      "step": 1473
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3211845102505694e-05,
+      "loss": 3.7348,
+      "step": 1474
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.32004555808656e-05,
+      "loss": 3.4273,
+      "step": 1475
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.318906605922552e-05,
+      "loss": 3.0749,
+      "step": 1476
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3177676537585425e-05,
+      "loss": 3.0157,
+      "step": 1477
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3166287015945326e-05,
+      "loss": 4.0944,
+      "step": 1478
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.315489749430524e-05,
+      "loss": 2.6539,
+      "step": 1479
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.314350797266515e-05,
+      "loss": 2.3148,
+      "step": 1480
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.313211845102506e-05,
+      "loss": 4.4766,
+      "step": 1481
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312072892938497e-05,
+      "loss": 3.3643,
+      "step": 1482
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3109339407744874e-05,
+      "loss": 4.5015,
+      "step": 1483
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.309794988610478e-05,
+      "loss": 2.653,
+      "step": 1484
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.308656036446469e-05,
+      "loss": 3.7364,
+      "step": 1485
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3075170842824606e-05,
+      "loss": 2.4379,
+      "step": 1486
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3063781321184514e-05,
+      "loss": 2.111,
+      "step": 1487
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305239179954442e-05,
+      "loss": 3.7125,
+      "step": 1488
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.304100227790433e-05,
+      "loss": 2.442,
+      "step": 1489
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.302961275626424e-05,
+      "loss": 3.2785,
+      "step": 1490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.301822323462415e-05,
+      "loss": 4.9819,
+      "step": 1491
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.300683371298406e-05,
+      "loss": 3.7154,
+      "step": 1492
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.299544419134396e-05,
+      "loss": 3.2143,
+      "step": 1493
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.298405466970387e-05,
+      "loss": 4.9498,
+      "step": 1494
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297266514806378e-05,
+      "loss": 4.6746,
+      "step": 1495
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2961275626423694e-05,
+      "loss": 4.457,
+      "step": 1496
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.29498861047836e-05,
+      "loss": 3.7867,
+      "step": 1497
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.293849658314351e-05,
+      "loss": 3.4591,
+      "step": 1498
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.292710706150342e-05,
+      "loss": 1.9091,
+      "step": 1499
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291571753986333e-05,
+      "loss": 3.2591,
+      "step": 1500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2904328018223236e-05,
+      "loss": 3.2701,
+      "step": 1501
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.289293849658315e-05,
+      "loss": 3.5629,
+      "step": 1502
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.288154897494305e-05,
+      "loss": 4.3215,
+      "step": 1503
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.287015945330296e-05,
+      "loss": 3.8457,
+      "step": 1504
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285876993166287e-05,
+      "loss": 4.1651,
+      "step": 1505
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.284738041002278e-05,
+      "loss": 3.776,
+      "step": 1506
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.283599088838269e-05,
+      "loss": 2.9386,
+      "step": 1507
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.28246013667426e-05,
+      "loss": 4.5317,
+      "step": 1508
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281321184510251e-05,
+      "loss": 3.2703,
+      "step": 1509
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2801822323462416e-05,
+      "loss": 3.4839,
+      "step": 1510
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2790432801822324e-05,
+      "loss": 2.1538,
+      "step": 1511
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.277904328018224e-05,
+      "loss": 2.8942,
+      "step": 1512
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276765375854214e-05,
+      "loss": 3.0613,
+      "step": 1513
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.275626423690205e-05,
+      "loss": 4.2821,
+      "step": 1514
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.274487471526196e-05,
+      "loss": 3.9091,
+      "step": 1515
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.273348519362187e-05,
+      "loss": 3.5384,
+      "step": 1516
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272209567198178e-05,
+      "loss": 3.6395,
+      "step": 1517
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271070615034169e-05,
+      "loss": 4.2068,
+      "step": 1518
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.26993166287016e-05,
+      "loss": 3.6311,
+      "step": 1519
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2687927107061505e-05,
+      "loss": 3.1192,
+      "step": 1520
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.267653758542141e-05,
+      "loss": 3.1039,
+      "step": 1521
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266514806378132e-05,
+      "loss": 3.1263,
+      "step": 1522
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.265375854214123e-05,
+      "loss": 2.0527,
+      "step": 1523
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264236902050114e-05,
+      "loss": 2.2847,
+      "step": 1524
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2630979498861046e-05,
+      "loss": 3.6335,
+      "step": 1525
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261958997722096e-05,
+      "loss": 5.0008,
+      "step": 1526
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.260820045558087e-05,
+      "loss": 4.3056,
+      "step": 1527
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.259681093394078e-05,
+      "loss": 5.0321,
+      "step": 1528
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2585421412300686e-05,
+      "loss": 4.8505,
+      "step": 1529
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2574031890660594e-05,
+      "loss": 3.9916,
+      "step": 1530
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25626423690205e-05,
+      "loss": 3.5501,
+      "step": 1531
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.255125284738041e-05,
+      "loss": 2.4659,
+      "step": 1532
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.253986332574032e-05,
+      "loss": 4.541,
+      "step": 1533
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.252847380410023e-05,
+      "loss": 3.1075,
+      "step": 1534
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2517084282460135e-05,
+      "loss": 6.012,
+      "step": 1535
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.250569476082005e-05,
+      "loss": 4.4673,
+      "step": 1536
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.249430523917996e-05,
+      "loss": 3.5484,
+      "step": 1537
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2482915717539866e-05,
+      "loss": 4.2984,
+      "step": 1538
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2471526195899774e-05,
+      "loss": 3.9299,
+      "step": 1539
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.246013667425968e-05,
+      "loss": 3.2651,
+      "step": 1540
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.244874715261959e-05,
+      "loss": 3.187,
+      "step": 1541
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.24373576309795e-05,
+      "loss": 2.8429,
+      "step": 1542
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2425968109339414e-05,
+      "loss": 2.7657,
+      "step": 1543
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2414578587699315e-05,
+      "loss": 3.9324,
+      "step": 1544
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2403189066059224e-05,
+      "loss": 3.1407,
+      "step": 1545
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.239179954441914e-05,
+      "loss": 5.2545,
+      "step": 1546
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.238041002277905e-05,
+      "loss": 2.5094,
+      "step": 1547
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2369020501138955e-05,
+      "loss": 3.8536,
+      "step": 1548
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.235763097949886e-05,
+      "loss": 2.9809,
+      "step": 1549
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.234624145785877e-05,
+      "loss": 3.3436,
+      "step": 1550
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.233485193621868e-05,
+      "loss": 5.759,
+      "step": 1551
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.232346241457859e-05,
+      "loss": 4.1713,
+      "step": 1552
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23120728929385e-05,
+      "loss": 2.2042,
+      "step": 1553
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2300683371298404e-05,
+      "loss": 3.6943,
+      "step": 1554
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.228929384965831e-05,
+      "loss": 2.8803,
+      "step": 1555
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.227790432801823e-05,
+      "loss": 4.2886,
+      "step": 1556
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2266514806378136e-05,
+      "loss": 5.602,
+      "step": 1557
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2255125284738044e-05,
+      "loss": 3.2288,
+      "step": 1558
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.224373576309795e-05,
+      "loss": 3.1248,
+      "step": 1559
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.223234624145786e-05,
+      "loss": 4.2938,
+      "step": 1560
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222095671981777e-05,
+      "loss": 3.5685,
+      "step": 1561
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.220956719817768e-05,
+      "loss": 3.8314,
+      "step": 1562
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.219817767653759e-05,
+      "loss": 4.7161,
+      "step": 1563
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.218678815489749e-05,
+      "loss": 3.829,
+      "step": 1564
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.21753986332574e-05,
+      "loss": 4.4536,
+      "step": 1565
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2164009111617316e-05,
+      "loss": 1.3494,
+      "step": 1566
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2152619589977225e-05,
+      "loss": 3.3832,
+      "step": 1567
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.214123006833713e-05,
+      "loss": 3.6154,
+      "step": 1568
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.212984054669704e-05,
+      "loss": 3.5555,
+      "step": 1569
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211845102505695e-05,
+      "loss": 3.7649,
+      "step": 1570
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.210706150341686e-05,
+      "loss": 4.7004,
+      "step": 1571
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2095671981776766e-05,
+      "loss": 2.9952,
+      "step": 1572
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.208428246013668e-05,
+      "loss": 3.2759,
+      "step": 1573
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207289293849658e-05,
+      "loss": 4.2699,
+      "step": 1574
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.206150341685649e-05,
+      "loss": 3.2543,
+      "step": 1575
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2050113895216405e-05,
+      "loss": 4.0741,
+      "step": 1576
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.203872437357631e-05,
+      "loss": 2.7018,
+      "step": 1577
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.202733485193622e-05,
+      "loss": 2.5774,
+      "step": 1578
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201594533029613e-05,
+      "loss": 3.6971,
+      "step": 1579
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.200455580865604e-05,
+      "loss": 2.7743,
+      "step": 1580
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.1993166287015946e-05,
+      "loss": 3.3904,
+      "step": 1581
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.1981776765375854e-05,
+      "loss": 1.7499,
+      "step": 1582
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197038724373577e-05,
+      "loss": 2.5419,
+      "step": 1583
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.195899772209567e-05,
+      "loss": 4.7309,
+      "step": 1584
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.194760820045558e-05,
+      "loss": 3.6122,
+      "step": 1585
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1936218678815494e-05,
+      "loss": 3.0166,
+      "step": 1586
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.19248291571754e-05,
+      "loss": 1.6438,
+      "step": 1587
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.191343963553531e-05,
+      "loss": 3.8119,
+      "step": 1588
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.190205011389522e-05,
+      "loss": 2.8015,
+      "step": 1589
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.189066059225513e-05,
+      "loss": 2.9901,
+      "step": 1590
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1879271070615035e-05,
+      "loss": 3.9637,
+      "step": 1591
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186788154897494e-05,
+      "loss": 2.9581,
+      "step": 1592
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.185649202733486e-05,
+      "loss": 5.0127,
+      "step": 1593
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.184510250569476e-05,
+      "loss": 4.8894,
+      "step": 1594
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.183371298405467e-05,
+      "loss": 3.324,
+      "step": 1595
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182232346241458e-05,
+      "loss": 2.0338,
+      "step": 1596
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181093394077449e-05,
+      "loss": 2.3338,
+      "step": 1597
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.17995444191344e-05,
+      "loss": 3.8097,
+      "step": 1598
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.178815489749431e-05,
+      "loss": 2.0692,
+      "step": 1599
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1776765375854216e-05,
+      "loss": 3.8163,
+      "step": 1600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1765375854214124e-05,
+      "loss": 5.1269,
+      "step": 1601
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.175398633257403e-05,
+      "loss": 2.5417,
+      "step": 1602
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.174259681093395e-05,
+      "loss": 4.2916,
+      "step": 1603
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.173120728929385e-05,
+      "loss": 3.651,
+      "step": 1604
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.171981776765376e-05,
+      "loss": 1.658,
+      "step": 1605
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.170842824601367e-05,
+      "loss": 2.5786,
+      "step": 1606
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.169703872437358e-05,
+      "loss": 4.3185,
+      "step": 1607
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.168564920273349e-05,
+      "loss": 3.3983,
+      "step": 1608
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1674259681093396e-05,
+      "loss": 4.1991,
+      "step": 1609
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1662870159453304e-05,
+      "loss": 3.559,
+      "step": 1610
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.165148063781321e-05,
+      "loss": 3.5935,
+      "step": 1611
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.164009111617312e-05,
+      "loss": 2.83,
+      "step": 1612
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1628701594533036e-05,
+      "loss": 4.2328,
+      "step": 1613
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161731207289294e-05,
+      "loss": 5.3789,
+      "step": 1614
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1605922551252846e-05,
+      "loss": 3.0781,
+      "step": 1615
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1594533029612754e-05,
+      "loss": 3.741,
+      "step": 1616
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.158314350797267e-05,
+      "loss": 4.8335,
+      "step": 1617
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157175398633258e-05,
+      "loss": 4.041,
+      "step": 1618
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1560364464692485e-05,
+      "loss": 3.4096,
+      "step": 1619
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.154897494305239e-05,
+      "loss": 1.9529,
+      "step": 1620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.15375854214123e-05,
+      "loss": 4.0962,
+      "step": 1621
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.152619589977221e-05,
+      "loss": 4.3004,
+      "step": 1622
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1514806378132125e-05,
+      "loss": 2.6927,
+      "step": 1623
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.150341685649203e-05,
+      "loss": 4.0756,
+      "step": 1624
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1492027334851934e-05,
+      "loss": 4.3546,
+      "step": 1625
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.148063781321184e-05,
+      "loss": 2.8441,
+      "step": 1626
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146924829157176e-05,
+      "loss": 2.1008,
+      "step": 1627
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1457858769931666e-05,
+      "loss": 4.8485,
+      "step": 1628
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1446469248291574e-05,
+      "loss": 3.7995,
+      "step": 1629
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.143507972665148e-05,
+      "loss": 3.2007,
+      "step": 1630
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142369020501139e-05,
+      "loss": 4.3346,
+      "step": 1631
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14123006833713e-05,
+      "loss": 3.17,
+      "step": 1632
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1400911161731214e-05,
+      "loss": 2.8998,
+      "step": 1633
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.138952164009112e-05,
+      "loss": 3.5112,
+      "step": 1634
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.137813211845102e-05,
+      "loss": 3.4854,
+      "step": 1635
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136674259681093e-05,
+      "loss": 2.8627,
+      "step": 1636
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1355353075170846e-05,
+      "loss": 4.5535,
+      "step": 1637
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1343963553530755e-05,
+      "loss": 4.2194,
+      "step": 1638
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.133257403189066e-05,
+      "loss": 4.704,
+      "step": 1639
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132118451025057e-05,
+      "loss": 4.4801,
+      "step": 1640
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130979498861048e-05,
+      "loss": 2.4141,
+      "step": 1641
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.129840546697039e-05,
+      "loss": 3.6155,
+      "step": 1642
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.12870159453303e-05,
+      "loss": 2.1188,
+      "step": 1643
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.127562642369021e-05,
+      "loss": 4.347,
+      "step": 1644
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126423690205011e-05,
+      "loss": 2.9106,
+      "step": 1645
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125284738041002e-05,
+      "loss": 3.967,
+      "step": 1646
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1241457858769935e-05,
+      "loss": 2.5394,
+      "step": 1647
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1230068337129843e-05,
+      "loss": 4.7577,
+      "step": 1648
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121867881548975e-05,
+      "loss": 3.8813,
+      "step": 1649
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120728929384966e-05,
+      "loss": 3.7168,
+      "step": 1650
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.119589977220957e-05,
+      "loss": 4.0295,
+      "step": 1651
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1184510250569476e-05,
+      "loss": 3.917,
+      "step": 1652
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117312072892939e-05,
+      "loss": 4.5727,
+      "step": 1653
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.11617312072893e-05,
+      "loss": 2.1691,
+      "step": 1654
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.11503416856492e-05,
+      "loss": 2.9885,
+      "step": 1655
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.113895216400911e-05,
+      "loss": 4.8093,
+      "step": 1656
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1127562642369024e-05,
+      "loss": 3.969,
+      "step": 1657
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111617312072893e-05,
+      "loss": 5.0399,
+      "step": 1658
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.110478359908884e-05,
+      "loss": 2.7362,
+      "step": 1659
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.109339407744875e-05,
+      "loss": 2.063,
+      "step": 1660
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.108200455580866e-05,
+      "loss": 2.2006,
+      "step": 1661
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1070615034168565e-05,
+      "loss": 2.7233,
+      "step": 1662
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105922551252847e-05,
+      "loss": 3.8622,
+      "step": 1663
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.104783599088839e-05,
+      "loss": 2.7839,
+      "step": 1664
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.103644646924829e-05,
+      "loss": 3.7543,
+      "step": 1665
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.10250569476082e-05,
+      "loss": 2.5621,
+      "step": 1666
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.101366742596811e-05,
+      "loss": 2.7158,
+      "step": 1667
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.100227790432802e-05,
+      "loss": 1.8935,
+      "step": 1668
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.099088838268793e-05,
+      "loss": 3.2518,
+      "step": 1669
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.097949886104784e-05,
+      "loss": 1.9469,
+      "step": 1670
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0968109339407746e-05,
+      "loss": 3.2486,
+      "step": 1671
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0956719817767654e-05,
+      "loss": 3.8925,
+      "step": 1672
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.094533029612756e-05,
+      "loss": 4.1475,
+      "step": 1673
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.093394077448748e-05,
+      "loss": 1.7796,
+      "step": 1674
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.092255125284738e-05,
+      "loss": 2.6905,
+      "step": 1675
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091116173120729e-05,
+      "loss": 3.9213,
+      "step": 1676
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.08997722095672e-05,
+      "loss": 2.6703,
+      "step": 1677
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.088838268792711e-05,
+      "loss": 3.6417,
+      "step": 1678
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.087699316628702e-05,
+      "loss": 3.5865,
+      "step": 1679
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0865603644646926e-05,
+      "loss": 3.5033,
+      "step": 1680
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0854214123006835e-05,
+      "loss": 4.2488,
+      "step": 1681
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.084282460136674e-05,
+      "loss": 3.6356,
+      "step": 1682
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.083143507972665e-05,
+      "loss": 3.4438,
+      "step": 1683
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0820045558086566e-05,
+      "loss": 3.2657,
+      "step": 1684
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080865603644647e-05,
+      "loss": 4.2209,
+      "step": 1685
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0797266514806376e-05,
+      "loss": 3.9741,
+      "step": 1686
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.078587699316629e-05,
+      "loss": 3.748,
+      "step": 1687
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.07744874715262e-05,
+      "loss": 2.81,
+      "step": 1688
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076309794988611e-05,
+      "loss": 1.8068,
+      "step": 1689
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0751708428246015e-05,
+      "loss": 4.6411,
+      "step": 1690
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.074031890660592e-05,
+      "loss": 2.6506,
+      "step": 1691
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.072892938496583e-05,
+      "loss": 4.4823,
+      "step": 1692
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071753986332574e-05,
+      "loss": 3.5161,
+      "step": 1693
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0706150341685655e-05,
+      "loss": 4.7268,
+      "step": 1694
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0694760820045556e-05,
+      "loss": 3.9321,
+      "step": 1695
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0683371298405464e-05,
+      "loss": 3.622,
+      "step": 1696
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.067198177676538e-05,
+      "loss": 4.5508,
+      "step": 1697
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066059225512529e-05,
+      "loss": 2.6535,
+      "step": 1698
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0649202733485196e-05,
+      "loss": 4.0988,
+      "step": 1699
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0637813211845104e-05,
+      "loss": 2.9467,
+      "step": 1700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.062642369020501e-05,
+      "loss": 3.302,
+      "step": 1701
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061503416856492e-05,
+      "loss": 2.6598,
+      "step": 1702
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060364464692483e-05,
+      "loss": 4.9405,
+      "step": 1703
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0592255125284744e-05,
+      "loss": 3.8981,
+      "step": 1704
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0580865603644645e-05,
+      "loss": 2.5359,
+      "step": 1705
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.056947608200455e-05,
+      "loss": 2.8922,
+      "step": 1706
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055808656036447e-05,
+      "loss": 2.2919,
+      "step": 1707
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0546697038724376e-05,
+      "loss": 3.0505,
+      "step": 1708
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0535307517084285e-05,
+      "loss": 4.5591,
+      "step": 1709
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052391799544419e-05,
+      "loss": 1.9845,
+      "step": 1710
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0512528473804104e-05,
+      "loss": 4.5254,
+      "step": 1711
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.050113895216401e-05,
+      "loss": 4.0582,
+      "step": 1712
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0489749430523917e-05,
+      "loss": 3.4664,
+      "step": 1713
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.047835990888383e-05,
+      "loss": 4.8229,
+      "step": 1714
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0466970387243737e-05,
+      "loss": 2.3077,
+      "step": 1715
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0455580865603645e-05,
+      "loss": 2.3055,
+      "step": 1716
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0444191343963557e-05,
+      "loss": 2.8004,
+      "step": 1717
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0432801822323465e-05,
+      "loss": 2.9396,
+      "step": 1718
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0421412300683373e-05,
+      "loss": 6.159,
+      "step": 1719
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0410022779043278e-05,
+      "loss": 3.0192,
+      "step": 1720
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0398633257403193e-05,
+      "loss": 2.9798,
+      "step": 1721
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0387243735763098e-05,
+      "loss": 4.0296,
+      "step": 1722
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0375854214123006e-05,
+      "loss": 2.6443,
+      "step": 1723
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0364464692482918e-05,
+      "loss": 4.2772,
+      "step": 1724
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0353075170842826e-05,
+      "loss": 4.0743,
+      "step": 1725
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0341685649202734e-05,
+      "loss": 4.2554,
+      "step": 1726
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0330296127562646e-05,
+      "loss": 1.6963,
+      "step": 1727
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0318906605922554e-05,
+      "loss": 4.0569,
+      "step": 1728
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0307517084282462e-05,
+      "loss": 2.9641,
+      "step": 1729
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0296127562642367e-05,
+      "loss": 4.0232,
+      "step": 1730
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0284738041002282e-05,
+      "loss": 3.8383,
+      "step": 1731
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0273348519362187e-05,
+      "loss": 1.6307,
+      "step": 1732
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0261958997722095e-05,
+      "loss": 2.9063,
+      "step": 1733
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0250569476082007e-05,
+      "loss": 5.2088,
+      "step": 1734
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0239179954441915e-05,
+      "loss": 4.0192,
+      "step": 1735
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0227790432801823e-05,
+      "loss": 3.5774,
+      "step": 1736
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0216400911161735e-05,
+      "loss": 1.7171,
+      "step": 1737
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0205011389521643e-05,
+      "loss": 3.1061,
+      "step": 1738
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.019362186788155e-05,
+      "loss": 1.7606,
+      "step": 1739
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0182232346241456e-05,
+      "loss": 4.082,
+      "step": 1740
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017084282460137e-05,
+      "loss": 4.1274,
+      "step": 1741
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0159453302961276e-05,
+      "loss": 2.9414,
+      "step": 1742
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0148063781321184e-05,
+      "loss": 1.9697,
+      "step": 1743
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0136674259681096e-05,
+      "loss": 4.2681,
+      "step": 1744
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0125284738041004e-05,
+      "loss": 4.1848,
+      "step": 1745
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0113895216400912e-05,
+      "loss": 3.4241,
+      "step": 1746
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0102505694760824e-05,
+      "loss": 2.3137,
+      "step": 1747
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0091116173120732e-05,
+      "loss": 3.0795,
+      "step": 1748
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.007972665148064e-05,
+      "loss": 4.0543,
+      "step": 1749
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0068337129840545e-05,
+      "loss": 2.0734,
+      "step": 1750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.005694760820046e-05,
+      "loss": 1.7059,
+      "step": 1751
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0045558086560365e-05,
+      "loss": 4.774,
+      "step": 1752
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0034168564920273e-05,
+      "loss": 3.8448,
+      "step": 1753
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0022779043280184e-05,
+      "loss": 1.589,
+      "step": 1754
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0011389521640093e-05,
+      "loss": 4.02,
+      "step": 1755
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3e-05,
+      "loss": 3.349,
+      "step": 1756
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.998861047835991e-05,
+      "loss": 3.063,
+      "step": 1757
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.997722095671982e-05,
+      "loss": 4.7281,
+      "step": 1758
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996583143507973e-05,
+      "loss": 2.682,
+      "step": 1759
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9954441913439634e-05,
+      "loss": 4.3727,
+      "step": 1760
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.994305239179955e-05,
+      "loss": 3.5759,
+      "step": 1761
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9931662870159453e-05,
+      "loss": 4.2703,
+      "step": 1762
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992027334851936e-05,
+      "loss": 2.7993,
+      "step": 1763
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9908883826879273e-05,
+      "loss": 6.4453,
+      "step": 1764
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.989749430523918e-05,
+      "loss": 2.0156,
+      "step": 1765
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.988610478359909e-05,
+      "loss": 3.7758,
+      "step": 1766
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9874715261958998e-05,
+      "loss": 3.2891,
+      "step": 1767
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986332574031891e-05,
+      "loss": 4.2409,
+      "step": 1768
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9851936218678818e-05,
+      "loss": 3.339,
+      "step": 1769
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9840546697038722e-05,
+      "loss": 2.4873,
+      "step": 1770
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9829157175398637e-05,
+      "loss": 4.2136,
+      "step": 1771
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9817767653758542e-05,
+      "loss": 4.0845,
+      "step": 1772
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.980637813211845e-05,
+      "loss": 2.986,
+      "step": 1773
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9794988610478365e-05,
+      "loss": 2.4726,
+      "step": 1774
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.978359908883827e-05,
+      "loss": 3.5047,
+      "step": 1775
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.977220956719818e-05,
+      "loss": 4.7481,
+      "step": 1776
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9760820045558087e-05,
+      "loss": 4.9307,
+      "step": 1777
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9749430523917998e-05,
+      "loss": 2.3753,
+      "step": 1778
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9738041002277906e-05,
+      "loss": 3.5283,
+      "step": 1779
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.972665148063781e-05,
+      "loss": 3.914,
+      "step": 1780
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9715261958997726e-05,
+      "loss": 2.7843,
+      "step": 1781
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970387243735763e-05,
+      "loss": 3.796,
+      "step": 1782
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.969248291571754e-05,
+      "loss": 3.1595,
+      "step": 1783
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9681093394077454e-05,
+      "loss": 4.2547,
+      "step": 1784
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966970387243736e-05,
+      "loss": 4.3356,
+      "step": 1785
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9658314350797267e-05,
+      "loss": 3.1174,
+      "step": 1786
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9646924829157175e-05,
+      "loss": 3.5504,
+      "step": 1787
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9635535307517087e-05,
+      "loss": 2.7798,
+      "step": 1788
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9624145785876995e-05,
+      "loss": 1.9911,
+      "step": 1789
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96127562642369e-05,
+      "loss": 2.9903,
+      "step": 1790
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9601366742596815e-05,
+      "loss": 4.5372,
+      "step": 1791
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9589977220956723e-05,
+      "loss": 4.4554,
+      "step": 1792
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9578587699316628e-05,
+      "loss": 3.5643,
+      "step": 1793
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9567198177676543e-05,
+      "loss": 3.6783,
+      "step": 1794
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9555808656036448e-05,
+      "loss": 3.6881,
+      "step": 1795
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9544419134396356e-05,
+      "loss": 3.0707,
+      "step": 1796
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9533029612756264e-05,
+      "loss": 2.3764,
+      "step": 1797
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9521640091116176e-05,
+      "loss": 4.1684,
+      "step": 1798
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9510250569476084e-05,
+      "loss": 4.5422,
+      "step": 1799
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.949886104783599e-05,
+      "loss": 4.1297,
+      "step": 1800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9487471526195904e-05,
+      "loss": 1.5166,
+      "step": 1801
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9476082004555812e-05,
+      "loss": 2.5589,
+      "step": 1802
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9464692482915717e-05,
+      "loss": 4.2999,
+      "step": 1803
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9453302961275632e-05,
+      "loss": 3.064,
+      "step": 1804
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9441913439635537e-05,
+      "loss": 2.6984,
+      "step": 1805
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9430523917995445e-05,
+      "loss": 2.0452,
+      "step": 1806
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9419134396355353e-05,
+      "loss": 2.2503,
+      "step": 1807
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9407744874715265e-05,
+      "loss": 3.3425,
+      "step": 1808
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9396355353075173e-05,
+      "loss": 3.2044,
+      "step": 1809
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.938496583143508e-05,
+      "loss": 4.2772,
+      "step": 1810
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9373576309794993e-05,
+      "loss": 1.2798,
+      "step": 1811
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.93621867881549e-05,
+      "loss": 3.6632,
+      "step": 1812
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9350797266514806e-05,
+      "loss": 3.0944,
+      "step": 1813
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9339407744874714e-05,
+      "loss": 3.3594,
+      "step": 1814
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9328018223234626e-05,
+      "loss": 3.2802,
+      "step": 1815
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9316628701594534e-05,
+      "loss": 3.753,
+      "step": 1816
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9305239179954442e-05,
+      "loss": 5.0069,
+      "step": 1817
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9293849658314354e-05,
+      "loss": 3.0088,
+      "step": 1818
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9282460136674262e-05,
+      "loss": 2.7332,
+      "step": 1819
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.927107061503417e-05,
+      "loss": 3.2023,
+      "step": 1820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.925968109339408e-05,
+      "loss": 2.6766,
+      "step": 1821
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.924829157175399e-05,
+      "loss": 3.0732,
+      "step": 1822
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9236902050113895e-05,
+      "loss": 3.4994,
+      "step": 1823
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9225512528473803e-05,
+      "loss": 2.318,
+      "step": 1824
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9214123006833714e-05,
+      "loss": 4.1539,
+      "step": 1825
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9202733485193623e-05,
+      "loss": 4.5437,
+      "step": 1826
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.919134396355353e-05,
+      "loss": 2.3476,
+      "step": 1827
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9179954441913442e-05,
+      "loss": 3.6018,
+      "step": 1828
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.916856492027335e-05,
+      "loss": 4.175,
+      "step": 1829
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.915717539863326e-05,
+      "loss": 5.0416,
+      "step": 1830
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.914578587699317e-05,
+      "loss": 3.8875,
+      "step": 1831
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.913439635535308e-05,
+      "loss": 3.3608,
+      "step": 1832
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9123006833712983e-05,
+      "loss": 5.8489,
+      "step": 1833
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911161731207289e-05,
+      "loss": 3.4899,
+      "step": 1834
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9100227790432803e-05,
+      "loss": 3.731,
+      "step": 1835
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.908883826879271e-05,
+      "loss": 3.0148,
+      "step": 1836
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.907744874715262e-05,
+      "loss": 2.5215,
+      "step": 1837
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906605922551253e-05,
+      "loss": 2.4846,
+      "step": 1838
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905466970387244e-05,
+      "loss": 3.419,
+      "step": 1839
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9043280182232348e-05,
+      "loss": 2.5534,
+      "step": 1840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.903189066059226e-05,
+      "loss": 3.1772,
+      "step": 1841
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9020501138952167e-05,
+      "loss": 5.1305,
+      "step": 1842
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9009111617312072e-05,
+      "loss": 3.1295,
+      "step": 1843
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.899772209567198e-05,
+      "loss": 2.4841,
+      "step": 1844
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.8986332574031892e-05,
+      "loss": 3.5073,
+      "step": 1845
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89749430523918e-05,
+      "loss": 4.1223,
+      "step": 1846
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.896355353075171e-05,
+      "loss": 2.6979,
+      "step": 1847
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895216400911162e-05,
+      "loss": 4.5321,
+      "step": 1848
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8940774487471528e-05,
+      "loss": 3.6009,
+      "step": 1849
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8929384965831436e-05,
+      "loss": 3.958,
+      "step": 1850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8917995444191348e-05,
+      "loss": 3.8344,
+      "step": 1851
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8906605922551256e-05,
+      "loss": 3.8941,
+      "step": 1852
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.889521640091116e-05,
+      "loss": 4.1577,
+      "step": 1853
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.888382687927107e-05,
+      "loss": 2.4114,
+      "step": 1854
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887243735763098e-05,
+      "loss": 2.8944,
+      "step": 1855
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886104783599089e-05,
+      "loss": 2.868,
+      "step": 1856
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8849658314350797e-05,
+      "loss": 2.9432,
+      "step": 1857
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.883826879271071e-05,
+      "loss": 3.0408,
+      "step": 1858
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8826879271070617e-05,
+      "loss": 4.175,
+      "step": 1859
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8815489749430525e-05,
+      "loss": 2.3572,
+      "step": 1860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.880410022779043e-05,
+      "loss": 3.3112,
+      "step": 1861
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8792710706150345e-05,
+      "loss": 4.264,
+      "step": 1862
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.878132118451025e-05,
+      "loss": 3.2389,
+      "step": 1863
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8769931662870158e-05,
+      "loss": 3.864,
+      "step": 1864
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8758542141230073e-05,
+      "loss": 5.0958,
+      "step": 1865
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8747152619589978e-05,
+      "loss": 3.424,
+      "step": 1866
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8735763097949886e-05,
+      "loss": 3.1079,
+      "step": 1867
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8724373576309798e-05,
+      "loss": 3.2803,
+      "step": 1868
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8712984054669706e-05,
+      "loss": 3.0576,
+      "step": 1869
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8701594533029614e-05,
+      "loss": 3.8393,
+      "step": 1870
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.869020501138952e-05,
+      "loss": 3.0707,
+      "step": 1871
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8678815489749434e-05,
+      "loss": 5.1583,
+      "step": 1872
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8667425968109342e-05,
+      "loss": 2.9999,
+      "step": 1873
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8656036446469247e-05,
+      "loss": 3.7726,
+      "step": 1874
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8644646924829162e-05,
+      "loss": 3.479,
+      "step": 1875
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8633257403189067e-05,
+      "loss": 2.2104,
+      "step": 1876
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8621867881548975e-05,
+      "loss": 4.1707,
+      "step": 1877
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8610478359908887e-05,
+      "loss": 4.4649,
+      "step": 1878
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8599088838268795e-05,
+      "loss": 1.2872,
+      "step": 1879
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8587699316628703e-05,
+      "loss": 3.5096,
+      "step": 1880
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8576309794988608e-05,
+      "loss": 4.2222,
+      "step": 1881
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8564920273348523e-05,
+      "loss": 4.2315,
+      "step": 1882
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855353075170843e-05,
+      "loss": 3.5118,
+      "step": 1883
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8542141230068336e-05,
+      "loss": 4.5485,
+      "step": 1884
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.853075170842825e-05,
+      "loss": 3.4801,
+      "step": 1885
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8519362186788156e-05,
+      "loss": 2.0104,
+      "step": 1886
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8507972665148064e-05,
+      "loss": 2.9333,
+      "step": 1887
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8496583143507975e-05,
+      "loss": 3.6804,
+      "step": 1888
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8485193621867884e-05,
+      "loss": 3.9311,
+      "step": 1889
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8473804100227792e-05,
+      "loss": 2.7446,
+      "step": 1890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.84624145785877e-05,
+      "loss": 4.4437,
+      "step": 1891
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845102505694761e-05,
+      "loss": 2.8179,
+      "step": 1892
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.843963553530752e-05,
+      "loss": 2.0533,
+      "step": 1893
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8428246013667425e-05,
+      "loss": 1.5821,
+      "step": 1894
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.841685649202734e-05,
+      "loss": 4.4017,
+      "step": 1895
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8405466970387244e-05,
+      "loss": 4.1255,
+      "step": 1896
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8394077448747153e-05,
+      "loss": 1.5041,
+      "step": 1897
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8382687927107064e-05,
+      "loss": 3.6021,
+      "step": 1898
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8371298405466972e-05,
+      "loss": 3.1163,
+      "step": 1899
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.835990888382688e-05,
+      "loss": 3.9066,
+      "step": 1900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.834851936218679e-05,
+      "loss": 1.752,
+      "step": 1901
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.83371298405467e-05,
+      "loss": 1.9983,
+      "step": 1902
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.832574031890661e-05,
+      "loss": 3.466,
+      "step": 1903
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8314350797266513e-05,
+      "loss": 3.7528,
+      "step": 1904
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.830296127562643e-05,
+      "loss": 3.5554,
+      "step": 1905
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8291571753986333e-05,
+      "loss": 2.0106,
+      "step": 1906
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.828018223234624e-05,
+      "loss": 2.9733,
+      "step": 1907
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.826879271070615e-05,
+      "loss": 4.1282,
+      "step": 1908
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.825740318906606e-05,
+      "loss": 3.9331,
+      "step": 1909
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.824601366742597e-05,
+      "loss": 5.1215,
+      "step": 1910
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8234624145785878e-05,
+      "loss": 2.7447,
+      "step": 1911
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.822323462414579e-05,
+      "loss": 2.9901,
+      "step": 1912
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8211845102505697e-05,
+      "loss": 4.254,
+      "step": 1913
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8200455580865602e-05,
+      "loss": 2.6313,
+      "step": 1914
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8189066059225517e-05,
+      "loss": 3.4485,
+      "step": 1915
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8177676537585422e-05,
+      "loss": 3.1867,
+      "step": 1916
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816628701594533e-05,
+      "loss": 4.3149,
+      "step": 1917
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.815489749430524e-05,
+      "loss": 3.3709,
+      "step": 1918
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.814350797266515e-05,
+      "loss": 3.22,
+      "step": 1919
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8132118451025058e-05,
+      "loss": 1.977,
+      "step": 1920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8120728929384967e-05,
+      "loss": 3.9805,
+      "step": 1921
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8109339407744878e-05,
+      "loss": 3.9829,
+      "step": 1922
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8097949886104786e-05,
+      "loss": 2.6831,
+      "step": 1923
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.808656036446469e-05,
+      "loss": 4.5905,
+      "step": 1924
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8075170842824606e-05,
+      "loss": 3.4316,
+      "step": 1925
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806378132118451e-05,
+      "loss": 2.8954,
+      "step": 1926
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805239179954442e-05,
+      "loss": 3.9408,
+      "step": 1927
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8041002277904327e-05,
+      "loss": 1.773,
+      "step": 1928
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.802961275626424e-05,
+      "loss": 3.9143,
+      "step": 1929
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8018223234624147e-05,
+      "loss": 4.5582,
+      "step": 1930
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8006833712984055e-05,
+      "loss": 3.0305,
+      "step": 1931
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.7995444191343967e-05,
+      "loss": 2.0888,
+      "step": 1932
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.7984054669703875e-05,
+      "loss": 3.9947,
+      "step": 1933
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797266514806378e-05,
+      "loss": 2.7701,
+      "step": 1934
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7961275626423695e-05,
+      "loss": 3.8113,
+      "step": 1935
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.79498861047836e-05,
+      "loss": 2.3374,
+      "step": 1936
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7938496583143508e-05,
+      "loss": 2.9015,
+      "step": 1937
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7927107061503416e-05,
+      "loss": 4.2044,
+      "step": 1938
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7915717539863328e-05,
+      "loss": 2.3334,
+      "step": 1939
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7904328018223236e-05,
+      "loss": 2.7124,
+      "step": 1940
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7892938496583144e-05,
+      "loss": 2.9748,
+      "step": 1941
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7881548974943056e-05,
+      "loss": 4.3685,
+      "step": 1942
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7870159453302964e-05,
+      "loss": 1.9496,
+      "step": 1943
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.785876993166287e-05,
+      "loss": 4.4861,
+      "step": 1944
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7847380410022784e-05,
+      "loss": 2.8892,
+      "step": 1945
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7835990888382692e-05,
+      "loss": 3.7099,
+      "step": 1946
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7824601366742597e-05,
+      "loss": 2.5168,
+      "step": 1947
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7813211845102505e-05,
+      "loss": 4.4188,
+      "step": 1948
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7801822323462417e-05,
+      "loss": 3.9717,
+      "step": 1949
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7790432801822325e-05,
+      "loss": 1.0952,
+      "step": 1950
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7779043280182233e-05,
+      "loss": 3.4218,
+      "step": 1951
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7767653758542145e-05,
+      "loss": 3.397,
+      "step": 1952
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7756264236902053e-05,
+      "loss": 2.0651,
+      "step": 1953
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7744874715261958e-05,
+      "loss": 4.9018,
+      "step": 1954
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7733485193621866e-05,
+      "loss": 4.1579,
+      "step": 1955
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772209567198178e-05,
+      "loss": 3.9543,
+      "step": 1956
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7710706150341686e-05,
+      "loss": 3.2913,
+      "step": 1957
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7699316628701594e-05,
+      "loss": 3.3567,
+      "step": 1958
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7687927107061505e-05,
+      "loss": 3.3322,
+      "step": 1959
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7676537585421414e-05,
+      "loss": 4.5983,
+      "step": 1960
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7665148063781322e-05,
+      "loss": 4.9845,
+      "step": 1961
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7653758542141233e-05,
+      "loss": 3.1109,
+      "step": 1962
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.764236902050114e-05,
+      "loss": 3.4181,
+      "step": 1963
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.763097949886105e-05,
+      "loss": 4.0905,
+      "step": 1964
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7619589977220955e-05,
+      "loss": 2.5504,
+      "step": 1965
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.760820045558087e-05,
+      "loss": 4.5646,
+      "step": 1966
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7596810933940774e-05,
+      "loss": 4.2651,
+      "step": 1967
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7585421412300683e-05,
+      "loss": 2.373,
+      "step": 1968
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7574031890660594e-05,
+      "loss": 2.7952,
+      "step": 1969
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7562642369020502e-05,
+      "loss": 3.0963,
+      "step": 1970
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755125284738041e-05,
+      "loss": 3.851,
+      "step": 1971
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7539863325740322e-05,
+      "loss": 3.7139,
+      "step": 1972
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.752847380410023e-05,
+      "loss": 3.4962,
+      "step": 1973
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751708428246014e-05,
+      "loss": 2.0681,
+      "step": 1974
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7505694760820043e-05,
+      "loss": 3.0548,
+      "step": 1975
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.749430523917996e-05,
+      "loss": 5.6989,
+      "step": 1976
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7482915717539863e-05,
+      "loss": 3.5432,
+      "step": 1977
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.747152619589977e-05,
+      "loss": 4.3594,
+      "step": 1978
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7460136674259683e-05,
+      "loss": 4.5852,
+      "step": 1979
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.744874715261959e-05,
+      "loss": 3.165,
+      "step": 1980
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74373576309795e-05,
+      "loss": 4.5615,
+      "step": 1981
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.742596810933941e-05,
+      "loss": 2.1827,
+      "step": 1982
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741457858769932e-05,
+      "loss": 3.5452,
+      "step": 1983
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7403189066059228e-05,
+      "loss": 3.6689,
+      "step": 1984
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7391799544419132e-05,
+      "loss": 2.9616,
+      "step": 1985
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7380410022779047e-05,
+      "loss": 4.2696,
+      "step": 1986
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7369020501138952e-05,
+      "loss": 2.3291,
+      "step": 1987
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.735763097949886e-05,
+      "loss": 2.6393,
+      "step": 1988
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7346241457858772e-05,
+      "loss": 3.9905,
+      "step": 1989
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.733485193621868e-05,
+      "loss": 3.1473,
+      "step": 1990
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.732346241457859e-05,
+      "loss": 3.5733,
+      "step": 1991
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.73120728929385e-05,
+      "loss": 3.527,
+      "step": 1992
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7300683371298408e-05,
+      "loss": 2.8304,
+      "step": 1993
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7289293849658316e-05,
+      "loss": 4.6614,
+      "step": 1994
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.727790432801822e-05,
+      "loss": 1.3204,
+      "step": 1995
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7266514806378136e-05,
+      "loss": 2.8552,
+      "step": 1996
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725512528473804e-05,
+      "loss": 3.9506,
+      "step": 1997
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.724373576309795e-05,
+      "loss": 4.8598,
+      "step": 1998
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.723234624145786e-05,
+      "loss": 3.6281,
+      "step": 1999
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722095671981777e-05,
+      "loss": 3.7816,
+      "step": 2000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7209567198177677e-05,
+      "loss": 1.8309,
+      "step": 2001
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7198177676537585e-05,
+      "loss": 4.4625,
+      "step": 2002
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7186788154897497e-05,
+      "loss": 4.1164,
+      "step": 2003
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7175398633257405e-05,
+      "loss": 3.5947,
+      "step": 2004
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716400911161731e-05,
+      "loss": 1.9769,
+      "step": 2005
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7152619589977225e-05,
+      "loss": 2.2074,
+      "step": 2006
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.714123006833713e-05,
+      "loss": 1.8324,
+      "step": 2007
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7129840546697038e-05,
+      "loss": 3.195,
+      "step": 2008
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711845102505695e-05,
+      "loss": 4.4578,
+      "step": 2009
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7107061503416858e-05,
+      "loss": 4.1396,
+      "step": 2010
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7095671981776766e-05,
+      "loss": 3.9212,
+      "step": 2011
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7084282460136674e-05,
+      "loss": 3.9623,
+      "step": 2012
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7072892938496586e-05,
+      "loss": 3.9507,
+      "step": 2013
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7061503416856494e-05,
+      "loss": 2.837,
+      "step": 2014
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.70501138952164e-05,
+      "loss": 2.5455,
+      "step": 2015
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7038724373576314e-05,
+      "loss": 2.813,
+      "step": 2016
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.702733485193622e-05,
+      "loss": 3.6521,
+      "step": 2017
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7015945330296127e-05,
+      "loss": 2.9811,
+      "step": 2018
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7004555808656042e-05,
+      "loss": 3.2062,
+      "step": 2019
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6993166287015947e-05,
+      "loss": 1.3645,
+      "step": 2020
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6981776765375855e-05,
+      "loss": 3.7327,
+      "step": 2021
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6970387243735763e-05,
+      "loss": 3.5542,
+      "step": 2022
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6958997722095675e-05,
+      "loss": 5.0417,
+      "step": 2023
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6947608200455583e-05,
+      "loss": 4.9188,
+      "step": 2024
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6936218678815488e-05,
+      "loss": 4.496,
+      "step": 2025
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6924829157175403e-05,
+      "loss": 4.8797,
+      "step": 2026
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6913439635535307e-05,
+      "loss": 3.561,
+      "step": 2027
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6902050113895216e-05,
+      "loss": 1.9303,
+      "step": 2028
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.689066059225513e-05,
+      "loss": 4.4183,
+      "step": 2029
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6879271070615035e-05,
+      "loss": 2.601,
+      "step": 2030
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6867881548974944e-05,
+      "loss": 3.1475,
+      "step": 2031
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6856492027334852e-05,
+      "loss": 3.7333,
+      "step": 2032
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6845102505694763e-05,
+      "loss": 2.9115,
+      "step": 2033
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.683371298405467e-05,
+      "loss": 2.8021,
+      "step": 2034
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6822323462414576e-05,
+      "loss": 3.1039,
+      "step": 2035
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.681093394077449e-05,
+      "loss": 3.5087,
+      "step": 2036
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.67995444191344e-05,
+      "loss": 2.8079,
+      "step": 2037
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6788154897494304e-05,
+      "loss": 2.803,
+      "step": 2038
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.677676537585422e-05,
+      "loss": 2.082,
+      "step": 2039
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6765375854214124e-05,
+      "loss": 3.7184,
+      "step": 2040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6753986332574032e-05,
+      "loss": 2.9669,
+      "step": 2041
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.674259681093394e-05,
+      "loss": 3.6613,
+      "step": 2042
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6731207289293852e-05,
+      "loss": 3.8572,
+      "step": 2043
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671981776765376e-05,
+      "loss": 1.3859,
+      "step": 2044
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6708428246013665e-05,
+      "loss": 3.4015,
+      "step": 2045
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.669703872437358e-05,
+      "loss": 2.0688,
+      "step": 2046
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.668564920273349e-05,
+      "loss": 1.5228,
+      "step": 2047
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6674259681093393e-05,
+      "loss": 3.0594,
+      "step": 2048
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.66628701594533e-05,
+      "loss": 2.9318,
+      "step": 2049
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6651480637813213e-05,
+      "loss": 3.7698,
+      "step": 2050
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.664009111617312e-05,
+      "loss": 2.3144,
+      "step": 2051
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.662870159453303e-05,
+      "loss": 3.769,
+      "step": 2052
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661731207289294e-05,
+      "loss": 2.3544,
+      "step": 2053
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660592255125285e-05,
+      "loss": 3.1658,
+      "step": 2054
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6594533029612758e-05,
+      "loss": 2.96,
+      "step": 2055
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.658314350797267e-05,
+      "loss": 4.5067,
+      "step": 2056
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6571753986332577e-05,
+      "loss": 4.3593,
+      "step": 2057
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6560364464692482e-05,
+      "loss": 4.7264,
+      "step": 2058
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.654897494305239e-05,
+      "loss": 3.7575,
+      "step": 2059
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6537585421412302e-05,
+      "loss": 3.2755,
+      "step": 2060
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.652619589977221e-05,
+      "loss": 5.2212,
+      "step": 2061
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651480637813212e-05,
+      "loss": 3.0214,
+      "step": 2062
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650341685649203e-05,
+      "loss": 1.8158,
+      "step": 2063
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6492027334851938e-05,
+      "loss": 2.5674,
+      "step": 2064
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6480637813211846e-05,
+      "loss": 2.667,
+      "step": 2065
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6469248291571758e-05,
+      "loss": 2.0244,
+      "step": 2066
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6457858769931666e-05,
+      "loss": 4.7039,
+      "step": 2067
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.644646924829157e-05,
+      "loss": 2.7653,
+      "step": 2068
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.643507972665148e-05,
+      "loss": 3.9848,
+      "step": 2069
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.642369020501139e-05,
+      "loss": 5.4409,
+      "step": 2070
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.64123006833713e-05,
+      "loss": 3.2843,
+      "step": 2071
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6400911161731207e-05,
+      "loss": 2.8262,
+      "step": 2072
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.638952164009112e-05,
+      "loss": 2.9346,
+      "step": 2073
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6378132118451027e-05,
+      "loss": 1.7742,
+      "step": 2074
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6366742596810935e-05,
+      "loss": 3.7951,
+      "step": 2075
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6355353075170847e-05,
+      "loss": 3.0544,
+      "step": 2076
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6343963553530755e-05,
+      "loss": 3.5141,
+      "step": 2077
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.633257403189066e-05,
+      "loss": 3.5251,
+      "step": 2078
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6321184510250568e-05,
+      "loss": 2.8763,
+      "step": 2079
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.630979498861048e-05,
+      "loss": 3.3469,
+      "step": 2080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6298405466970388e-05,
+      "loss": 3.4961,
+      "step": 2081
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6287015945330296e-05,
+      "loss": 3.6414,
+      "step": 2082
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6275626423690208e-05,
+      "loss": 2.7016,
+      "step": 2083
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6264236902050116e-05,
+      "loss": 1.6544,
+      "step": 2084
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6252847380410024e-05,
+      "loss": 4.6043,
+      "step": 2085
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6241457858769936e-05,
+      "loss": 4.7468,
+      "step": 2086
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6230068337129844e-05,
+      "loss": 4.6032,
+      "step": 2087
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.621867881548975e-05,
+      "loss": 3.4272,
+      "step": 2088
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6207289293849657e-05,
+      "loss": 3.8968,
+      "step": 2089
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.619589977220957e-05,
+      "loss": 3.9014,
+      "step": 2090
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6184510250569477e-05,
+      "loss": 4.1688,
+      "step": 2091
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6173120728929385e-05,
+      "loss": 4.0998,
+      "step": 2092
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6161731207289296e-05,
+      "loss": 4.1357,
+      "step": 2093
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6150341685649205e-05,
+      "loss": 4.5315,
+      "step": 2094
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6138952164009113e-05,
+      "loss": 2.8546,
+      "step": 2095
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6127562642369024e-05,
+      "loss": 4.1192,
+      "step": 2096
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6116173120728933e-05,
+      "loss": 3.8682,
+      "step": 2097
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6104783599088837e-05,
+      "loss": 3.7005,
+      "step": 2098
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6093394077448746e-05,
+      "loss": 5.1105,
+      "step": 2099
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.608200455580866e-05,
+      "loss": 2.219,
+      "step": 2100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6070615034168565e-05,
+      "loss": 3.5035,
+      "step": 2101
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6059225512528474e-05,
+      "loss": 4.1917,
+      "step": 2102
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6047835990888385e-05,
+      "loss": 3.9892,
+      "step": 2103
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6036446469248293e-05,
+      "loss": 3.3222,
+      "step": 2104
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6025056947608202e-05,
+      "loss": 4.236,
+      "step": 2105
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6013667425968107e-05,
+      "loss": 2.5865,
+      "step": 2106
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600227790432802e-05,
+      "loss": 3.0681,
+      "step": 2107
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.5990888382687926e-05,
+      "loss": 2.3699,
+      "step": 2108
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.5979498861047835e-05,
+      "loss": 2.9721,
+      "step": 2109
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.596810933940775e-05,
+      "loss": 3.7534,
+      "step": 2110
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5956719817767654e-05,
+      "loss": 3.806,
+      "step": 2111
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5945330296127563e-05,
+      "loss": 2.1768,
+      "step": 2112
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5933940774487474e-05,
+      "loss": 4.469,
+      "step": 2113
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5922551252847382e-05,
+      "loss": 3.2825,
+      "step": 2114
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.591116173120729e-05,
+      "loss": 3.0413,
+      "step": 2115
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5899772209567195e-05,
+      "loss": 4.6734,
+      "step": 2116
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.588838268792711e-05,
+      "loss": 1.8513,
+      "step": 2117
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.587699316628702e-05,
+      "loss": 4.2285,
+      "step": 2118
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5865603644646923e-05,
+      "loss": 3.3378,
+      "step": 2119
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585421412300684e-05,
+      "loss": 3.374,
+      "step": 2120
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5842824601366743e-05,
+      "loss": 2.7069,
+      "step": 2121
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.583143507972665e-05,
+      "loss": 1.5329,
+      "step": 2122
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5820045558086563e-05,
+      "loss": 3.4548,
+      "step": 2123
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580865603644647e-05,
+      "loss": 5.7696,
+      "step": 2124
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.579726651480638e-05,
+      "loss": 3.4131,
+      "step": 2125
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5785876993166284e-05,
+      "loss": 2.7706,
+      "step": 2126
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.57744874715262e-05,
+      "loss": 3.2774,
+      "step": 2127
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5763097949886107e-05,
+      "loss": 5.6206,
+      "step": 2128
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5751708428246012e-05,
+      "loss": 4.7283,
+      "step": 2129
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5740318906605927e-05,
+      "loss": 3.308,
+      "step": 2130
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5728929384965832e-05,
+      "loss": 3.9242,
+      "step": 2131
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.571753986332574e-05,
+      "loss": 3.7497,
+      "step": 2132
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5706150341685652e-05,
+      "loss": 3.4545,
+      "step": 2133
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.569476082004556e-05,
+      "loss": 2.9617,
+      "step": 2134
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5683371298405468e-05,
+      "loss": 3.4748,
+      "step": 2135
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5671981776765376e-05,
+      "loss": 2.519,
+      "step": 2136
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5660592255125288e-05,
+      "loss": 1.0792,
+      "step": 2137
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5649202733485196e-05,
+      "loss": 3.0809,
+      "step": 2138
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.56378132118451e-05,
+      "loss": 2.7808,
+      "step": 2139
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5626423690205016e-05,
+      "loss": 3.9111,
+      "step": 2140
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.561503416856492e-05,
+      "loss": 3.6042,
+      "step": 2141
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560364464692483e-05,
+      "loss": 5.3414,
+      "step": 2142
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.559225512528474e-05,
+      "loss": 3.5515,
+      "step": 2143
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.558086560364465e-05,
+      "loss": 3.5679,
+      "step": 2144
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5569476082004557e-05,
+      "loss": 2.8639,
+      "step": 2145
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5558086560364465e-05,
+      "loss": 4.6879,
+      "step": 2146
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5546697038724377e-05,
+      "loss": 4.2953,
+      "step": 2147
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5535307517084285e-05,
+      "loss": 4.6155,
+      "step": 2148
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.552391799544419e-05,
+      "loss": 3.7888,
+      "step": 2149
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5512528473804105e-05,
+      "loss": 3.3074,
+      "step": 2150
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550113895216401e-05,
+      "loss": 4.2218,
+      "step": 2151
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5489749430523918e-05,
+      "loss": 2.7893,
+      "step": 2152
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5478359908883826e-05,
+      "loss": 2.0562,
+      "step": 2153
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5466970387243738e-05,
+      "loss": 2.8348,
+      "step": 2154
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5455580865603646e-05,
+      "loss": 3.6517,
+      "step": 2155
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5444191343963554e-05,
+      "loss": 3.3921,
+      "step": 2156
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5432801822323466e-05,
+      "loss": 2.7105,
+      "step": 2157
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5421412300683374e-05,
+      "loss": 3.8214,
+      "step": 2158
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541002277904328e-05,
+      "loss": 3.0982,
+      "step": 2159
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5398633257403194e-05,
+      "loss": 2.4284,
+      "step": 2160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.53872437357631e-05,
+      "loss": 2.4083,
+      "step": 2161
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5375854214123007e-05,
+      "loss": 3.9425,
+      "step": 2162
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5364464692482915e-05,
+      "loss": 4.2461,
+      "step": 2163
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5353075170842826e-05,
+      "loss": 1.6375,
+      "step": 2164
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5341685649202735e-05,
+      "loss": 3.2849,
+      "step": 2165
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5330296127562643e-05,
+      "loss": 4.2099,
+      "step": 2166
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5318906605922554e-05,
+      "loss": 3.2833,
+      "step": 2167
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5307517084282463e-05,
+      "loss": 2.8904,
+      "step": 2168
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5296127562642368e-05,
+      "loss": 3.4482,
+      "step": 2169
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5284738041002283e-05,
+      "loss": 3.4706,
+      "step": 2170
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5273348519362187e-05,
+      "loss": 2.9812,
+      "step": 2171
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5261958997722096e-05,
+      "loss": 2.6741,
+      "step": 2172
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5250569476082004e-05,
+      "loss": 3.1941,
+      "step": 2173
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5239179954441915e-05,
+      "loss": 4.5775,
+      "step": 2174
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5227790432801824e-05,
+      "loss": 2.856,
+      "step": 2175
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5216400911161732e-05,
+      "loss": 3.4905,
+      "step": 2176
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5205011389521643e-05,
+      "loss": 2.7788,
+      "step": 2177
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.519362186788155e-05,
+      "loss": 4.8416,
+      "step": 2178
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5182232346241456e-05,
+      "loss": 2.761,
+      "step": 2179
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.517084282460137e-05,
+      "loss": 3.7016,
+      "step": 2180
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5159453302961276e-05,
+      "loss": 3.291,
+      "step": 2181
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5148063781321184e-05,
+      "loss": 4.8477,
+      "step": 2182
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5136674259681093e-05,
+      "loss": 2.6204,
+      "step": 2183
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5125284738041004e-05,
+      "loss": 3.4144,
+      "step": 2184
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5113895216400912e-05,
+      "loss": 3.8525,
+      "step": 2185
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.510250569476082e-05,
+      "loss": 2.7143,
+      "step": 2186
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5091116173120732e-05,
+      "loss": 3.4165,
+      "step": 2187
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.507972665148064e-05,
+      "loss": 4.299,
+      "step": 2188
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5068337129840545e-05,
+      "loss": 2.8785,
+      "step": 2189
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.505694760820046e-05,
+      "loss": 4.6585,
+      "step": 2190
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.504555808656037e-05,
+      "loss": 3.5849,
+      "step": 2191
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5034168564920273e-05,
+      "loss": 2.2054,
+      "step": 2192
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.502277904328018e-05,
+      "loss": 1.2294,
+      "step": 2193
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5011389521640093e-05,
+      "loss": 4.6958,
+      "step": 2194
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5e-05,
+      "loss": 2.3682,
+      "step": 2195
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4988610478359913e-05,
+      "loss": 3.5542,
+      "step": 2196
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4977220956719818e-05,
+      "loss": 3.2124,
+      "step": 2197
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.496583143507973e-05,
+      "loss": 3.2631,
+      "step": 2198
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4954441913439634e-05,
+      "loss": 4.4021,
+      "step": 2199
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4943052391799546e-05,
+      "loss": 2.2227,
+      "step": 2200
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4931662870159457e-05,
+      "loss": 3.8802,
+      "step": 2201
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4920273348519362e-05,
+      "loss": 3.6588,
+      "step": 2202
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4908883826879274e-05,
+      "loss": 4.0166,
+      "step": 2203
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4897494305239182e-05,
+      "loss": 3.213,
+      "step": 2204
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.488610478359909e-05,
+      "loss": 2.9406,
+      "step": 2205
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4874715261959e-05,
+      "loss": 3.6935,
+      "step": 2206
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4863325740318906e-05,
+      "loss": 3.3173,
+      "step": 2207
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4851936218678818e-05,
+      "loss": 1.7156,
+      "step": 2208
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4840546697038726e-05,
+      "loss": 2.8198,
+      "step": 2209
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4829157175398634e-05,
+      "loss": 2.9913,
+      "step": 2210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4817767653758546e-05,
+      "loss": 3.4118,
+      "step": 2211
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.480637813211845e-05,
+      "loss": 2.9119,
+      "step": 2212
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4794988610478362e-05,
+      "loss": 2.4402,
+      "step": 2213
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.478359908883827e-05,
+      "loss": 2.0643,
+      "step": 2214
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.477220956719818e-05,
+      "loss": 4.6431,
+      "step": 2215
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.476082004555809e-05,
+      "loss": 2.7375,
+      "step": 2216
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4749430523917995e-05,
+      "loss": 4.4466,
+      "step": 2217
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4738041002277907e-05,
+      "loss": 2.3245,
+      "step": 2218
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4726651480637815e-05,
+      "loss": 2.0501,
+      "step": 2219
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4715261958997723e-05,
+      "loss": 2.4885,
+      "step": 2220
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.470387243735763e-05,
+      "loss": 3.7398,
+      "step": 2221
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.469248291571754e-05,
+      "loss": 4.2208,
+      "step": 2222
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.468109339407745e-05,
+      "loss": 3.1782,
+      "step": 2223
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.466970387243736e-05,
+      "loss": 3.43,
+      "step": 2224
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4658314350797268e-05,
+      "loss": 3.5663,
+      "step": 2225
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4646924829157176e-05,
+      "loss": 2.9184,
+      "step": 2226
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4635535307517084e-05,
+      "loss": 2.7097,
+      "step": 2227
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4624145785876996e-05,
+      "loss": 2.4479,
+      "step": 2228
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4612756264236904e-05,
+      "loss": 2.2454,
+      "step": 2229
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4601366742596812e-05,
+      "loss": 3.0021,
+      "step": 2230
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.458997722095672e-05,
+      "loss": 2.6,
+      "step": 2231
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.457858769931663e-05,
+      "loss": 2.9615,
+      "step": 2232
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.456719817767654e-05,
+      "loss": 2.6591,
+      "step": 2233
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.455580865603645e-05,
+      "loss": 2.9457,
+      "step": 2234
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4544419134396357e-05,
+      "loss": 3.3863,
+      "step": 2235
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4533029612756265e-05,
+      "loss": 2.5566,
+      "step": 2236
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4521640091116173e-05,
+      "loss": 1.9373,
+      "step": 2237
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4510250569476085e-05,
+      "loss": 3.5465,
+      "step": 2238
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4498861047835993e-05,
+      "loss": 3.9035,
+      "step": 2239
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.44874715261959e-05,
+      "loss": 2.4625,
+      "step": 2240
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.447608200455581e-05,
+      "loss": 4.6833,
+      "step": 2241
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4464692482915717e-05,
+      "loss": 3.2507,
+      "step": 2242
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.445330296127563e-05,
+      "loss": 2.3157,
+      "step": 2243
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4441913439635537e-05,
+      "loss": 2.7658,
+      "step": 2244
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4430523917995445e-05,
+      "loss": 3.6241,
+      "step": 2245
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4419134396355354e-05,
+      "loss": 3.3321,
+      "step": 2246
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4407744874715262e-05,
+      "loss": 3.0748,
+      "step": 2247
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4396355353075173e-05,
+      "loss": 2.1909,
+      "step": 2248
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.438496583143508e-05,
+      "loss": 3.4375,
+      "step": 2249
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.437357630979499e-05,
+      "loss": 3.0135,
+      "step": 2250
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4362186788154898e-05,
+      "loss": 2.6746,
+      "step": 2251
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4350797266514806e-05,
+      "loss": 4.5584,
+      "step": 2252
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4339407744874718e-05,
+      "loss": 5.3308,
+      "step": 2253
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4328018223234626e-05,
+      "loss": 4.5514,
+      "step": 2254
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4316628701594534e-05,
+      "loss": 2.7666,
+      "step": 2255
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4305239179954442e-05,
+      "loss": 4.5807,
+      "step": 2256
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.429384965831435e-05,
+      "loss": 2.4827,
+      "step": 2257
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4282460136674262e-05,
+      "loss": 5.2867,
+      "step": 2258
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.427107061503417e-05,
+      "loss": 4.4729,
+      "step": 2259
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.425968109339408e-05,
+      "loss": 2.7735,
+      "step": 2260
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4248291571753987e-05,
+      "loss": 2.3093,
+      "step": 2261
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4236902050113895e-05,
+      "loss": 4.4055,
+      "step": 2262
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4225512528473807e-05,
+      "loss": 3.6136,
+      "step": 2263
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4214123006833715e-05,
+      "loss": 2.7924,
+      "step": 2264
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4202733485193623e-05,
+      "loss": 3.3523,
+      "step": 2265
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.419134396355353e-05,
+      "loss": 3.5612,
+      "step": 2266
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.417995444191344e-05,
+      "loss": 2.0343,
+      "step": 2267
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.416856492027335e-05,
+      "loss": 3.3786,
+      "step": 2268
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.415717539863326e-05,
+      "loss": 4.0351,
+      "step": 2269
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4145785876993167e-05,
+      "loss": 3.443,
+      "step": 2270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4134396355353076e-05,
+      "loss": 2.4399,
+      "step": 2271
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4123006833712984e-05,
+      "loss": 3.9378,
+      "step": 2272
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4111617312072892e-05,
+      "loss": 1.3919,
+      "step": 2273
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4100227790432804e-05,
+      "loss": 2.2872,
+      "step": 2274
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4088838268792712e-05,
+      "loss": 3.645,
+      "step": 2275
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.407744874715262e-05,
+      "loss": 1.6136,
+      "step": 2276
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.406605922551253e-05,
+      "loss": 2.6567,
+      "step": 2277
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4054669703872436e-05,
+      "loss": 4.9626,
+      "step": 2278
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4043280182232348e-05,
+      "loss": 2.8333,
+      "step": 2279
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4031890660592256e-05,
+      "loss": 4.2293,
+      "step": 2280
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4020501138952164e-05,
+      "loss": 2.0164,
+      "step": 2281
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4009111617312076e-05,
+      "loss": 3.6245,
+      "step": 2282
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.399772209567198e-05,
+      "loss": 0.8862,
+      "step": 2283
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.3986332574031892e-05,
+      "loss": 4.3777,
+      "step": 2284
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.39749430523918e-05,
+      "loss": 3.9595,
+      "step": 2285
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.396355353075171e-05,
+      "loss": 4.0484,
+      "step": 2286
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.395216400911162e-05,
+      "loss": 3.2071,
+      "step": 2287
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3940774487471525e-05,
+      "loss": 3.8764,
+      "step": 2288
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3929384965831437e-05,
+      "loss": 3.5917,
+      "step": 2289
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3917995444191345e-05,
+      "loss": 4.46,
+      "step": 2290
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3906605922551253e-05,
+      "loss": 3.8877,
+      "step": 2291
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3895216400911165e-05,
+      "loss": 2.2216,
+      "step": 2292
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.388382687927107e-05,
+      "loss": 2.4608,
+      "step": 2293
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.387243735763098e-05,
+      "loss": 3.2715,
+      "step": 2294
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.386104783599089e-05,
+      "loss": 1.1803,
+      "step": 2295
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3849658314350798e-05,
+      "loss": 3.259,
+      "step": 2296
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.383826879271071e-05,
+      "loss": 3.4562,
+      "step": 2297
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3826879271070614e-05,
+      "loss": 7.0273,
+      "step": 2298
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3815489749430526e-05,
+      "loss": 4.7608,
+      "step": 2299
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3804100227790434e-05,
+      "loss": 2.4251,
+      "step": 2300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3792710706150342e-05,
+      "loss": 3.2751,
+      "step": 2301
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3781321184510254e-05,
+      "loss": 2.8581,
+      "step": 2302
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.376993166287016e-05,
+      "loss": 3.872,
+      "step": 2303
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.375854214123007e-05,
+      "loss": 3.5304,
+      "step": 2304
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.374715261958998e-05,
+      "loss": 3.5479,
+      "step": 2305
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3735763097949887e-05,
+      "loss": 3.4617,
+      "step": 2306
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3724373576309798e-05,
+      "loss": 3.332,
+      "step": 2307
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3712984054669703e-05,
+      "loss": 2.7533,
+      "step": 2308
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3701594533029615e-05,
+      "loss": 2.2714,
+      "step": 2309
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3690205011389523e-05,
+      "loss": 3.3176,
+      "step": 2310
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.367881548974943e-05,
+      "loss": 4.2153,
+      "step": 2311
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3667425968109343e-05,
+      "loss": 3.0795,
+      "step": 2312
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3656036446469247e-05,
+      "loss": 2.3043,
+      "step": 2313
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.364464692482916e-05,
+      "loss": 1.2881,
+      "step": 2314
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3633257403189067e-05,
+      "loss": 2.3284,
+      "step": 2315
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3621867881548975e-05,
+      "loss": 4.1919,
+      "step": 2316
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3610478359908887e-05,
+      "loss": 2.0623,
+      "step": 2317
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3599088838268792e-05,
+      "loss": 4.8358,
+      "step": 2318
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3587699316628703e-05,
+      "loss": 4.3844,
+      "step": 2319
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.357630979498861e-05,
+      "loss": 2.9953,
+      "step": 2320
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.356492027334852e-05,
+      "loss": 3.4167,
+      "step": 2321
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.355353075170843e-05,
+      "loss": 4.2951,
+      "step": 2322
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3542141230068336e-05,
+      "loss": 2.3005,
+      "step": 2323
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3530751708428248e-05,
+      "loss": 2.6828,
+      "step": 2324
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3519362186788156e-05,
+      "loss": 2.6573,
+      "step": 2325
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3507972665148064e-05,
+      "loss": 3.2592,
+      "step": 2326
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3496583143507976e-05,
+      "loss": 3.1492,
+      "step": 2327
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.348519362186788e-05,
+      "loss": 3.6075,
+      "step": 2328
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3473804100227792e-05,
+      "loss": 2.4614,
+      "step": 2329
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.34624145785877e-05,
+      "loss": 2.0568,
+      "step": 2330
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.345102505694761e-05,
+      "loss": 3.4257,
+      "step": 2331
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.343963553530752e-05,
+      "loss": 2.1817,
+      "step": 2332
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3428246013667425e-05,
+      "loss": 2.4995,
+      "step": 2333
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3416856492027337e-05,
+      "loss": 1.8566,
+      "step": 2334
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3405466970387245e-05,
+      "loss": 1.9899,
+      "step": 2335
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3394077448747153e-05,
+      "loss": 3.9488,
+      "step": 2336
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3382687927107065e-05,
+      "loss": 2.1181,
+      "step": 2337
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.337129840546697e-05,
+      "loss": 3.8681,
+      "step": 2338
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.335990888382688e-05,
+      "loss": 4.8442,
+      "step": 2339
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.334851936218679e-05,
+      "loss": 3.4938,
+      "step": 2340
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3337129840546697e-05,
+      "loss": 3.6613,
+      "step": 2341
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.332574031890661e-05,
+      "loss": 4.5497,
+      "step": 2342
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3314350797266514e-05,
+      "loss": 3.6492,
+      "step": 2343
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3302961275626425e-05,
+      "loss": 4.4635,
+      "step": 2344
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3291571753986334e-05,
+      "loss": 2.3769,
+      "step": 2345
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3280182232346242e-05,
+      "loss": 1.4126,
+      "step": 2346
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3268792710706153e-05,
+      "loss": 4.1159,
+      "step": 2347
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.325740318906606e-05,
+      "loss": 2.5789,
+      "step": 2348
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.324601366742597e-05,
+      "loss": 3.1395,
+      "step": 2349
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3234624145785878e-05,
+      "loss": 2.3413,
+      "step": 2350
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3223234624145786e-05,
+      "loss": 2.9916,
+      "step": 2351
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3211845102505698e-05,
+      "loss": 3.7715,
+      "step": 2352
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3200455580865603e-05,
+      "loss": 2.7037,
+      "step": 2353
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3189066059225514e-05,
+      "loss": 2.1915,
+      "step": 2354
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3177676537585423e-05,
+      "loss": 3.9811,
+      "step": 2355
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.316628701594533e-05,
+      "loss": 4.057,
+      "step": 2356
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3154897494305242e-05,
+      "loss": 4.1326,
+      "step": 2357
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3143507972665147e-05,
+      "loss": 3.2761,
+      "step": 2358
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.313211845102506e-05,
+      "loss": 3.9848,
+      "step": 2359
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3120728929384967e-05,
+      "loss": 1.7875,
+      "step": 2360
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3109339407744875e-05,
+      "loss": 3.4747,
+      "step": 2361
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3097949886104787e-05,
+      "loss": 3.0356,
+      "step": 2362
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3086560364464695e-05,
+      "loss": 4.6586,
+      "step": 2363
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3075170842824603e-05,
+      "loss": 2.6775,
+      "step": 2364
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.306378132118451e-05,
+      "loss": 1.4547,
+      "step": 2365
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.305239179954442e-05,
+      "loss": 3.2284,
+      "step": 2366
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3041002277904328e-05,
+      "loss": 3.6856,
+      "step": 2367
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.302961275626424e-05,
+      "loss": 3.0076,
+      "step": 2368
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3018223234624148e-05,
+      "loss": 4.2824,
+      "step": 2369
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3006833712984056e-05,
+      "loss": 4.3881,
+      "step": 2370
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.2995444191343964e-05,
+      "loss": 3.5258,
+      "step": 2371
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.2984054669703872e-05,
+      "loss": 2.9673,
+      "step": 2372
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2972665148063784e-05,
+      "loss": 6.0065,
+      "step": 2373
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2961275626423692e-05,
+      "loss": 3.0103,
+      "step": 2374
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.29498861047836e-05,
+      "loss": 2.1263,
+      "step": 2375
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.293849658314351e-05,
+      "loss": 2.5063,
+      "step": 2376
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2927107061503417e-05,
+      "loss": 2.4824,
+      "step": 2377
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2915717539863328e-05,
+      "loss": 3.8861,
+      "step": 2378
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2904328018223236e-05,
+      "loss": 2.0355,
+      "step": 2379
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2892938496583145e-05,
+      "loss": 3.337,
+      "step": 2380
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2881548974943053e-05,
+      "loss": 2.57,
+      "step": 2381
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.287015945330296e-05,
+      "loss": 3.0936,
+      "step": 2382
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2858769931662873e-05,
+      "loss": 1.1698,
+      "step": 2383
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.284738041002278e-05,
+      "loss": 3.9984,
+      "step": 2384
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.283599088838269e-05,
+      "loss": 3.114,
+      "step": 2385
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2824601366742597e-05,
+      "loss": 4.1452,
+      "step": 2386
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2813211845102505e-05,
+      "loss": 2.1007,
+      "step": 2387
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2801822323462417e-05,
+      "loss": 3.1682,
+      "step": 2388
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2790432801822325e-05,
+      "loss": 3.8729,
+      "step": 2389
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2779043280182233e-05,
+      "loss": 2.8728,
+      "step": 2390
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.276765375854214e-05,
+      "loss": 3.9663,
+      "step": 2391
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.275626423690205e-05,
+      "loss": 4.5187,
+      "step": 2392
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.274487471526196e-05,
+      "loss": 1.9517,
+      "step": 2393
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.273348519362187e-05,
+      "loss": 3.0697,
+      "step": 2394
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2722095671981778e-05,
+      "loss": 4.0802,
+      "step": 2395
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2710706150341686e-05,
+      "loss": 1.1643,
+      "step": 2396
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2699316628701594e-05,
+      "loss": 2.7745,
+      "step": 2397
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2687927107061506e-05,
+      "loss": 4.8988,
+      "step": 2398
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2676537585421414e-05,
+      "loss": 4.9231,
+      "step": 2399
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2665148063781322e-05,
+      "loss": 2.9114,
+      "step": 2400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.265375854214123e-05,
+      "loss": 2.0829,
+      "step": 2401
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.264236902050114e-05,
+      "loss": 2.6893,
+      "step": 2402
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.263097949886105e-05,
+      "loss": 4.8604,
+      "step": 2403
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.261958997722096e-05,
+      "loss": 0.8125,
+      "step": 2404
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2608200455580867e-05,
+      "loss": 1.3994,
+      "step": 2405
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2596810933940775e-05,
+      "loss": 2.6984,
+      "step": 2406
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2585421412300683e-05,
+      "loss": 2.9522,
+      "step": 2407
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2574031890660595e-05,
+      "loss": 4.3238,
+      "step": 2408
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2562642369020503e-05,
+      "loss": 2.8133,
+      "step": 2409
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.255125284738041e-05,
+      "loss": 4.459,
+      "step": 2410
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.253986332574032e-05,
+      "loss": 2.1825,
+      "step": 2411
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2528473804100228e-05,
+      "loss": 3.8187,
+      "step": 2412
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.251708428246014e-05,
+      "loss": 2.7158,
+      "step": 2413
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2505694760820047e-05,
+      "loss": 4.4157,
+      "step": 2414
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2494305239179956e-05,
+      "loss": 2.9716,
+      "step": 2415
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2482915717539864e-05,
+      "loss": 3.3829,
+      "step": 2416
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2471526195899772e-05,
+      "loss": 3.4301,
+      "step": 2417
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2460136674259684e-05,
+      "loss": 2.2302,
+      "step": 2418
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.244874715261959e-05,
+      "loss": 2.3991,
+      "step": 2419
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.24373576309795e-05,
+      "loss": 3.3679,
+      "step": 2420
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2425968109339408e-05,
+      "loss": 4.1409,
+      "step": 2421
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2414578587699316e-05,
+      "loss": 3.3349,
+      "step": 2422
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2403189066059228e-05,
+      "loss": 2.5701,
+      "step": 2423
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2391799544419133e-05,
+      "loss": 2.7814,
+      "step": 2424
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2380410022779044e-05,
+      "loss": 2.224,
+      "step": 2425
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2369020501138953e-05,
+      "loss": 0.8401,
+      "step": 2426
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.235763097949886e-05,
+      "loss": 3.4474,
+      "step": 2427
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2346241457858772e-05,
+      "loss": 3.3189,
+      "step": 2428
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2334851936218677e-05,
+      "loss": 2.6854,
+      "step": 2429
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.232346241457859e-05,
+      "loss": 3.969,
+      "step": 2430
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.23120728929385e-05,
+      "loss": 3.3373,
+      "step": 2431
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2300683371298405e-05,
+      "loss": 3.6304,
+      "step": 2432
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2289293849658317e-05,
+      "loss": 2.5663,
+      "step": 2433
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.227790432801822e-05,
+      "loss": 2.332,
+      "step": 2434
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2266514806378133e-05,
+      "loss": 3.4025,
+      "step": 2435
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2255125284738045e-05,
+      "loss": 1.8524,
+      "step": 2436
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.224373576309795e-05,
+      "loss": 2.7032,
+      "step": 2437
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.223234624145786e-05,
+      "loss": 3.2103,
+      "step": 2438
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2220956719817766e-05,
+      "loss": 3.5293,
+      "step": 2439
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2209567198177678e-05,
+      "loss": 3.1616,
+      "step": 2440
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.219817767653759e-05,
+      "loss": 4.5313,
+      "step": 2441
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2186788154897494e-05,
+      "loss": 3.0628,
+      "step": 2442
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2175398633257406e-05,
+      "loss": 2.4035,
+      "step": 2443
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.216400911161731e-05,
+      "loss": 4.0057,
+      "step": 2444
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2152619589977222e-05,
+      "loss": 2.0299,
+      "step": 2445
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2141230068337134e-05,
+      "loss": 3.089,
+      "step": 2446
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.212984054669704e-05,
+      "loss": 4.6802,
+      "step": 2447
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.211845102505695e-05,
+      "loss": 4.0112,
+      "step": 2448
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2107061503416858e-05,
+      "loss": 2.7931,
+      "step": 2449
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2095671981776766e-05,
+      "loss": 2.442,
+      "step": 2450
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2084282460136678e-05,
+      "loss": 2.8463,
+      "step": 2451
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2072892938496583e-05,
+      "loss": 1.9908,
+      "step": 2452
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2061503416856494e-05,
+      "loss": 3.1662,
+      "step": 2453
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2050113895216403e-05,
+      "loss": 3.2106,
+      "step": 2454
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.203872437357631e-05,
+      "loss": 2.6007,
+      "step": 2455
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2027334851936222e-05,
+      "loss": 2.9658,
+      "step": 2456
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2015945330296127e-05,
+      "loss": 1.7737,
+      "step": 2457
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.200455580865604e-05,
+      "loss": 3.9862,
+      "step": 2458
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.1993166287015947e-05,
+      "loss": 3.3424,
+      "step": 2459
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.1981776765375855e-05,
+      "loss": 1.9472,
+      "step": 2460
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1970387243735767e-05,
+      "loss": 2.827,
+      "step": 2461
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.195899772209567e-05,
+      "loss": 3.8475,
+      "step": 2462
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1947608200455583e-05,
+      "loss": 3.7048,
+      "step": 2463
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.193621867881549e-05,
+      "loss": 2.8916,
+      "step": 2464
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.19248291571754e-05,
+      "loss": 4.1998,
+      "step": 2465
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1913439635535308e-05,
+      "loss": 3.2201,
+      "step": 2466
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1902050113895216e-05,
+      "loss": 2.8161,
+      "step": 2467
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1890660592255128e-05,
+      "loss": 2.1489,
+      "step": 2468
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1879271070615036e-05,
+      "loss": 2.7514,
+      "step": 2469
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1867881548974944e-05,
+      "loss": 4.9154,
+      "step": 2470
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1856492027334852e-05,
+      "loss": 1.9516,
+      "step": 2471
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.184510250569476e-05,
+      "loss": 3.1875,
+      "step": 2472
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1833712984054672e-05,
+      "loss": 3.2232,
+      "step": 2473
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.182232346241458e-05,
+      "loss": 4.7718,
+      "step": 2474
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.181093394077449e-05,
+      "loss": 2.6181,
+      "step": 2475
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1799544419134397e-05,
+      "loss": 1.5862,
+      "step": 2476
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1788154897494305e-05,
+      "loss": 5.1422,
+      "step": 2477
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1776765375854217e-05,
+      "loss": 2.9118,
+      "step": 2478
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1765375854214125e-05,
+      "loss": 2.9328,
+      "step": 2479
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1753986332574033e-05,
+      "loss": 3.0448,
+      "step": 2480
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.174259681093394e-05,
+      "loss": 4.1542,
+      "step": 2481
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.173120728929385e-05,
+      "loss": 2.5515,
+      "step": 2482
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.171981776765376e-05,
+      "loss": 2.8562,
+      "step": 2483
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.170842824601367e-05,
+      "loss": 3.1034,
+      "step": 2484
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1697038724373577e-05,
+      "loss": 3.2627,
+      "step": 2485
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1685649202733486e-05,
+      "loss": 4.6369,
+      "step": 2486
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1674259681093394e-05,
+      "loss": 2.1993,
+      "step": 2487
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1662870159453305e-05,
+      "loss": 1.3793,
+      "step": 2488
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1651480637813214e-05,
+      "loss": 2.371,
+      "step": 2489
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1640091116173122e-05,
+      "loss": 3.8276,
+      "step": 2490
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.162870159453303e-05,
+      "loss": 3.3344,
+      "step": 2491
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1617312072892938e-05,
+      "loss": 3.812,
+      "step": 2492
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.160592255125285e-05,
+      "loss": 3.0952,
+      "step": 2493
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1594533029612758e-05,
+      "loss": 3.9453,
+      "step": 2494
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1583143507972666e-05,
+      "loss": 2.8846,
+      "step": 2495
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1571753986332574e-05,
+      "loss": 4.3939,
+      "step": 2496
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1560364464692483e-05,
+      "loss": 3.4697,
+      "step": 2497
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1548974943052394e-05,
+      "loss": 2.9032,
+      "step": 2498
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1537585421412302e-05,
+      "loss": 3.8426,
+      "step": 2499
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.152619589977221e-05,
+      "loss": 3.2431,
+      "step": 2500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.151480637813212e-05,
+      "loss": 2.2852,
+      "step": 2501
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1503416856492027e-05,
+      "loss": 3.1631,
+      "step": 2502
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.149202733485194e-05,
+      "loss": 2.5291,
+      "step": 2503
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1480637813211847e-05,
+      "loss": 3.2818,
+      "step": 2504
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1469248291571755e-05,
+      "loss": 3.8408,
+      "step": 2505
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1457858769931663e-05,
+      "loss": 2.7921,
+      "step": 2506
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.144646924829157e-05,
+      "loss": 3.4085,
+      "step": 2507
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1435079726651483e-05,
+      "loss": 2.9056,
+      "step": 2508
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.142369020501139e-05,
+      "loss": 2.6152,
+      "step": 2509
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.14123006833713e-05,
+      "loss": 4.27,
+      "step": 2510
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1400911161731208e-05,
+      "loss": 1.8438,
+      "step": 2511
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1389521640091116e-05,
+      "loss": 2.7101,
+      "step": 2512
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1378132118451024e-05,
+      "loss": 5.1405,
+      "step": 2513
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1366742596810936e-05,
+      "loss": 4.3067,
+      "step": 2514
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1355353075170844e-05,
+      "loss": 3.0863,
+      "step": 2515
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1343963553530752e-05,
+      "loss": 3.531,
+      "step": 2516
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.133257403189066e-05,
+      "loss": 3.3899,
+      "step": 2517
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.132118451025057e-05,
+      "loss": 2.1417,
+      "step": 2518
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.130979498861048e-05,
+      "loss": 3.9174,
+      "step": 2519
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1298405466970388e-05,
+      "loss": 4.2828,
+      "step": 2520
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1287015945330296e-05,
+      "loss": 4.1478,
+      "step": 2521
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1275626423690208e-05,
+      "loss": 3.3569,
+      "step": 2522
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1264236902050113e-05,
+      "loss": 4.4702,
+      "step": 2523
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1252847380410024e-05,
+      "loss": 4.1933,
+      "step": 2524
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1241457858769933e-05,
+      "loss": 2.0121,
+      "step": 2525
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.123006833712984e-05,
+      "loss": 3.6323,
+      "step": 2526
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1218678815489752e-05,
+      "loss": 2.1846,
+      "step": 2527
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1207289293849657e-05,
+      "loss": 3.7879,
+      "step": 2528
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.119589977220957e-05,
+      "loss": 2.5056,
+      "step": 2529
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1184510250569477e-05,
+      "loss": 3.5966,
+      "step": 2530
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1173120728929385e-05,
+      "loss": 3.4689,
+      "step": 2531
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1161731207289297e-05,
+      "loss": 3.0148,
+      "step": 2532
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1150341685649202e-05,
+      "loss": 2.1763,
+      "step": 2533
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1138952164009113e-05,
+      "loss": 2.7187,
+      "step": 2534
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.112756264236902e-05,
+      "loss": 2.7274,
+      "step": 2535
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.111617312072893e-05,
+      "loss": 4.1862,
+      "step": 2536
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.110478359908884e-05,
+      "loss": 3.3388,
+      "step": 2537
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1093394077448746e-05,
+      "loss": 3.7335,
+      "step": 2538
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1082004555808658e-05,
+      "loss": 1.7138,
+      "step": 2539
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1070615034168566e-05,
+      "loss": 5.9442,
+      "step": 2540
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1059225512528474e-05,
+      "loss": 3.7588,
+      "step": 2541
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1047835990888386e-05,
+      "loss": 2.2677,
+      "step": 2542
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.103644646924829e-05,
+      "loss": 3.2486,
+      "step": 2543
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1025056947608202e-05,
+      "loss": 1.5948,
+      "step": 2544
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.101366742596811e-05,
+      "loss": 3.1465,
+      "step": 2545
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.100227790432802e-05,
+      "loss": 2.8512,
+      "step": 2546
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.099088838268793e-05,
+      "loss": 3.3035,
+      "step": 2547
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.0979498861047835e-05,
+      "loss": 2.8289,
+      "step": 2548
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0968109339407747e-05,
+      "loss": 1.6807,
+      "step": 2549
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0956719817767655e-05,
+      "loss": 3.3507,
+      "step": 2550
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0945330296127563e-05,
+      "loss": 4.5496,
+      "step": 2551
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0933940774487475e-05,
+      "loss": 2.9551,
+      "step": 2552
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.092255125284738e-05,
+      "loss": 5.8199,
+      "step": 2553
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.091116173120729e-05,
+      "loss": 3.0683,
+      "step": 2554
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.08997722095672e-05,
+      "loss": 3.2608,
+      "step": 2555
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0888382687927107e-05,
+      "loss": 3.64,
+      "step": 2556
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.087699316628702e-05,
+      "loss": 3.3923,
+      "step": 2557
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0865603644646924e-05,
+      "loss": 3.2138,
+      "step": 2558
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0854214123006835e-05,
+      "loss": 4.8137,
+      "step": 2559
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0842824601366744e-05,
+      "loss": 2.9257,
+      "step": 2560
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0831435079726652e-05,
+      "loss": 4.0794,
+      "step": 2561
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0820045558086563e-05,
+      "loss": 3.271,
+      "step": 2562
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0808656036446468e-05,
+      "loss": 3.2086,
+      "step": 2563
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.079726651480638e-05,
+      "loss": 3.075,
+      "step": 2564
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0785876993166288e-05,
+      "loss": 3.3948,
+      "step": 2565
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0774487471526196e-05,
+      "loss": 2.6054,
+      "step": 2566
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0763097949886108e-05,
+      "loss": 2.9011,
+      "step": 2567
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0751708428246013e-05,
+      "loss": 4.0671,
+      "step": 2568
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0740318906605924e-05,
+      "loss": 2.595,
+      "step": 2569
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0728929384965832e-05,
+      "loss": 2.9836,
+      "step": 2570
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.071753986332574e-05,
+      "loss": 2.0806,
+      "step": 2571
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0706150341685652e-05,
+      "loss": 2.6453,
+      "step": 2572
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0694760820045557e-05,
+      "loss": 3.7553,
+      "step": 2573
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.068337129840547e-05,
+      "loss": 4.1354,
+      "step": 2574
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0671981776765377e-05,
+      "loss": 2.251,
+      "step": 2575
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0660592255125285e-05,
+      "loss": 4.6047,
+      "step": 2576
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0649202733485197e-05,
+      "loss": 4.7607,
+      "step": 2577
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.06378132118451e-05,
+      "loss": 2.1512,
+      "step": 2578
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0626423690205013e-05,
+      "loss": 2.4676,
+      "step": 2579
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.061503416856492e-05,
+      "loss": 2.1474,
+      "step": 2580
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.060364464692483e-05,
+      "loss": 2.5345,
+      "step": 2581
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.059225512528474e-05,
+      "loss": 2.7549,
+      "step": 2582
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0580865603644646e-05,
+      "loss": 2.1759,
+      "step": 2583
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0569476082004557e-05,
+      "loss": 3.5658,
+      "step": 2584
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0558086560364466e-05,
+      "loss": 4.6259,
+      "step": 2585
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0546697038724374e-05,
+      "loss": 2.961,
+      "step": 2586
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0535307517084285e-05,
+      "loss": 2.435,
+      "step": 2587
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.052391799544419e-05,
+      "loss": 2.6427,
+      "step": 2588
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0512528473804102e-05,
+      "loss": 3.6852,
+      "step": 2589
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.050113895216401e-05,
+      "loss": 2.3736,
+      "step": 2590
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0489749430523918e-05,
+      "loss": 3.2239,
+      "step": 2591
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.047835990888383e-05,
+      "loss": 4.2117,
+      "step": 2592
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0466970387243735e-05,
+      "loss": 2.3726,
+      "step": 2593
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0455580865603646e-05,
+      "loss": 5.1191,
+      "step": 2594
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0444191343963555e-05,
+      "loss": 2.9999,
+      "step": 2595
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0432801822323463e-05,
+      "loss": 3.5256,
+      "step": 2596
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0421412300683374e-05,
+      "loss": 2.6551,
+      "step": 2597
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.041002277904328e-05,
+      "loss": 2.919,
+      "step": 2598
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.039863325740319e-05,
+      "loss": 3.324,
+      "step": 2599
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.03872437357631e-05,
+      "loss": 3.1531,
+      "step": 2600
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0375854214123007e-05,
+      "loss": 2.9291,
+      "step": 2601
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.036446469248292e-05,
+      "loss": 3.8079,
+      "step": 2602
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0353075170842824e-05,
+      "loss": 3.1192,
+      "step": 2603
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0341685649202735e-05,
+      "loss": 3.8863,
+      "step": 2604
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0330296127562643e-05,
+      "loss": 2.3789,
+      "step": 2605
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.031890660592255e-05,
+      "loss": 1.9647,
+      "step": 2606
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0307517084282463e-05,
+      "loss": 3.5929,
+      "step": 2607
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.029612756264237e-05,
+      "loss": 1.7114,
+      "step": 2608
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.028473804100228e-05,
+      "loss": 2.2406,
+      "step": 2609
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0273348519362188e-05,
+      "loss": 3.0559,
+      "step": 2610
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0261958997722096e-05,
+      "loss": 3.8273,
+      "step": 2611
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0250569476082004e-05,
+      "loss": 2.6672,
+      "step": 2612
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0239179954441916e-05,
+      "loss": 2.855,
+      "step": 2613
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0227790432801824e-05,
+      "loss": 3.1623,
+      "step": 2614
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0216400911161732e-05,
+      "loss": 3.5333,
+      "step": 2615
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.020501138952164e-05,
+      "loss": 4.452,
+      "step": 2616
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.019362186788155e-05,
+      "loss": 2.3264,
+      "step": 2617
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.018223234624146e-05,
+      "loss": 4.4106,
+      "step": 2618
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.017084282460137e-05,
+      "loss": 1.5245,
+      "step": 2619
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0159453302961277e-05,
+      "loss": 4.5281,
+      "step": 2620
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0148063781321185e-05,
+      "loss": 3.3411,
+      "step": 2621
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0136674259681093e-05,
+      "loss": 3.311,
+      "step": 2622
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0125284738041005e-05,
+      "loss": 3.2443,
+      "step": 2623
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0113895216400913e-05,
+      "loss": 3.0692,
+      "step": 2624
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.010250569476082e-05,
+      "loss": 2.0773,
+      "step": 2625
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.009111617312073e-05,
+      "loss": 3.5057,
+      "step": 2626
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0079726651480637e-05,
+      "loss": 2.4607,
+      "step": 2627
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.006833712984055e-05,
+      "loss": 2.9482,
+      "step": 2628
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0056947608200457e-05,
+      "loss": 3.3792,
+      "step": 2629
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0045558086560365e-05,
+      "loss": 2.2524,
+      "step": 2630
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0034168564920274e-05,
+      "loss": 3.2672,
+      "step": 2631
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0022779043280182e-05,
+      "loss": 2.7069,
+      "step": 2632
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0011389521640093e-05,
+      "loss": 3.5563,
+      "step": 2633
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2e-05,
+      "loss": 1.5373,
+      "step": 2634
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.998861047835991e-05,
+      "loss": 1.2627,
+      "step": 2635
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.9977220956719818e-05,
+      "loss": 3.0104,
+      "step": 2636
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9965831435079726e-05,
+      "loss": 2.6388,
+      "step": 2637
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9954441913439638e-05,
+      "loss": 4.4016,
+      "step": 2638
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9943052391799546e-05,
+      "loss": 3.4119,
+      "step": 2639
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9931662870159454e-05,
+      "loss": 3.7087,
+      "step": 2640
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9920273348519362e-05,
+      "loss": 2.7671,
+      "step": 2641
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.990888382687927e-05,
+      "loss": 1.2052,
+      "step": 2642
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9897494305239182e-05,
+      "loss": 2.8077,
+      "step": 2643
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.988610478359909e-05,
+      "loss": 3.5269,
+      "step": 2644
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9874715261959e-05,
+      "loss": 3.6279,
+      "step": 2645
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9863325740318907e-05,
+      "loss": 2.6613,
+      "step": 2646
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9851936218678815e-05,
+      "loss": 4.2297,
+      "step": 2647
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9840546697038727e-05,
+      "loss": 3.2518,
+      "step": 2648
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9829157175398635e-05,
+      "loss": 2.4811,
+      "step": 2649
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9817767653758543e-05,
+      "loss": 3.6055,
+      "step": 2650
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.980637813211845e-05,
+      "loss": 3.6632,
+      "step": 2651
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.979498861047836e-05,
+      "loss": 2.7249,
+      "step": 2652
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.978359908883827e-05,
+      "loss": 2.246,
+      "step": 2653
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.977220956719818e-05,
+      "loss": 1.6139,
+      "step": 2654
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9760820045558088e-05,
+      "loss": 3.1328,
+      "step": 2655
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9749430523917996e-05,
+      "loss": 3.0356,
+      "step": 2656
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9738041002277904e-05,
+      "loss": 3.398,
+      "step": 2657
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9726651480637816e-05,
+      "loss": 3.3723,
+      "step": 2658
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.971526195899772e-05,
+      "loss": 2.7407,
+      "step": 2659
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9703872437357632e-05,
+      "loss": 4.1391,
+      "step": 2660
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.969248291571754e-05,
+      "loss": 2.8094,
+      "step": 2661
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.968109339407745e-05,
+      "loss": 2.7854,
+      "step": 2662
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.966970387243736e-05,
+      "loss": 2.5488,
+      "step": 2663
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9658314350797265e-05,
+      "loss": 2.219,
+      "step": 2664
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9646924829157176e-05,
+      "loss": 2.7514,
+      "step": 2665
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9635535307517085e-05,
+      "loss": 3.1255,
+      "step": 2666
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9624145785876993e-05,
+      "loss": 2.4488,
+      "step": 2667
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9612756264236904e-05,
+      "loss": 2.0248,
+      "step": 2668
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.960136674259681e-05,
+      "loss": 4.3058,
+      "step": 2669
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.958997722095672e-05,
+      "loss": 3.3326,
+      "step": 2670
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.957858769931663e-05,
+      "loss": 1.1047,
+      "step": 2671
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9567198177676537e-05,
+      "loss": 3.0751,
+      "step": 2672
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.955580865603645e-05,
+      "loss": 2.5836,
+      "step": 2673
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9544419134396354e-05,
+      "loss": 2.4178,
+      "step": 2674
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9533029612756265e-05,
+      "loss": 4.8139,
+      "step": 2675
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9521640091116177e-05,
+      "loss": 1.7226,
+      "step": 2676
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.951025056947608e-05,
+      "loss": 3.169,
+      "step": 2677
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9498861047835993e-05,
+      "loss": 1.9337,
+      "step": 2678
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9487471526195898e-05,
+      "loss": 3.3713,
+      "step": 2679
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.947608200455581e-05,
+      "loss": 2.2479,
+      "step": 2680
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.946469248291572e-05,
+      "loss": 3.1156,
+      "step": 2681
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9453302961275626e-05,
+      "loss": 3.5614,
+      "step": 2682
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9441913439635538e-05,
+      "loss": 2.4267,
+      "step": 2683
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9430523917995442e-05,
+      "loss": 3.9293,
+      "step": 2684
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9419134396355354e-05,
+      "loss": 2.8924,
+      "step": 2685
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9407744874715266e-05,
+      "loss": 2.542,
+      "step": 2686
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.939635535307517e-05,
+      "loss": 3.1274,
+      "step": 2687
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9384965831435082e-05,
+      "loss": 2.4803,
+      "step": 2688
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9373576309794987e-05,
+      "loss": 1.5148,
+      "step": 2689
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.93621867881549e-05,
+      "loss": 2.4811,
+      "step": 2690
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.935079726651481e-05,
+      "loss": 5.4898,
+      "step": 2691
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9339407744874715e-05,
+      "loss": 4.9867,
+      "step": 2692
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9328018223234626e-05,
+      "loss": 2.9684,
+      "step": 2693
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9316628701594535e-05,
+      "loss": 2.4075,
+      "step": 2694
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9305239179954443e-05,
+      "loss": 3.6174,
+      "step": 2695
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9293849658314354e-05,
+      "loss": 3.5645,
+      "step": 2696
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.928246013667426e-05,
+      "loss": 2.8961,
+      "step": 2697
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.927107061503417e-05,
+      "loss": 3.2435,
+      "step": 2698
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.925968109339408e-05,
+      "loss": 2.6333,
+      "step": 2699
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9248291571753987e-05,
+      "loss": 2.7794,
+      "step": 2700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.92369020501139e-05,
+      "loss": 2.5434,
+      "step": 2701
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9225512528473804e-05,
+      "loss": 1.5801,
+      "step": 2702
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9214123006833715e-05,
+      "loss": 3.6585,
+      "step": 2703
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9202733485193623e-05,
+      "loss": 3.0877,
+      "step": 2704
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.919134396355353e-05,
+      "loss": 3.2656,
+      "step": 2705
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9179954441913443e-05,
+      "loss": 2.771,
+      "step": 2706
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9168564920273348e-05,
+      "loss": 2.5274,
+      "step": 2707
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.915717539863326e-05,
+      "loss": 2.2426,
+      "step": 2708
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9145785876993168e-05,
+      "loss": 3.6486,
+      "step": 2709
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9134396355353076e-05,
+      "loss": 3.7022,
+      "step": 2710
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9123006833712984e-05,
+      "loss": 2.059,
+      "step": 2711
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9111617312072892e-05,
+      "loss": 3.9894,
+      "step": 2712
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9100227790432804e-05,
+      "loss": 2.9035,
+      "step": 2713
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9088838268792712e-05,
+      "loss": 3.4916,
+      "step": 2714
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.907744874715262e-05,
+      "loss": 4.1253,
+      "step": 2715
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.906605922551253e-05,
+      "loss": 2.9265,
+      "step": 2716
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9054669703872437e-05,
+      "loss": 3.6099,
+      "step": 2717
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.904328018223235e-05,
+      "loss": 1.6325,
+      "step": 2718
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9031890660592257e-05,
+      "loss": 2.5924,
+      "step": 2719
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9020501138952165e-05,
+      "loss": 3.1222,
+      "step": 2720
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9009111617312073e-05,
+      "loss": 2.3241,
+      "step": 2721
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.899772209567198e-05,
+      "loss": 3.7,
+      "step": 2722
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.8986332574031893e-05,
+      "loss": 4.4353,
+      "step": 2723
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.89749430523918e-05,
+      "loss": 3.12,
+      "step": 2724
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.896355353075171e-05,
+      "loss": 3.5821,
+      "step": 2725
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8952164009111618e-05,
+      "loss": 2.1877,
+      "step": 2726
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8940774487471526e-05,
+      "loss": 3.049,
+      "step": 2727
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8929384965831437e-05,
+      "loss": 5.4912,
+      "step": 2728
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8917995444191346e-05,
+      "loss": 2.9667,
+      "step": 2729
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8906605922551254e-05,
+      "loss": 2.9991,
+      "step": 2730
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8895216400911162e-05,
+      "loss": 3.941,
+      "step": 2731
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.888382687927107e-05,
+      "loss": 2.6516,
+      "step": 2732
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8872437357630982e-05,
+      "loss": 3.6786,
+      "step": 2733
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.886104783599089e-05,
+      "loss": 2.1042,
+      "step": 2734
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8849658314350798e-05,
+      "loss": 2.5527,
+      "step": 2735
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8838268792710706e-05,
+      "loss": 3.9883,
+      "step": 2736
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8826879271070615e-05,
+      "loss": 3.2163,
+      "step": 2737
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8815489749430526e-05,
+      "loss": 3.4775,
+      "step": 2738
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8804100227790434e-05,
+      "loss": 3.7371,
+      "step": 2739
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8792710706150343e-05,
+      "loss": 3.115,
+      "step": 2740
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.878132118451025e-05,
+      "loss": 3.5357,
+      "step": 2741
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.876993166287016e-05,
+      "loss": 1.5733,
+      "step": 2742
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.875854214123007e-05,
+      "loss": 4.4635,
+      "step": 2743
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.874715261958998e-05,
+      "loss": 2.1109,
+      "step": 2744
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8735763097949887e-05,
+      "loss": 2.0895,
+      "step": 2745
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8724373576309795e-05,
+      "loss": 2.643,
+      "step": 2746
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8712984054669703e-05,
+      "loss": 3.4105,
+      "step": 2747
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8701594533029615e-05,
+      "loss": 3.6136,
+      "step": 2748
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8690205011389523e-05,
+      "loss": 4.1554,
+      "step": 2749
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.867881548974943e-05,
+      "loss": 2.0566,
+      "step": 2750
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.866742596810934e-05,
+      "loss": 3.5028,
+      "step": 2751
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8656036446469248e-05,
+      "loss": 3.0521,
+      "step": 2752
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.864464692482916e-05,
+      "loss": 2.4142,
+      "step": 2753
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8633257403189068e-05,
+      "loss": 3.9518,
+      "step": 2754
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8621867881548976e-05,
+      "loss": 3.261,
+      "step": 2755
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8610478359908884e-05,
+      "loss": 2.4914,
+      "step": 2756
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8599088838268792e-05,
+      "loss": 3.5245,
+      "step": 2757
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.85876993166287e-05,
+      "loss": 2.6941,
+      "step": 2758
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8576309794988612e-05,
+      "loss": 2.1779,
+      "step": 2759
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.856492027334852e-05,
+      "loss": 2.6977,
+      "step": 2760
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.855353075170843e-05,
+      "loss": 3.6377,
+      "step": 2761
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.854214123006834e-05,
+      "loss": 3.7739,
+      "step": 2762
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8530751708428245e-05,
+      "loss": 3.4818,
+      "step": 2763
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8519362186788156e-05,
+      "loss": 5.2471,
+      "step": 2764
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8507972665148065e-05,
+      "loss": 3.2241,
+      "step": 2765
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8496583143507973e-05,
+      "loss": 3.3015,
+      "step": 2766
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8485193621867884e-05,
+      "loss": 3.2847,
+      "step": 2767
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.847380410022779e-05,
+      "loss": 2.774,
+      "step": 2768
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.84624145785877e-05,
+      "loss": 1.4564,
+      "step": 2769
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.845102505694761e-05,
+      "loss": 2.8364,
+      "step": 2770
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8439635535307517e-05,
+      "loss": 3.6,
+      "step": 2771
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.842824601366743e-05,
+      "loss": 4.8277,
+      "step": 2772
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8416856492027334e-05,
+      "loss": 3.696,
+      "step": 2773
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8405466970387245e-05,
+      "loss": 4.6147,
+      "step": 2774
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8394077448747154e-05,
+      "loss": 2.6129,
+      "step": 2775
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8382687927107062e-05,
+      "loss": 3.1467,
+      "step": 2776
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8371298405466973e-05,
+      "loss": 2.2465,
+      "step": 2777
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8359908883826878e-05,
+      "loss": 3.5605,
+      "step": 2778
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.834851936218679e-05,
+      "loss": 2.441,
+      "step": 2779
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8337129840546698e-05,
+      "loss": 3.3575,
+      "step": 2780
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8325740318906606e-05,
+      "loss": 3.5765,
+      "step": 2781
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8314350797266518e-05,
+      "loss": 3.5885,
+      "step": 2782
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8302961275626423e-05,
+      "loss": 4.4406,
+      "step": 2783
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8291571753986334e-05,
+      "loss": 3.4646,
+      "step": 2784
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8280182232346242e-05,
+      "loss": 3.5986,
+      "step": 2785
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.826879271070615e-05,
+      "loss": 3.7158,
+      "step": 2786
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8257403189066062e-05,
+      "loss": 4.3858,
+      "step": 2787
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8246013667425967e-05,
+      "loss": 2.2713,
+      "step": 2788
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.823462414578588e-05,
+      "loss": 3.1886,
+      "step": 2789
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8223234624145787e-05,
+      "loss": 2.7598,
+      "step": 2790
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8211845102505695e-05,
+      "loss": 1.8039,
+      "step": 2791
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8200455580865607e-05,
+      "loss": 2.8836,
+      "step": 2792
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.818906605922551e-05,
+      "loss": 3.7514,
+      "step": 2793
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8177676537585423e-05,
+      "loss": 3.6745,
+      "step": 2794
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.816628701594533e-05,
+      "loss": 3.4908,
+      "step": 2795
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.815489749430524e-05,
+      "loss": 3.3694,
+      "step": 2796
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.814350797266515e-05,
+      "loss": 5.1155,
+      "step": 2797
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8132118451025056e-05,
+      "loss": 3.2413,
+      "step": 2798
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8120728929384967e-05,
+      "loss": 3.2291,
+      "step": 2799
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8109339407744876e-05,
+      "loss": 3.8913,
+      "step": 2800
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8097949886104784e-05,
+      "loss": 1.8081,
+      "step": 2801
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8086560364464695e-05,
+      "loss": 3.6681,
+      "step": 2802
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.80751708428246e-05,
+      "loss": 2.5446,
+      "step": 2803
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8063781321184512e-05,
+      "loss": 2.2965,
+      "step": 2804
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.805239179954442e-05,
+      "loss": 2.7843,
+      "step": 2805
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8041002277904328e-05,
+      "loss": 2.4395,
+      "step": 2806
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.802961275626424e-05,
+      "loss": 3.8791,
+      "step": 2807
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8018223234624145e-05,
+      "loss": 2.7686,
+      "step": 2808
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8006833712984056e-05,
+      "loss": 2.6712,
+      "step": 2809
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.7995444191343964e-05,
+      "loss": 2.6275,
+      "step": 2810
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.7984054669703873e-05,
+      "loss": 2.3657,
+      "step": 2811
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7972665148063784e-05,
+      "loss": 2.825,
+      "step": 2812
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.796127562642369e-05,
+      "loss": 5.5859,
+      "step": 2813
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.79498861047836e-05,
+      "loss": 1.7941,
+      "step": 2814
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.793849658314351e-05,
+      "loss": 1.9665,
+      "step": 2815
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7927107061503417e-05,
+      "loss": 4.2722,
+      "step": 2816
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.791571753986333e-05,
+      "loss": 1.8234,
+      "step": 2817
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7904328018223233e-05,
+      "loss": 5.8155,
+      "step": 2818
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7892938496583145e-05,
+      "loss": 2.3948,
+      "step": 2819
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7881548974943053e-05,
+      "loss": 2.8139,
+      "step": 2820
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.787015945330296e-05,
+      "loss": 1.686,
+      "step": 2821
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7858769931662873e-05,
+      "loss": 1.9851,
+      "step": 2822
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7847380410022778e-05,
+      "loss": 3.6332,
+      "step": 2823
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.783599088838269e-05,
+      "loss": 4.21,
+      "step": 2824
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7824601366742598e-05,
+      "loss": 2.8651,
+      "step": 2825
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7813211845102506e-05,
+      "loss": 3.5857,
+      "step": 2826
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7801822323462417e-05,
+      "loss": 3.9247,
+      "step": 2827
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7790432801822322e-05,
+      "loss": 3.8756,
+      "step": 2828
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7779043280182234e-05,
+      "loss": 0.9915,
+      "step": 2829
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7767653758542142e-05,
+      "loss": 3.251,
+      "step": 2830
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.775626423690205e-05,
+      "loss": 3.2074,
+      "step": 2831
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7744874715261962e-05,
+      "loss": 1.1357,
+      "step": 2832
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7733485193621867e-05,
+      "loss": 2.9051,
+      "step": 2833
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7722095671981778e-05,
+      "loss": 3.9882,
+      "step": 2834
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7710706150341686e-05,
+      "loss": 2.8419,
+      "step": 2835
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7699316628701595e-05,
+      "loss": 3.3994,
+      "step": 2836
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7687927107061506e-05,
+      "loss": 2.8804,
+      "step": 2837
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.767653758542141e-05,
+      "loss": 2.3254,
+      "step": 2838
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7665148063781323e-05,
+      "loss": 3.028,
+      "step": 2839
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.765375854214123e-05,
+      "loss": 2.3649,
+      "step": 2840
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.764236902050114e-05,
+      "loss": 2.753,
+      "step": 2841
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.763097949886105e-05,
+      "loss": 2.9261,
+      "step": 2842
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7619589977220956e-05,
+      "loss": 1.9572,
+      "step": 2843
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7608200455580867e-05,
+      "loss": 1.7277,
+      "step": 2844
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7596810933940775e-05,
+      "loss": 2.6867,
+      "step": 2845
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7585421412300684e-05,
+      "loss": 1.9435,
+      "step": 2846
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7574031890660595e-05,
+      "loss": 3.3062,
+      "step": 2847
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.75626423690205e-05,
+      "loss": 3.967,
+      "step": 2848
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.755125284738041e-05,
+      "loss": 2.7886,
+      "step": 2849
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.753986332574032e-05,
+      "loss": 3.8765,
+      "step": 2850
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7528473804100228e-05,
+      "loss": 3.4985,
+      "step": 2851
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.751708428246014e-05,
+      "loss": 2.2314,
+      "step": 2852
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7505694760820048e-05,
+      "loss": 3.915,
+      "step": 2853
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7494305239179956e-05,
+      "loss": 1.8593,
+      "step": 2854
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7482915717539864e-05,
+      "loss": 2.2733,
+      "step": 2855
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7471526195899772e-05,
+      "loss": 1.9853,
+      "step": 2856
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.746013667425968e-05,
+      "loss": 3.3891,
+      "step": 2857
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7448747152619592e-05,
+      "loss": 4.3868,
+      "step": 2858
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.74373576309795e-05,
+      "loss": 1.597,
+      "step": 2859
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.742596810933941e-05,
+      "loss": 3.8876,
+      "step": 2860
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7414578587699317e-05,
+      "loss": 3.3316,
+      "step": 2861
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7403189066059225e-05,
+      "loss": 3.5711,
+      "step": 2862
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7391799544419137e-05,
+      "loss": 1.6323,
+      "step": 2863
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7380410022779045e-05,
+      "loss": 4.73,
+      "step": 2864
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7369020501138953e-05,
+      "loss": 1.7465,
+      "step": 2865
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.735763097949886e-05,
+      "loss": 3.1047,
+      "step": 2866
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.734624145785877e-05,
+      "loss": 2.6665,
+      "step": 2867
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.733485193621868e-05,
+      "loss": 4.2915,
+      "step": 2868
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.732346241457859e-05,
+      "loss": 2.6068,
+      "step": 2869
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7312072892938497e-05,
+      "loss": 4.2186,
+      "step": 2870
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7300683371298406e-05,
+      "loss": 3.521,
+      "step": 2871
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7289293849658314e-05,
+      "loss": 2.8496,
+      "step": 2872
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7277904328018225e-05,
+      "loss": 3.3673,
+      "step": 2873
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7266514806378134e-05,
+      "loss": 1.1988,
+      "step": 2874
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7255125284738042e-05,
+      "loss": 3.0733,
+      "step": 2875
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.724373576309795e-05,
+      "loss": 1.6067,
+      "step": 2876
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7232346241457858e-05,
+      "loss": 2.7849,
+      "step": 2877
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.722095671981777e-05,
+      "loss": 4.3309,
+      "step": 2878
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7209567198177678e-05,
+      "loss": 3.7134,
+      "step": 2879
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7198177676537586e-05,
+      "loss": 3.6955,
+      "step": 2880
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7186788154897494e-05,
+      "loss": 2.9707,
+      "step": 2881
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7175398633257403e-05,
+      "loss": 3.5,
+      "step": 2882
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7164009111617314e-05,
+      "loss": 1.5584,
+      "step": 2883
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7152619589977222e-05,
+      "loss": 3.5263,
+      "step": 2884
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.714123006833713e-05,
+      "loss": 1.9105,
+      "step": 2885
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.712984054669704e-05,
+      "loss": 3.0666,
+      "step": 2886
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7118451025056947e-05,
+      "loss": 3.5916,
+      "step": 2887
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.710706150341686e-05,
+      "loss": 2.788,
+      "step": 2888
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7095671981776767e-05,
+      "loss": 2.5085,
+      "step": 2889
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7084282460136675e-05,
+      "loss": 1.7092,
+      "step": 2890
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7072892938496583e-05,
+      "loss": 1.9868,
+      "step": 2891
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.706150341685649e-05,
+      "loss": 3.2442,
+      "step": 2892
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7050113895216403e-05,
+      "loss": 2.071,
+      "step": 2893
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.703872437357631e-05,
+      "loss": 4.6185,
+      "step": 2894
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.702733485193622e-05,
+      "loss": 2.6898,
+      "step": 2895
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7015945330296128e-05,
+      "loss": 2.9744,
+      "step": 2896
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7004555808656036e-05,
+      "loss": 4.6291,
+      "step": 2897
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6993166287015947e-05,
+      "loss": 2.7505,
+      "step": 2898
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6981776765375856e-05,
+      "loss": 1.8545,
+      "step": 2899
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6970387243735764e-05,
+      "loss": 2.3232,
+      "step": 2900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6958997722095672e-05,
+      "loss": 1.8009,
+      "step": 2901
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.694760820045558e-05,
+      "loss": 0.9131,
+      "step": 2902
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6936218678815492e-05,
+      "loss": 2.5098,
+      "step": 2903
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6924829157175397e-05,
+      "loss": 3.482,
+      "step": 2904
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.691343963553531e-05,
+      "loss": 5.774,
+      "step": 2905
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6902050113895217e-05,
+      "loss": 3.8258,
+      "step": 2906
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6890660592255125e-05,
+      "loss": 2.5691,
+      "step": 2907
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6879271070615036e-05,
+      "loss": 3.2375,
+      "step": 2908
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.686788154897494e-05,
+      "loss": 2.3532,
+      "step": 2909
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6856492027334853e-05,
+      "loss": 1.9691,
+      "step": 2910
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.684510250569476e-05,
+      "loss": 4.7891,
+      "step": 2911
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.683371298405467e-05,
+      "loss": 2.5466,
+      "step": 2912
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.682232346241458e-05,
+      "loss": 4.3316,
+      "step": 2913
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6810933940774486e-05,
+      "loss": 3.7403,
+      "step": 2914
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6799544419134397e-05,
+      "loss": 2.6319,
+      "step": 2915
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6788154897494305e-05,
+      "loss": 4.011,
+      "step": 2916
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6776765375854214e-05,
+      "loss": 2.6926,
+      "step": 2917
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6765375854214125e-05,
+      "loss": 4.1331,
+      "step": 2918
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.675398633257403e-05,
+      "loss": 1.1712,
+      "step": 2919
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.674259681093394e-05,
+      "loss": 3.2211,
+      "step": 2920
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6731207289293853e-05,
+      "loss": 3.7859,
+      "step": 2921
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6719817767653758e-05,
+      "loss": 2.7841,
+      "step": 2922
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.670842824601367e-05,
+      "loss": 3.7156,
+      "step": 2923
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6697038724373574e-05,
+      "loss": 1.5166,
+      "step": 2924
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6685649202733486e-05,
+      "loss": 2.0997,
+      "step": 2925
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6674259681093398e-05,
+      "loss": 3.8508,
+      "step": 2926
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6662870159453302e-05,
+      "loss": 1.9413,
+      "step": 2927
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6651480637813214e-05,
+      "loss": 3.7242,
+      "step": 2928
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.664009111617312e-05,
+      "loss": 2.6111,
+      "step": 2929
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.662870159453303e-05,
+      "loss": 2.7985,
+      "step": 2930
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6617312072892942e-05,
+      "loss": 2.8748,
+      "step": 2931
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6605922551252847e-05,
+      "loss": 3.6555,
+      "step": 2932
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.659453302961276e-05,
+      "loss": 2.6469,
+      "step": 2933
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6583143507972663e-05,
+      "loss": 2.2941,
+      "step": 2934
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6571753986332575e-05,
+      "loss": 1.8816,
+      "step": 2935
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6560364464692486e-05,
+      "loss": 4.9252,
+      "step": 2936
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.654897494305239e-05,
+      "loss": 2.8097,
+      "step": 2937
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6537585421412303e-05,
+      "loss": 2.7807,
+      "step": 2938
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.652619589977221e-05,
+      "loss": 3.329,
+      "step": 2939
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.651480637813212e-05,
+      "loss": 3.5781,
+      "step": 2940
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.650341685649203e-05,
+      "loss": 1.5408,
+      "step": 2941
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6492027334851936e-05,
+      "loss": 2.34,
+      "step": 2942
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6480637813211847e-05,
+      "loss": 2.6166,
+      "step": 2943
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6469248291571755e-05,
+      "loss": 2.4088,
+      "step": 2944
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6457858769931664e-05,
+      "loss": 2.6674,
+      "step": 2945
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6446469248291575e-05,
+      "loss": 2.8888,
+      "step": 2946
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.643507972665148e-05,
+      "loss": 2.3762,
+      "step": 2947
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.642369020501139e-05,
+      "loss": 2.519,
+      "step": 2948
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.64123006833713e-05,
+      "loss": 0.8197,
+      "step": 2949
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6400911161731208e-05,
+      "loss": 3.2271,
+      "step": 2950
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.638952164009112e-05,
+      "loss": 3.0847,
+      "step": 2951
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6378132118451024e-05,
+      "loss": 3.1636,
+      "step": 2952
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6366742596810936e-05,
+      "loss": 2.9117,
+      "step": 2953
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6355353075170844e-05,
+      "loss": 2.8497,
+      "step": 2954
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6343963553530752e-05,
+      "loss": 4.4561,
+      "step": 2955
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.633257403189066e-05,
+      "loss": 3.8365,
+      "step": 2956
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.632118451025057e-05,
+      "loss": 1.7722,
+      "step": 2957
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.630979498861048e-05,
+      "loss": 4.9933,
+      "step": 2958
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.629840546697039e-05,
+      "loss": 2.492,
+      "step": 2959
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6287015945330297e-05,
+      "loss": 4.1598,
+      "step": 2960
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6275626423690205e-05,
+      "loss": 3.0657,
+      "step": 2961
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6264236902050113e-05,
+      "loss": 4.2921,
+      "step": 2962
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6252847380410025e-05,
+      "loss": 2.9145,
+      "step": 2963
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6241457858769933e-05,
+      "loss": 2.1735,
+      "step": 2964
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.623006833712984e-05,
+      "loss": 1.5861,
+      "step": 2965
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.621867881548975e-05,
+      "loss": 2.7201,
+      "step": 2966
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6207289293849658e-05,
+      "loss": 3.4681,
+      "step": 2967
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.619589977220957e-05,
+      "loss": 3.2317,
+      "step": 2968
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6184510250569478e-05,
+      "loss": 4.5744,
+      "step": 2969
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6173120728929386e-05,
+      "loss": 3.7021,
+      "step": 2970
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6161731207289294e-05,
+      "loss": 3.4064,
+      "step": 2971
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6150341685649202e-05,
+      "loss": 2.2577,
+      "step": 2972
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6138952164009114e-05,
+      "loss": 3.4386,
+      "step": 2973
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6127562642369022e-05,
+      "loss": 3.0858,
+      "step": 2974
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.611617312072893e-05,
+      "loss": 3.4053,
+      "step": 2975
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.610478359908884e-05,
+      "loss": 3.257,
+      "step": 2976
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6093394077448747e-05,
+      "loss": 2.6965,
+      "step": 2977
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6082004555808658e-05,
+      "loss": 2.4942,
+      "step": 2978
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6070615034168566e-05,
+      "loss": 3.9956,
+      "step": 2979
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6059225512528475e-05,
+      "loss": 1.0542,
+      "step": 2980
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6047835990888383e-05,
+      "loss": 3.9287,
+      "step": 2981
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.603644646924829e-05,
+      "loss": 3.0929,
+      "step": 2982
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6025056947608203e-05,
+      "loss": 3.7007,
+      "step": 2983
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.601366742596811e-05,
+      "loss": 0.9882,
+      "step": 2984
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.600227790432802e-05,
+      "loss": 1.7227,
+      "step": 2985
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5990888382687927e-05,
+      "loss": 2.4185,
+      "step": 2986
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5979498861047835e-05,
+      "loss": 1.7864,
+      "step": 2987
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5968109339407747e-05,
+      "loss": 4.0723,
+      "step": 2988
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5956719817767655e-05,
+      "loss": 2.9758,
+      "step": 2989
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5945330296127563e-05,
+      "loss": 3.8256,
+      "step": 2990
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.593394077448747e-05,
+      "loss": 3.8272,
+      "step": 2991
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.592255125284738e-05,
+      "loss": 2.7089,
+      "step": 2992
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.591116173120729e-05,
+      "loss": 2.3904,
+      "step": 2993
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.58997722095672e-05,
+      "loss": 4.0039,
+      "step": 2994
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5888382687927108e-05,
+      "loss": 3.4226,
+      "step": 2995
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5876993166287016e-05,
+      "loss": 3.6311,
+      "step": 2996
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5865603644646924e-05,
+      "loss": 2.3112,
+      "step": 2997
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5854214123006836e-05,
+      "loss": 1.7164,
+      "step": 2998
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5842824601366744e-05,
+      "loss": 3.2857,
+      "step": 2999
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5831435079726652e-05,
+      "loss": 1.9448,
+      "step": 3000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.582004555808656e-05,
+      "loss": 3.1861,
+      "step": 3001
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.580865603644647e-05,
+      "loss": 3.2853,
+      "step": 3002
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5797266514806377e-05,
+      "loss": 3.916,
+      "step": 3003
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.578587699316629e-05,
+      "loss": 2.8869,
+      "step": 3004
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5774487471526197e-05,
+      "loss": 2.7584,
+      "step": 3005
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5763097949886105e-05,
+      "loss": 2.6072,
+      "step": 3006
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5751708428246016e-05,
+      "loss": 3.588,
+      "step": 3007
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.574031890660592e-05,
+      "loss": 3.6738,
+      "step": 3008
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5728929384965833e-05,
+      "loss": 3.2056,
+      "step": 3009
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.571753986332574e-05,
+      "loss": 2.9084,
+      "step": 3010
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.570615034168565e-05,
+      "loss": 2.2959,
+      "step": 3011
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.569476082004556e-05,
+      "loss": 2.2819,
+      "step": 3012
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5683371298405466e-05,
+      "loss": 4.118,
+      "step": 3013
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5671981776765377e-05,
+      "loss": 3.5363,
+      "step": 3014
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5660592255125285e-05,
+      "loss": 2.7712,
+      "step": 3015
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5649202733485194e-05,
+      "loss": 4.233,
+      "step": 3016
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5637813211845105e-05,
+      "loss": 3.3727,
+      "step": 3017
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.562642369020501e-05,
+      "loss": 3.154,
+      "step": 3018
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5615034168564922e-05,
+      "loss": 2.9323,
+      "step": 3019
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.560364464692483e-05,
+      "loss": 3.7446,
+      "step": 3020
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5592255125284738e-05,
+      "loss": 3.5303,
+      "step": 3021
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.558086560364465e-05,
+      "loss": 3.5919,
+      "step": 3022
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5569476082004555e-05,
+      "loss": 2.9972,
+      "step": 3023
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5558086560364466e-05,
+      "loss": 3.5567,
+      "step": 3024
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5546697038724374e-05,
+      "loss": 3.2026,
+      "step": 3025
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5535307517084283e-05,
+      "loss": 3.395,
+      "step": 3026
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5523917995444194e-05,
+      "loss": 4.6036,
+      "step": 3027
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.55125284738041e-05,
+      "loss": 3.1596,
+      "step": 3028
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.550113895216401e-05,
+      "loss": 3.1869,
+      "step": 3029
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.548974943052392e-05,
+      "loss": 2.5383,
+      "step": 3030
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5478359908883827e-05,
+      "loss": 4.3053,
+      "step": 3031
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.546697038724374e-05,
+      "loss": 2.106,
+      "step": 3032
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5455580865603643e-05,
+      "loss": 4.0371,
+      "step": 3033
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5444191343963555e-05,
+      "loss": 1.9825,
+      "step": 3034
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5432801822323463e-05,
+      "loss": 2.8544,
+      "step": 3035
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.542141230068337e-05,
+      "loss": 2.8371,
+      "step": 3036
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5410022779043283e-05,
+      "loss": 3.6644,
+      "step": 3037
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5398633257403188e-05,
+      "loss": 3.0677,
+      "step": 3038
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.53872437357631e-05,
+      "loss": 3.1541,
+      "step": 3039
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5375854214123008e-05,
+      "loss": 2.9369,
+      "step": 3040
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5364464692482916e-05,
+      "loss": 3.6574,
+      "step": 3041
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5353075170842827e-05,
+      "loss": 2.609,
+      "step": 3042
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5341685649202732e-05,
+      "loss": 2.6671,
+      "step": 3043
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5330296127562644e-05,
+      "loss": 3.4493,
+      "step": 3044
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5318906605922552e-05,
+      "loss": 2.1013,
+      "step": 3045
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.530751708428246e-05,
+      "loss": 3.4864,
+      "step": 3046
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5296127562642372e-05,
+      "loss": 2.3925,
+      "step": 3047
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5284738041002277e-05,
+      "loss": 2.4401,
+      "step": 3048
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5273348519362188e-05,
+      "loss": 3.5887,
+      "step": 3049
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5261958997722096e-05,
+      "loss": 3.9033,
+      "step": 3050
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5250569476082005e-05,
+      "loss": 2.8584,
+      "step": 3051
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5239179954441915e-05,
+      "loss": 3.693,
+      "step": 3052
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5227790432801823e-05,
+      "loss": 2.2639,
+      "step": 3053
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5216400911161733e-05,
+      "loss": 1.7155,
+      "step": 3054
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5205011389521639e-05,
+      "loss": 4.1288,
+      "step": 3055
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5193621867881549e-05,
+      "loss": 2.5716,
+      "step": 3056
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5182232346241459e-05,
+      "loss": 2.4606,
+      "step": 3057
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5170842824601367e-05,
+      "loss": 3.5746,
+      "step": 3058
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5159453302961277e-05,
+      "loss": 3.7405,
+      "step": 3059
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5148063781321184e-05,
+      "loss": 2.8275,
+      "step": 3060
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5136674259681093e-05,
+      "loss": 3.2271,
+      "step": 3061
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5125284738041003e-05,
+      "loss": 5.4438,
+      "step": 3062
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5113895216400912e-05,
+      "loss": 2.3064,
+      "step": 3063
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5102505694760821e-05,
+      "loss": 4.066,
+      "step": 3064
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5091116173120728e-05,
+      "loss": 4.1082,
+      "step": 3065
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5079726651480638e-05,
+      "loss": 2.5642,
+      "step": 3066
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5068337129840548e-05,
+      "loss": 2.7212,
+      "step": 3067
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5056947608200456e-05,
+      "loss": 3.079,
+      "step": 3068
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5045558086560366e-05,
+      "loss": 2.8375,
+      "step": 3069
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5034168564920272e-05,
+      "loss": 3.3284,
+      "step": 3070
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5022779043280182e-05,
+      "loss": 3.3013,
+      "step": 3071
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5011389521640092e-05,
+      "loss": 3.5083,
+      "step": 3072
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.2177,
+      "step": 3073
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.498861047835991e-05,
+      "loss": 2.9291,
+      "step": 3074
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.4977220956719817e-05,
+      "loss": 2.0091,
+      "step": 3075
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4965831435079727e-05,
+      "loss": 2.5785,
+      "step": 3076
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4954441913439637e-05,
+      "loss": 2.0378,
+      "step": 3077
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4943052391799545e-05,
+      "loss": 2.6861,
+      "step": 3078
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4931662870159455e-05,
+      "loss": 4.2783,
+      "step": 3079
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4920273348519361e-05,
+      "loss": 3.4121,
+      "step": 3080
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4908883826879271e-05,
+      "loss": 5.5332,
+      "step": 3081
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4897494305239183e-05,
+      "loss": 2.7297,
+      "step": 3082
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.488610478359909e-05,
+      "loss": 2.8173,
+      "step": 3083
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4874715261958999e-05,
+      "loss": 2.6603,
+      "step": 3084
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4863325740318906e-05,
+      "loss": 2.0971,
+      "step": 3085
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4851936218678816e-05,
+      "loss": 3.2321,
+      "step": 3086
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4840546697038727e-05,
+      "loss": 2.31,
+      "step": 3087
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4829157175398634e-05,
+      "loss": 3.405,
+      "step": 3088
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4817767653758544e-05,
+      "loss": 3.7553,
+      "step": 3089
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.480637813211845e-05,
+      "loss": 1.9954,
+      "step": 3090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4794988610478362e-05,
+      "loss": 2.6306,
+      "step": 3091
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4783599088838272e-05,
+      "loss": 3.3837,
+      "step": 3092
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4772209567198178e-05,
+      "loss": 3.0936,
+      "step": 3093
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4760820045558088e-05,
+      "loss": 1.8299,
+      "step": 3094
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4749430523917994e-05,
+      "loss": 3.6543,
+      "step": 3095
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4738041002277906e-05,
+      "loss": 2.9743,
+      "step": 3096
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4726651480637816e-05,
+      "loss": 3.8586,
+      "step": 3097
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4715261958997722e-05,
+      "loss": 2.7536,
+      "step": 3098
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4703872437357632e-05,
+      "loss": 1.8797,
+      "step": 3099
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.469248291571754e-05,
+      "loss": 2.5468,
+      "step": 3100
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.468109339407745e-05,
+      "loss": 3.0059,
+      "step": 3101
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4669703872437357e-05,
+      "loss": 3.1155,
+      "step": 3102
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4658314350797267e-05,
+      "loss": 1.9237,
+      "step": 3103
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4646924829157177e-05,
+      "loss": 2.9873,
+      "step": 3104
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4635535307517085e-05,
+      "loss": 3.3309,
+      "step": 3105
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4624145785876995e-05,
+      "loss": 3.699,
+      "step": 3106
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4612756264236901e-05,
+      "loss": 2.8734,
+      "step": 3107
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4601366742596811e-05,
+      "loss": 3.6264,
+      "step": 3108
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4589977220956721e-05,
+      "loss": 2.444,
+      "step": 3109
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.457858769931663e-05,
+      "loss": 2.2763,
+      "step": 3110
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.456719817767654e-05,
+      "loss": 4.2505,
+      "step": 3111
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4555808656036446e-05,
+      "loss": 2.827,
+      "step": 3112
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4544419134396356e-05,
+      "loss": 3.1507,
+      "step": 3113
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4533029612756266e-05,
+      "loss": 3.7534,
+      "step": 3114
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4521640091116174e-05,
+      "loss": 3.4239,
+      "step": 3115
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4510250569476084e-05,
+      "loss": 3.048,
+      "step": 3116
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.449886104783599e-05,
+      "loss": 3.2107,
+      "step": 3117
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.44874715261959e-05,
+      "loss": 4.9397,
+      "step": 3118
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.447608200455581e-05,
+      "loss": 3.8883,
+      "step": 3119
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4464692482915718e-05,
+      "loss": 1.9696,
+      "step": 3120
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4453302961275628e-05,
+      "loss": 2.8303,
+      "step": 3121
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4441913439635535e-05,
+      "loss": 2.11,
+      "step": 3122
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4430523917995445e-05,
+      "loss": 3.9284,
+      "step": 3123
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4419134396355354e-05,
+      "loss": 2.7458,
+      "step": 3124
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4407744874715263e-05,
+      "loss": 3.8462,
+      "step": 3125
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4396355353075173e-05,
+      "loss": 3.121,
+      "step": 3126
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4384965831435079e-05,
+      "loss": 3.4563,
+      "step": 3127
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4373576309794989e-05,
+      "loss": 1.8825,
+      "step": 3128
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4362186788154899e-05,
+      "loss": 4.2251,
+      "step": 3129
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4350797266514807e-05,
+      "loss": 2.4445,
+      "step": 3130
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4339407744874717e-05,
+      "loss": 2.9135,
+      "step": 3131
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4328018223234623e-05,
+      "loss": 2.7562,
+      "step": 3132
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4316628701594533e-05,
+      "loss": 0.8814,
+      "step": 3133
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4305239179954443e-05,
+      "loss": 3.9841,
+      "step": 3134
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4293849658314351e-05,
+      "loss": 2.7561,
+      "step": 3135
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4282460136674261e-05,
+      "loss": 2.589,
+      "step": 3136
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4271070615034168e-05,
+      "loss": 3.1386,
+      "step": 3137
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4259681093394078e-05,
+      "loss": 2.214,
+      "step": 3138
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4248291571753988e-05,
+      "loss": 2.9796,
+      "step": 3139
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4236902050113896e-05,
+      "loss": 3.5547,
+      "step": 3140
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4225512528473806e-05,
+      "loss": 2.4469,
+      "step": 3141
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4214123006833712e-05,
+      "loss": 3.195,
+      "step": 3142
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4202733485193622e-05,
+      "loss": 3.1385,
+      "step": 3143
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4191343963553532e-05,
+      "loss": 4.5993,
+      "step": 3144
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.417995444191344e-05,
+      "loss": 4.8665,
+      "step": 3145
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.416856492027335e-05,
+      "loss": 3.8559,
+      "step": 3146
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4157175398633257e-05,
+      "loss": 4.0278,
+      "step": 3147
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4145785876993167e-05,
+      "loss": 2.6232,
+      "step": 3148
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4134396355353075e-05,
+      "loss": 2.8685,
+      "step": 3149
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4123006833712985e-05,
+      "loss": 2.4884,
+      "step": 3150
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4111617312072895e-05,
+      "loss": 4.7961,
+      "step": 3151
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4100227790432801e-05,
+      "loss": 2.7978,
+      "step": 3152
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4088838268792711e-05,
+      "loss": 2.8624,
+      "step": 3153
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.407744874715262e-05,
+      "loss": 3.7372,
+      "step": 3154
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4066059225512529e-05,
+      "loss": 1.1814,
+      "step": 3155
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4054669703872439e-05,
+      "loss": 1.8097,
+      "step": 3156
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4043280182232346e-05,
+      "loss": 3.5281,
+      "step": 3157
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4031890660592255e-05,
+      "loss": 2.7231,
+      "step": 3158
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4020501138952164e-05,
+      "loss": 1.6094,
+      "step": 3159
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4009111617312074e-05,
+      "loss": 2.9563,
+      "step": 3160
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.3997722095671983e-05,
+      "loss": 3.1267,
+      "step": 3161
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.398633257403189e-05,
+      "loss": 2.805,
+      "step": 3162
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.39749430523918e-05,
+      "loss": 3.3267,
+      "step": 3163
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3963553530751708e-05,
+      "loss": 2.2,
+      "step": 3164
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3952164009111618e-05,
+      "loss": 3.0815,
+      "step": 3165
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3940774487471528e-05,
+      "loss": 4.4272,
+      "step": 3166
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3929384965831434e-05,
+      "loss": 2.6256,
+      "step": 3167
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3917995444191346e-05,
+      "loss": 1.5353,
+      "step": 3168
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3906605922551252e-05,
+      "loss": 2.6274,
+      "step": 3169
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3895216400911162e-05,
+      "loss": 4.7346,
+      "step": 3170
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3883826879271072e-05,
+      "loss": 2.4449,
+      "step": 3171
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3872437357630979e-05,
+      "loss": 1.3789,
+      "step": 3172
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.386104783599089e-05,
+      "loss": 2.6546,
+      "step": 3173
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3849658314350797e-05,
+      "loss": 2.6065,
+      "step": 3174
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3838268792710707e-05,
+      "loss": 3.2674,
+      "step": 3175
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3826879271070617e-05,
+      "loss": 5.6578,
+      "step": 3176
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3815489749430525e-05,
+      "loss": 2.7544,
+      "step": 3177
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3804100227790435e-05,
+      "loss": 3.1076,
+      "step": 3178
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3792710706150341e-05,
+      "loss": 1.4785,
+      "step": 3179
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3781321184510251e-05,
+      "loss": 4.5636,
+      "step": 3180
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3769931662870161e-05,
+      "loss": 2.8664,
+      "step": 3181
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.375854214123007e-05,
+      "loss": 2.3146,
+      "step": 3182
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.374715261958998e-05,
+      "loss": 2.9639,
+      "step": 3183
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3735763097949886e-05,
+      "loss": 3.767,
+      "step": 3184
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3724373576309796e-05,
+      "loss": 1.6593,
+      "step": 3185
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3712984054669706e-05,
+      "loss": 2.9697,
+      "step": 3186
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3701594533029614e-05,
+      "loss": 1.1034,
+      "step": 3187
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3690205011389524e-05,
+      "loss": 3.4116,
+      "step": 3188
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.367881548974943e-05,
+      "loss": 2.0196,
+      "step": 3189
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.366742596810934e-05,
+      "loss": 1.7579,
+      "step": 3190
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.365603644646925e-05,
+      "loss": 2.52,
+      "step": 3191
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3644646924829158e-05,
+      "loss": 2.9805,
+      "step": 3192
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3633257403189068e-05,
+      "loss": 2.0162,
+      "step": 3193
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3621867881548975e-05,
+      "loss": 1.5596,
+      "step": 3194
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3610478359908884e-05,
+      "loss": 2.9375,
+      "step": 3195
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3599088838268793e-05,
+      "loss": 2.939,
+      "step": 3196
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3587699316628703e-05,
+      "loss": 3.5423,
+      "step": 3197
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3576309794988612e-05,
+      "loss": 1.8455,
+      "step": 3198
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3564920273348519e-05,
+      "loss": 2.1613,
+      "step": 3199
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3553530751708429e-05,
+      "loss": 1.8204,
+      "step": 3200
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3542141230068337e-05,
+      "loss": 2.9351,
+      "step": 3201
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3530751708428247e-05,
+      "loss": 2.1278,
+      "step": 3202
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3519362186788157e-05,
+      "loss": 2.4726,
+      "step": 3203
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3507972665148063e-05,
+      "loss": 3.3353,
+      "step": 3204
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3496583143507973e-05,
+      "loss": 4.9907,
+      "step": 3205
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3485193621867882e-05,
+      "loss": 3.1104,
+      "step": 3206
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3473804100227791e-05,
+      "loss": 2.7792,
+      "step": 3207
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3462414578587701e-05,
+      "loss": 5.1795,
+      "step": 3208
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3451025056947608e-05,
+      "loss": 3.9637,
+      "step": 3209
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3439635535307518e-05,
+      "loss": 2.6229,
+      "step": 3210
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3428246013667426e-05,
+      "loss": 3.0132,
+      "step": 3211
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3416856492027336e-05,
+      "loss": 3.9524,
+      "step": 3212
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3405466970387246e-05,
+      "loss": 3.5585,
+      "step": 3213
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3394077448747152e-05,
+      "loss": 2.1179,
+      "step": 3214
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3382687927107062e-05,
+      "loss": 2.9216,
+      "step": 3215
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.337129840546697e-05,
+      "loss": 3.3679,
+      "step": 3216
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.335990888382688e-05,
+      "loss": 4.9873,
+      "step": 3217
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.334851936218679e-05,
+      "loss": 2.9866,
+      "step": 3218
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.3337129840546697e-05,
+      "loss": 2.2416,
+      "step": 3219
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.3325740318906607e-05,
+      "loss": 3.4121,
+      "step": 3220
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3314350797266515e-05,
+      "loss": 2.78,
+      "step": 3221
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3302961275626425e-05,
+      "loss": 3.7068,
+      "step": 3222
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3291571753986335e-05,
+      "loss": 3.3301,
+      "step": 3223
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3280182232346241e-05,
+      "loss": 3.034,
+      "step": 3224
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3268792710706151e-05,
+      "loss": 4.0155,
+      "step": 3225
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.325740318906606e-05,
+      "loss": 2.4872,
+      "step": 3226
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3246013667425969e-05,
+      "loss": 2.9565,
+      "step": 3227
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3234624145785879e-05,
+      "loss": 4.7031,
+      "step": 3228
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3223234624145785e-05,
+      "loss": 4.6455,
+      "step": 3229
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3211845102505695e-05,
+      "loss": 2.6889,
+      "step": 3230
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3200455580865604e-05,
+      "loss": 3.6603,
+      "step": 3231
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3189066059225513e-05,
+      "loss": 2.16,
+      "step": 3232
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3177676537585423e-05,
+      "loss": 2.0516,
+      "step": 3233
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.316628701594533e-05,
+      "loss": 3.6734,
+      "step": 3234
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.315489749430524e-05,
+      "loss": 3.9174,
+      "step": 3235
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3143507972665148e-05,
+      "loss": 3.0504,
+      "step": 3236
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3132118451025058e-05,
+      "loss": 2.2326,
+      "step": 3237
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3120728929384968e-05,
+      "loss": 4.3866,
+      "step": 3238
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3109339407744874e-05,
+      "loss": 1.3325,
+      "step": 3239
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3097949886104784e-05,
+      "loss": 2.8659,
+      "step": 3240
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3086560364464692e-05,
+      "loss": 4.2068,
+      "step": 3241
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3075170842824602e-05,
+      "loss": 3.0955,
+      "step": 3242
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3063781321184512e-05,
+      "loss": 4.8819,
+      "step": 3243
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3052391799544419e-05,
+      "loss": 2.5625,
+      "step": 3244
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.304100227790433e-05,
+      "loss": 3.8384,
+      "step": 3245
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3029612756264237e-05,
+      "loss": 1.8579,
+      "step": 3246
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3018223234624147e-05,
+      "loss": 4.0999,
+      "step": 3247
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3006833712984053e-05,
+      "loss": 3.2641,
+      "step": 3248
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.2995444191343963e-05,
+      "loss": 2.2446,
+      "step": 3249
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.2984054669703875e-05,
+      "loss": 2.0221,
+      "step": 3250
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2972665148063781e-05,
+      "loss": 2.1896,
+      "step": 3251
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2961275626423691e-05,
+      "loss": 2.9257,
+      "step": 3252
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2949886104783598e-05,
+      "loss": 2.7899,
+      "step": 3253
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.293849658314351e-05,
+      "loss": 3.0252,
+      "step": 3254
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.292710706150342e-05,
+      "loss": 3.5721,
+      "step": 3255
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2915717539863326e-05,
+      "loss": 3.5433,
+      "step": 3256
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2904328018223236e-05,
+      "loss": 3.4108,
+      "step": 3257
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2892938496583142e-05,
+      "loss": 2.5158,
+      "step": 3258
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2881548974943054e-05,
+      "loss": 2.4261,
+      "step": 3259
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2870159453302964e-05,
+      "loss": 1.8262,
+      "step": 3260
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.285876993166287e-05,
+      "loss": 2.7386,
+      "step": 3261
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.284738041002278e-05,
+      "loss": 3.7182,
+      "step": 3262
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2835990888382688e-05,
+      "loss": 1.7964,
+      "step": 3263
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2824601366742598e-05,
+      "loss": 3.2489,
+      "step": 3264
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2813211845102508e-05,
+      "loss": 3.0855,
+      "step": 3265
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2801822323462415e-05,
+      "loss": 1.7505,
+      "step": 3266
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2790432801822324e-05,
+      "loss": 2.944,
+      "step": 3267
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2779043280182233e-05,
+      "loss": 2.7019,
+      "step": 3268
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2767653758542143e-05,
+      "loss": 2.803,
+      "step": 3269
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2756264236902052e-05,
+      "loss": 4.0907,
+      "step": 3270
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2744874715261959e-05,
+      "loss": 3.3283,
+      "step": 3271
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2733485193621869e-05,
+      "loss": 1.6896,
+      "step": 3272
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2722095671981777e-05,
+      "loss": 2.5816,
+      "step": 3273
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2710706150341687e-05,
+      "loss": 1.0124,
+      "step": 3274
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2699316628701597e-05,
+      "loss": 3.0147,
+      "step": 3275
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2687927107061503e-05,
+      "loss": 4.0424,
+      "step": 3276
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2676537585421413e-05,
+      "loss": 1.5296,
+      "step": 3277
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2665148063781321e-05,
+      "loss": 3.1854,
+      "step": 3278
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2653758542141231e-05,
+      "loss": 2.7429,
+      "step": 3279
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2642369020501141e-05,
+      "loss": 4.4375,
+      "step": 3280
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2630979498861048e-05,
+      "loss": 3.674,
+      "step": 3281
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2619589977220958e-05,
+      "loss": 1.8135,
+      "step": 3282
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2608200455580866e-05,
+      "loss": 2.9467,
+      "step": 3283
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2596810933940776e-05,
+      "loss": 3.2462,
+      "step": 3284
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2585421412300686e-05,
+      "loss": 1.3319,
+      "step": 3285
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2574031890660592e-05,
+      "loss": 3.2687,
+      "step": 3286
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2562642369020502e-05,
+      "loss": 2.4731,
+      "step": 3287
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.255125284738041e-05,
+      "loss": 1.3768,
+      "step": 3288
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.253986332574032e-05,
+      "loss": 1.8871,
+      "step": 3289
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.252847380410023e-05,
+      "loss": 1.6943,
+      "step": 3290
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2517084282460137e-05,
+      "loss": 2.6078,
+      "step": 3291
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2505694760820046e-05,
+      "loss": 3.4281,
+      "step": 3292
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2494305239179956e-05,
+      "loss": 1.094,
+      "step": 3293
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2482915717539865e-05,
+      "loss": 2.7254,
+      "step": 3294
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2471526195899773e-05,
+      "loss": 4.7252,
+      "step": 3295
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2460136674259681e-05,
+      "loss": 3.5749,
+      "step": 3296
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2448747152619591e-05,
+      "loss": 2.1227,
+      "step": 3297
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.24373576309795e-05,
+      "loss": 4.8484,
+      "step": 3298
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2425968109339409e-05,
+      "loss": 4.2538,
+      "step": 3299
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2414578587699317e-05,
+      "loss": 3.4228,
+      "step": 3300
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2403189066059225e-05,
+      "loss": 3.3854,
+      "step": 3301
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2391799544419135e-05,
+      "loss": 1.6288,
+      "step": 3302
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2380410022779045e-05,
+      "loss": 4.1435,
+      "step": 3303
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2369020501138953e-05,
+      "loss": 2.8972,
+      "step": 3304
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2357630979498862e-05,
+      "loss": 3.1284,
+      "step": 3305
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.234624145785877e-05,
+      "loss": 3.4374,
+      "step": 3306
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.233485193621868e-05,
+      "loss": 2.6575,
+      "step": 3307
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2323462414578588e-05,
+      "loss": 4.384,
+      "step": 3308
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2312072892938498e-05,
+      "loss": 2.9164,
+      "step": 3309
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2300683371298406e-05,
+      "loss": 2.5393,
+      "step": 3310
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2289293849658314e-05,
+      "loss": 2.5602,
+      "step": 3311
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2277904328018224e-05,
+      "loss": 1.1649,
+      "step": 3312
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2266514806378132e-05,
+      "loss": 3.615,
+      "step": 3313
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2255125284738042e-05,
+      "loss": 2.5717,
+      "step": 3314
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.224373576309795e-05,
+      "loss": 3.4119,
+      "step": 3315
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2232346241457859e-05,
+      "loss": 2.8504,
+      "step": 3316
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2220956719817769e-05,
+      "loss": 2.924,
+      "step": 3317
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2209567198177677e-05,
+      "loss": 2.4722,
+      "step": 3318
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2198177676537587e-05,
+      "loss": 3.9962,
+      "step": 3319
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2186788154897495e-05,
+      "loss": 3.0145,
+      "step": 3320
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2175398633257403e-05,
+      "loss": 1.5782,
+      "step": 3321
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2164009111617313e-05,
+      "loss": 3.2134,
+      "step": 3322
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2152619589977221e-05,
+      "loss": 2.7087,
+      "step": 3323
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2141230068337131e-05,
+      "loss": 3.8259,
+      "step": 3324
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.212984054669704e-05,
+      "loss": 2.9813,
+      "step": 3325
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2118451025056948e-05,
+      "loss": 3.1692,
+      "step": 3326
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2107061503416857e-05,
+      "loss": 2.2753,
+      "step": 3327
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2095671981776766e-05,
+      "loss": 5.3322,
+      "step": 3328
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2084282460136676e-05,
+      "loss": 3.8758,
+      "step": 3329
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2072892938496584e-05,
+      "loss": 4.1062,
+      "step": 3330
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2061503416856492e-05,
+      "loss": 3.7858,
+      "step": 3331
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2050113895216402e-05,
+      "loss": 3.9504,
+      "step": 3332
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.203872437357631e-05,
+      "loss": 1.3278,
+      "step": 3333
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2027334851936218e-05,
+      "loss": 3.5386,
+      "step": 3334
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2015945330296128e-05,
+      "loss": 2.876,
+      "step": 3335
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2004555808656038e-05,
+      "loss": 3.8451,
+      "step": 3336
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.1993166287015946e-05,
+      "loss": 2.3914,
+      "step": 3337
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.1981776765375854e-05,
+      "loss": 2.1905,
+      "step": 3338
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1970387243735763e-05,
+      "loss": 3.8676,
+      "step": 3339
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1958997722095673e-05,
+      "loss": 4.9116,
+      "step": 3340
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1947608200455582e-05,
+      "loss": 3.722,
+      "step": 3341
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.193621867881549e-05,
+      "loss": 3.5588,
+      "step": 3342
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1924829157175399e-05,
+      "loss": 4.3318,
+      "step": 3343
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1913439635535307e-05,
+      "loss": 2.2037,
+      "step": 3344
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1902050113895217e-05,
+      "loss": 2.7012,
+      "step": 3345
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1890660592255127e-05,
+      "loss": 2.8429,
+      "step": 3346
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1879271070615035e-05,
+      "loss": 2.8383,
+      "step": 3347
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1867881548974943e-05,
+      "loss": 2.2722,
+      "step": 3348
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1856492027334851e-05,
+      "loss": 1.7064,
+      "step": 3349
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1845102505694761e-05,
+      "loss": 4.8311,
+      "step": 3350
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1833712984054671e-05,
+      "loss": 2.8375,
+      "step": 3351
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.182232346241458e-05,
+      "loss": 3.3989,
+      "step": 3352
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1810933940774488e-05,
+      "loss": 3.171,
+      "step": 3353
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1799544419134396e-05,
+      "loss": 1.303,
+      "step": 3354
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1788154897494306e-05,
+      "loss": 4.4291,
+      "step": 3355
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1776765375854216e-05,
+      "loss": 2.9186,
+      "step": 3356
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1765375854214124e-05,
+      "loss": 2.3581,
+      "step": 3357
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1753986332574032e-05,
+      "loss": 3.205,
+      "step": 3358
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.174259681093394e-05,
+      "loss": 3.3829,
+      "step": 3359
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.173120728929385e-05,
+      "loss": 2.5299,
+      "step": 3360
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.171981776765376e-05,
+      "loss": 4.2982,
+      "step": 3361
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1708428246013668e-05,
+      "loss": 3.6363,
+      "step": 3362
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1697038724373577e-05,
+      "loss": 1.8157,
+      "step": 3363
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1685649202733485e-05,
+      "loss": 3.4171,
+      "step": 3364
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1674259681093395e-05,
+      "loss": 1.9433,
+      "step": 3365
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1662870159453305e-05,
+      "loss": 1.6253,
+      "step": 3366
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1651480637813213e-05,
+      "loss": 2.4062,
+      "step": 3367
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1640091116173121e-05,
+      "loss": 3.1566,
+      "step": 3368
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.162870159453303e-05,
+      "loss": 2.7383,
+      "step": 3369
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1617312072892939e-05,
+      "loss": 2.6231,
+      "step": 3370
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1605922551252849e-05,
+      "loss": 2.8345,
+      "step": 3371
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1594533029612757e-05,
+      "loss": 3.8802,
+      "step": 3372
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1583143507972665e-05,
+      "loss": 3.3014,
+      "step": 3373
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1571753986332574e-05,
+      "loss": 4.0631,
+      "step": 3374
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1560364464692483e-05,
+      "loss": 1.1445,
+      "step": 3375
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1548974943052393e-05,
+      "loss": 3.6625,
+      "step": 3376
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1537585421412302e-05,
+      "loss": 3.736,
+      "step": 3377
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.152619589977221e-05,
+      "loss": 3.8411,
+      "step": 3378
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.151480637813212e-05,
+      "loss": 2.071,
+      "step": 3379
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1503416856492028e-05,
+      "loss": 2.3808,
+      "step": 3380
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1492027334851936e-05,
+      "loss": 1.6964,
+      "step": 3381
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1480637813211846e-05,
+      "loss": 3.7907,
+      "step": 3382
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1469248291571754e-05,
+      "loss": 4.989,
+      "step": 3383
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1457858769931664e-05,
+      "loss": 2.4904,
+      "step": 3384
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1446469248291572e-05,
+      "loss": 2.2521,
+      "step": 3385
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.143507972665148e-05,
+      "loss": 1.8961,
+      "step": 3386
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.142369020501139e-05,
+      "loss": 2.4781,
+      "step": 3387
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1412300683371299e-05,
+      "loss": 4.6428,
+      "step": 3388
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1400911161731209e-05,
+      "loss": 3.1212,
+      "step": 3389
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1389521640091117e-05,
+      "loss": 3.1786,
+      "step": 3390
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1378132118451025e-05,
+      "loss": 1.5812,
+      "step": 3391
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1366742596810935e-05,
+      "loss": 1.9978,
+      "step": 3392
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1355353075170843e-05,
+      "loss": 2.0298,
+      "step": 3393
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1343963553530753e-05,
+      "loss": 4.7618,
+      "step": 3394
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1332574031890661e-05,
+      "loss": 3.1684,
+      "step": 3395
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.132118451025057e-05,
+      "loss": 2.5381,
+      "step": 3396
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.130979498861048e-05,
+      "loss": 3.4487,
+      "step": 3397
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1298405466970387e-05,
+      "loss": 2.8577,
+      "step": 3398
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1287015945330297e-05,
+      "loss": 3.7337,
+      "step": 3399
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1275626423690206e-05,
+      "loss": 1.4548,
+      "step": 3400
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1264236902050114e-05,
+      "loss": 3.8976,
+      "step": 3401
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1252847380410024e-05,
+      "loss": 5.4643,
+      "step": 3402
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1241457858769932e-05,
+      "loss": 2.5614,
+      "step": 3403
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1230068337129842e-05,
+      "loss": 3.3805,
+      "step": 3404
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.121867881548975e-05,
+      "loss": 3.0641,
+      "step": 3405
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1207289293849658e-05,
+      "loss": 2.7107,
+      "step": 3406
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1195899772209566e-05,
+      "loss": 3.1404,
+      "step": 3407
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1184510250569476e-05,
+      "loss": 3.3978,
+      "step": 3408
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1173120728929386e-05,
+      "loss": 3.8524,
+      "step": 3409
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1161731207289294e-05,
+      "loss": 4.1745,
+      "step": 3410
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1150341685649203e-05,
+      "loss": 2.3119,
+      "step": 3411
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.113895216400911e-05,
+      "loss": 3.2208,
+      "step": 3412
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1127562642369022e-05,
+      "loss": 3.098,
+      "step": 3413
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.111617312072893e-05,
+      "loss": 2.1281,
+      "step": 3414
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1104783599088839e-05,
+      "loss": 4.6247,
+      "step": 3415
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1093394077448747e-05,
+      "loss": 3.6888,
+      "step": 3416
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1082004555808655e-05,
+      "loss": 4.0267,
+      "step": 3417
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1070615034168567e-05,
+      "loss": 3.3316,
+      "step": 3418
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1059225512528475e-05,
+      "loss": 2.0966,
+      "step": 3419
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1047835990888383e-05,
+      "loss": 2.5159,
+      "step": 3420
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1036446469248291e-05,
+      "loss": 4.159,
+      "step": 3421
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1025056947608201e-05,
+      "loss": 4.085,
+      "step": 3422
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.1013667425968111e-05,
+      "loss": 2.9536,
+      "step": 3423
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.100227790432802e-05,
+      "loss": 3.7735,
+      "step": 3424
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.0990888382687928e-05,
+      "loss": 3.6828,
+      "step": 3425
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.0979498861047836e-05,
+      "loss": 2.6309,
+      "step": 3426
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0968109339407746e-05,
+      "loss": 2.9918,
+      "step": 3427
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0956719817767654e-05,
+      "loss": 3.1538,
+      "step": 3428
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0945330296127564e-05,
+      "loss": 2.5625,
+      "step": 3429
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0933940774487472e-05,
+      "loss": 3.7187,
+      "step": 3430
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.092255125284738e-05,
+      "loss": 2.1556,
+      "step": 3431
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.091116173120729e-05,
+      "loss": 2.0278,
+      "step": 3432
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0899772209567198e-05,
+      "loss": 4.5628,
+      "step": 3433
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0888382687927108e-05,
+      "loss": 2.4322,
+      "step": 3434
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0876993166287016e-05,
+      "loss": 2.2497,
+      "step": 3435
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0865603644646925e-05,
+      "loss": 2.8602,
+      "step": 3436
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0854214123006835e-05,
+      "loss": 3.9662,
+      "step": 3437
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0842824601366743e-05,
+      "loss": 2.7007,
+      "step": 3438
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0831435079726653e-05,
+      "loss": 4.0493,
+      "step": 3439
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0820045558086561e-05,
+      "loss": 1.8066,
+      "step": 3440
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0808656036446469e-05,
+      "loss": 3.1366,
+      "step": 3441
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0797266514806379e-05,
+      "loss": 2.3236,
+      "step": 3442
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0785876993166287e-05,
+      "loss": 1.9678,
+      "step": 3443
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0774487471526197e-05,
+      "loss": 3.3805,
+      "step": 3444
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0763097949886105e-05,
+      "loss": 2.8133,
+      "step": 3445
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0751708428246014e-05,
+      "loss": 1.372,
+      "step": 3446
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0740318906605923e-05,
+      "loss": 3.15,
+      "step": 3447
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0728929384965832e-05,
+      "loss": 2.3623,
+      "step": 3448
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0717539863325742e-05,
+      "loss": 2.6486,
+      "step": 3449
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.070615034168565e-05,
+      "loss": 2.3385,
+      "step": 3450
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0694760820045558e-05,
+      "loss": 4.0267,
+      "step": 3451
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0683371298405468e-05,
+      "loss": 4.2628,
+      "step": 3452
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0671981776765376e-05,
+      "loss": 2.081,
+      "step": 3453
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0660592255125284e-05,
+      "loss": 2.8096,
+      "step": 3454
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0649202733485194e-05,
+      "loss": 2.9333,
+      "step": 3455
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0637813211845104e-05,
+      "loss": 3.0781,
+      "step": 3456
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0626423690205012e-05,
+      "loss": 3.1842,
+      "step": 3457
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.061503416856492e-05,
+      "loss": 2.455,
+      "step": 3458
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0603644646924829e-05,
+      "loss": 3.2516,
+      "step": 3459
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0592255125284739e-05,
+      "loss": 1.0259,
+      "step": 3460
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0580865603644648e-05,
+      "loss": 2.3893,
+      "step": 3461
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0569476082004557e-05,
+      "loss": 2.1625,
+      "step": 3462
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0558086560364465e-05,
+      "loss": 2.8332,
+      "step": 3463
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0546697038724373e-05,
+      "loss": 2.2264,
+      "step": 3464
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0535307517084283e-05,
+      "loss": 1.9287,
+      "step": 3465
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0523917995444193e-05,
+      "loss": 2.083,
+      "step": 3466
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0512528473804101e-05,
+      "loss": 3.0274,
+      "step": 3467
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.050113895216401e-05,
+      "loss": 3.4347,
+      "step": 3468
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0489749430523917e-05,
+      "loss": 2.4516,
+      "step": 3469
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0478359908883827e-05,
+      "loss": 1.091,
+      "step": 3470
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0466970387243737e-05,
+      "loss": 3.0185,
+      "step": 3471
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0455580865603645e-05,
+      "loss": 2.3746,
+      "step": 3472
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0444191343963554e-05,
+      "loss": 2.3607,
+      "step": 3473
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0432801822323462e-05,
+      "loss": 3.9295,
+      "step": 3474
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0421412300683372e-05,
+      "loss": 2.5532,
+      "step": 3475
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0410022779043282e-05,
+      "loss": 4.4921,
+      "step": 3476
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.039863325740319e-05,
+      "loss": 2.8805,
+      "step": 3477
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0387243735763098e-05,
+      "loss": 1.5059,
+      "step": 3478
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0375854214123006e-05,
+      "loss": 3.1061,
+      "step": 3479
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0364464692482916e-05,
+      "loss": 2.6487,
+      "step": 3480
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0353075170842826e-05,
+      "loss": 4.8689,
+      "step": 3481
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0341685649202734e-05,
+      "loss": 3.3902,
+      "step": 3482
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0330296127562643e-05,
+      "loss": 3.4602,
+      "step": 3483
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.031890660592255e-05,
+      "loss": 3.8937,
+      "step": 3484
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.030751708428246e-05,
+      "loss": 3.1245,
+      "step": 3485
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.029612756264237e-05,
+      "loss": 3.2823,
+      "step": 3486
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0284738041002279e-05,
+      "loss": 2.7035,
+      "step": 3487
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0273348519362187e-05,
+      "loss": 3.418,
+      "step": 3488
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0261958997722095e-05,
+      "loss": 1.3544,
+      "step": 3489
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0250569476082005e-05,
+      "loss": 3.1002,
+      "step": 3490
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0239179954441915e-05,
+      "loss": 2.0131,
+      "step": 3491
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0227790432801823e-05,
+      "loss": 3.8011,
+      "step": 3492
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.0216400911161731e-05,
+      "loss": 2.9814,
+      "step": 3493
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.020501138952164e-05,
+      "loss": 4.8025,
+      "step": 3494
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.019362186788155e-05,
+      "loss": 1.1792,
+      "step": 3495
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.018223234624146e-05,
+      "loss": 2.6159,
+      "step": 3496
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0170842824601368e-05,
+      "loss": 4.3485,
+      "step": 3497
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0159453302961276e-05,
+      "loss": 2.3739,
+      "step": 3498
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0148063781321186e-05,
+      "loss": 4.433,
+      "step": 3499
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0136674259681094e-05,
+      "loss": 2.8988,
+      "step": 3500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0125284738041002e-05,
+      "loss": 2.6017,
+      "step": 3501
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0113895216400912e-05,
+      "loss": 3.7286,
+      "step": 3502
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.010250569476082e-05,
+      "loss": 3.3221,
+      "step": 3503
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.009111617312073e-05,
+      "loss": 0.8642,
+      "step": 3504
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0079726651480638e-05,
+      "loss": 2.4734,
+      "step": 3505
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0068337129840547e-05,
+      "loss": 4.9708,
+      "step": 3506
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0056947608200456e-05,
+      "loss": 3.9824,
+      "step": 3507
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0045558086560365e-05,
+      "loss": 2.0894,
+      "step": 3508
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0034168564920275e-05,
+      "loss": 2.5177,
+      "step": 3509
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0022779043280183e-05,
+      "loss": 2.9656,
+      "step": 3510
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0011389521640091e-05,
+      "loss": 4.0893,
+      "step": 3511
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1e-05,
+      "loss": 2.7867,
+      "step": 3512
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.988610478359909e-06,
+      "loss": 4.0771,
+      "step": 3513
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.977220956719819e-06,
+      "loss": 2.6884,
+      "step": 3514
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.965831435079727e-06,
+      "loss": 2.0674,
+      "step": 3515
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.954441913439635e-06,
+      "loss": 2.964,
+      "step": 3516
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.943052391799545e-06,
+      "loss": 3.0298,
+      "step": 3517
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.931662870159453e-06,
+      "loss": 2.4619,
+      "step": 3518
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.920273348519363e-06,
+      "loss": 4.1926,
+      "step": 3519
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.908883826879272e-06,
+      "loss": 3.7532,
+      "step": 3520
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.89749430523918e-06,
+      "loss": 4.0312,
+      "step": 3521
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.88610478359909e-06,
+      "loss": 4.3728,
+      "step": 3522
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.874715261958998e-06,
+      "loss": 3.4611,
+      "step": 3523
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.863325740318908e-06,
+      "loss": 3.517,
+      "step": 3524
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.851936218678816e-06,
+      "loss": 3.1172,
+      "step": 3525
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.840546697038724e-06,
+      "loss": 4.937,
+      "step": 3526
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.829157175398632e-06,
+      "loss": 2.4175,
+      "step": 3527
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.817767653758542e-06,
+      "loss": 3.6157,
+      "step": 3528
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.806378132118452e-06,
+      "loss": 4.0185,
+      "step": 3529
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.79498861047836e-06,
+      "loss": 1.9772,
+      "step": 3530
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.783599088838269e-06,
+      "loss": 2.957,
+      "step": 3531
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.772209567198177e-06,
+      "loss": 3.4806,
+      "step": 3532
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.760820045558088e-06,
+      "loss": 4.3308,
+      "step": 3533
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.749430523917997e-06,
+      "loss": 2.8729,
+      "step": 3534
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.738041002277905e-06,
+      "loss": 3.6986,
+      "step": 3535
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.726651480637813e-06,
+      "loss": 3.4175,
+      "step": 3536
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.715261958997721e-06,
+      "loss": 2.0817,
+      "step": 3537
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.703872437357633e-06,
+      "loss": 2.5517,
+      "step": 3538
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.692482915717541e-06,
+      "loss": 2.0982,
+      "step": 3539
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.68109339407745e-06,
+      "loss": 0.9525,
+      "step": 3540
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.669703872437357e-06,
+      "loss": 4.1366,
+      "step": 3541
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.658314350797267e-06,
+      "loss": 1.978,
+      "step": 3542
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.646924829157177e-06,
+      "loss": 2.5424,
+      "step": 3543
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.635535307517085e-06,
+      "loss": 3.9194,
+      "step": 3544
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.624145785876994e-06,
+      "loss": 3.4545,
+      "step": 3545
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.612756264236902e-06,
+      "loss": 3.0287,
+      "step": 3546
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.601366742596812e-06,
+      "loss": 2.0911,
+      "step": 3547
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.589977220956722e-06,
+      "loss": 1.573,
+      "step": 3548
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.57858769931663e-06,
+      "loss": 3.7163,
+      "step": 3549
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.567198177676538e-06,
+      "loss": 1.0272,
+      "step": 3550
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.555808656036446e-06,
+      "loss": 1.8948,
+      "step": 3551
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.544419134396356e-06,
+      "loss": 1.2308,
+      "step": 3552
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.533029612756264e-06,
+      "loss": 3.0702,
+      "step": 3553
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.521640091116174e-06,
+      "loss": 2.7702,
+      "step": 3554
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.510250569476082e-06,
+      "loss": 4.5454,
+      "step": 3555
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.49886104783599e-06,
+      "loss": 2.9198,
+      "step": 3556
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.4874715261959e-06,
+      "loss": 3.1807,
+      "step": 3557
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.476082004555809e-06,
+      "loss": 2.4089,
+      "step": 3558
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.464692482915719e-06,
+      "loss": 1.3667,
+      "step": 3559
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.453302961275627e-06,
+      "loss": 1.8697,
+      "step": 3560
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.441913439635535e-06,
+      "loss": 3.6913,
+      "step": 3561
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.430523917995445e-06,
+      "loss": 1.904,
+      "step": 3562
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.419134396355353e-06,
+      "loss": 1.6796,
+      "step": 3563
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.407744874715263e-06,
+      "loss": 3.1162,
+      "step": 3564
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.396355353075171e-06,
+      "loss": 2.908,
+      "step": 3565
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.38496583143508e-06,
+      "loss": 2.8792,
+      "step": 3566
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.37357630979499e-06,
+      "loss": 2.1377,
+      "step": 3567
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.362186788154898e-06,
+      "loss": 2.8827,
+      "step": 3568
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.350797266514808e-06,
+      "loss": 1.9878,
+      "step": 3569
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.339407744874716e-06,
+      "loss": 2.1384,
+      "step": 3570
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.328018223234624e-06,
+      "loss": 1.3539,
+      "step": 3571
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.316628701594534e-06,
+      "loss": 3.0606,
+      "step": 3572
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.305239179954442e-06,
+      "loss": 2.8904,
+      "step": 3573
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.29384965831435e-06,
+      "loss": 1.8,
+      "step": 3574
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.28246013667426e-06,
+      "loss": 2.8624,
+      "step": 3575
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.27107061503417e-06,
+      "loss": 4.0602,
+      "step": 3576
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.259681093394078e-06,
+      "loss": 3.6994,
+      "step": 3577
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.248291571753986e-06,
+      "loss": 1.8952,
+      "step": 3578
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.236902050113895e-06,
+      "loss": 2.2434,
+      "step": 3579
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.225512528473805e-06,
+      "loss": 2.1127,
+      "step": 3580
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.214123006833714e-06,
+      "loss": 3.2725,
+      "step": 3581
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.202733485193623e-06,
+      "loss": 3.1991,
+      "step": 3582
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.191343963553531e-06,
+      "loss": 2.0534,
+      "step": 3583
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.179954441913439e-06,
+      "loss": 2.3407,
+      "step": 3584
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.168564920273349e-06,
+      "loss": 1.6514,
+      "step": 3585
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.157175398633259e-06,
+      "loss": 2.308,
+      "step": 3586
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.145785876993167e-06,
+      "loss": 2.79,
+      "step": 3587
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.134396355353075e-06,
+      "loss": 2.8917,
+      "step": 3588
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.123006833712983e-06,
+      "loss": 3.121,
+      "step": 3589
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.111617312072893e-06,
+      "loss": 2.9013,
+      "step": 3590
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.100227790432803e-06,
+      "loss": 3.396,
+      "step": 3591
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.088838268792711e-06,
+      "loss": 2.3567,
+      "step": 3592
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.07744874715262e-06,
+      "loss": 2.7524,
+      "step": 3593
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.066059225512528e-06,
+      "loss": 2.6,
+      "step": 3594
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.054669703872438e-06,
+      "loss": 2.7274,
+      "step": 3595
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.043280182232348e-06,
+      "loss": 3.5166,
+      "step": 3596
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.031890660592256e-06,
+      "loss": 1.63,
+      "step": 3597
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.020501138952164e-06,
+      "loss": 2.5159,
+      "step": 3598
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.009111617312072e-06,
+      "loss": 2.4986,
+      "step": 3599
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.997722095671982e-06,
+      "loss": 3.0451,
+      "step": 3600
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.986332574031892e-06,
+      "loss": 3.4799,
+      "step": 3601
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.9749430523918e-06,
+      "loss": 3.249,
+      "step": 3602
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.963553530751709e-06,
+      "loss": 1.9161,
+      "step": 3603
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.952164009111617e-06,
+      "loss": 2.1529,
+      "step": 3604
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.940774487471527e-06,
+      "loss": 2.3142,
+      "step": 3605
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.929384965831437e-06,
+      "loss": 2.8985,
+      "step": 3606
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.917995444191345e-06,
+      "loss": 3.8248,
+      "step": 3607
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.906605922551253e-06,
+      "loss": 2.2902,
+      "step": 3608
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.895216400911161e-06,
+      "loss": 3.1731,
+      "step": 3609
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.883826879271071e-06,
+      "loss": 3.3976,
+      "step": 3610
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.872437357630981e-06,
+      "loss": 3.7905,
+      "step": 3611
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.861047835990889e-06,
+      "loss": 1.2724,
+      "step": 3612
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.849658314350797e-06,
+      "loss": 2.12,
+      "step": 3613
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.838268792710706e-06,
+      "loss": 3.8789,
+      "step": 3614
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.826879271070615e-06,
+      "loss": 2.2769,
+      "step": 3615
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.815489749430525e-06,
+      "loss": 4.0606,
+      "step": 3616
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.804100227790434e-06,
+      "loss": 2.4548,
+      "step": 3617
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.792710706150342e-06,
+      "loss": 2.7546,
+      "step": 3618
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.78132118451025e-06,
+      "loss": 2.4621,
+      "step": 3619
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.76993166287016e-06,
+      "loss": 2.6548,
+      "step": 3620
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.75854214123007e-06,
+      "loss": 1.7758,
+      "step": 3621
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.747152619589978e-06,
+      "loss": 2.4723,
+      "step": 3622
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.735763097949886e-06,
+      "loss": 1.4444,
+      "step": 3623
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.724373576309796e-06,
+      "loss": 1.0214,
+      "step": 3624
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.712984054669704e-06,
+      "loss": 4.1713,
+      "step": 3625
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.701594533029612e-06,
+      "loss": 3.3378,
+      "step": 3626
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.690205011389522e-06,
+      "loss": 4.714,
+      "step": 3627
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.67881548974943e-06,
+      "loss": 2.5944,
+      "step": 3628
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.66742596810934e-06,
+      "loss": 2.6968,
+      "step": 3629
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.656036446469249e-06,
+      "loss": 3.1546,
+      "step": 3630
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.644646924829157e-06,
+      "loss": 2.248,
+      "step": 3631
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.633257403189067e-06,
+      "loss": 3.4698,
+      "step": 3632
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.621867881548975e-06,
+      "loss": 2.6924,
+      "step": 3633
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.610478359908885e-06,
+      "loss": 2.7076,
+      "step": 3634
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.599088838268793e-06,
+      "loss": 2.6233,
+      "step": 3635
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.587699316628701e-06,
+      "loss": 0.9064,
+      "step": 3636
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.576309794988611e-06,
+      "loss": 3.1209,
+      "step": 3637
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.56492027334852e-06,
+      "loss": 2.9194,
+      "step": 3638
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.55353075170843e-06,
+      "loss": 3.1539,
+      "step": 3639
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.542141230068338e-06,
+      "loss": 3.4794,
+      "step": 3640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.530751708428246e-06,
+      "loss": 2.0462,
+      "step": 3641
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.519362186788156e-06,
+      "loss": 3.279,
+      "step": 3642
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.507972665148064e-06,
+      "loss": 2.7793,
+      "step": 3643
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.496583143507974e-06,
+      "loss": 2.7408,
+      "step": 3644
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.485193621867882e-06,
+      "loss": 1.1971,
+      "step": 3645
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.47380410022779e-06,
+      "loss": 3.9858,
+      "step": 3646
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.462414578587698e-06,
+      "loss": 2.9353,
+      "step": 3647
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.451025056947608e-06,
+      "loss": 2.0868,
+      "step": 3648
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.439635535307518e-06,
+      "loss": 2.6369,
+      "step": 3649
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.428246013667426e-06,
+      "loss": 2.5067,
+      "step": 3650
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.416856492027335e-06,
+      "loss": 2.0097,
+      "step": 3651
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.405466970387243e-06,
+      "loss": 1.5464,
+      "step": 3652
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.394077448747153e-06,
+      "loss": 1.7196,
+      "step": 3653
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.382687927107063e-06,
+      "loss": 2.418,
+      "step": 3654
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.37129840546697e-06,
+      "loss": 3.8288,
+      "step": 3655
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.359908883826879e-06,
+      "loss": 2.9433,
+      "step": 3656
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.348519362186787e-06,
+      "loss": 4.3758,
+      "step": 3657
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.337129840546699e-06,
+      "loss": 2.6523,
+      "step": 3658
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.325740318906607e-06,
+      "loss": 3.4866,
+      "step": 3659
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.314350797266515e-06,
+      "loss": 3.8293,
+      "step": 3660
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.302961275626423e-06,
+      "loss": 3.1609,
+      "step": 3661
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.291571753986332e-06,
+      "loss": 2.2963,
+      "step": 3662
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.280182232346243e-06,
+      "loss": 3.3869,
+      "step": 3663
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.268792710706151e-06,
+      "loss": 3.1261,
+      "step": 3664
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.25740318906606e-06,
+      "loss": 3.9254,
+      "step": 3665
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.246013667425968e-06,
+      "loss": 2.3162,
+      "step": 3666
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.234624145785878e-06,
+      "loss": 2.1442,
+      "step": 3667
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.223234624145788e-06,
+      "loss": 4.0085,
+      "step": 3668
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.211845102505696e-06,
+      "loss": 3.8245,
+      "step": 3669
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.200455580865604e-06,
+      "loss": 3.2582,
+      "step": 3670
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.189066059225512e-06,
+      "loss": 2.8821,
+      "step": 3671
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.177676537585422e-06,
+      "loss": 2.0483,
+      "step": 3672
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.16628701594533e-06,
+      "loss": 2.8696,
+      "step": 3673
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.15489749430524e-06,
+      "loss": 3.2373,
+      "step": 3674
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.143507972665148e-06,
+      "loss": 3.7718,
+      "step": 3675
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.132118451025057e-06,
+      "loss": 3.1826,
+      "step": 3676
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.120728929384967e-06,
+      "loss": 2.84,
+      "step": 3677
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.109339407744875e-06,
+      "loss": 2.5698,
+      "step": 3678
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.097949886104785e-06,
+      "loss": 3.6892,
+      "step": 3679
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.086560364464693e-06,
+      "loss": 3.5801,
+      "step": 3680
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.075170842824601e-06,
+      "loss": 3.1904,
+      "step": 3681
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.063781321184511e-06,
+      "loss": 1.457,
+      "step": 3682
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.05239179954442e-06,
+      "loss": 3.0176,
+      "step": 3683
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.041002277904329e-06,
+      "loss": 4.4102,
+      "step": 3684
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.029612756264237e-06,
+      "loss": 2.368,
+      "step": 3685
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.018223234624145e-06,
+      "loss": 4.4434,
+      "step": 3686
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.006833712984055e-06,
+      "loss": 2.8591,
+      "step": 3687
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 7.995444191343964e-06,
+      "loss": 2.862,
+      "step": 3688
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 7.984054669703873e-06,
+      "loss": 2.6485,
+      "step": 3689
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.972665148063782e-06,
+      "loss": 4.4967,
+      "step": 3690
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.96127562642369e-06,
+      "loss": 2.7506,
+      "step": 3691
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.9498861047836e-06,
+      "loss": 3.5409,
+      "step": 3692
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.938496583143508e-06,
+      "loss": 2.2121,
+      "step": 3693
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.927107061503418e-06,
+      "loss": 4.2051,
+      "step": 3694
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.915717539863326e-06,
+      "loss": 3.9536,
+      "step": 3695
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.904328018223234e-06,
+      "loss": 1.5376,
+      "step": 3696
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.892938496583144e-06,
+      "loss": 2.4051,
+      "step": 3697
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.881548974943052e-06,
+      "loss": 4.2625,
+      "step": 3698
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.87015945330296e-06,
+      "loss": 2.0106,
+      "step": 3699
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.85876993166287e-06,
+      "loss": 3.4594,
+      "step": 3700
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.84738041002278e-06,
+      "loss": 2.4044,
+      "step": 3701
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.835990888382689e-06,
+      "loss": 3.059,
+      "step": 3702
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.824601366742597e-06,
+      "loss": 2.2133,
+      "step": 3703
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.813211845102505e-06,
+      "loss": 2.2776,
+      "step": 3704
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.801822323462415e-06,
+      "loss": 4.0714,
+      "step": 3705
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.790432801822325e-06,
+      "loss": 3.2978,
+      "step": 3706
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.779043280182233e-06,
+      "loss": 3.0056,
+      "step": 3707
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.767653758542141e-06,
+      "loss": 2.0501,
+      "step": 3708
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.75626423690205e-06,
+      "loss": 2.9017,
+      "step": 3709
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.74487471526196e-06,
+      "loss": 1.7232,
+      "step": 3710
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.73348519362187e-06,
+      "loss": 2.8013,
+      "step": 3711
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.722095671981777e-06,
+      "loss": 3.9618,
+      "step": 3712
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.710706150341686e-06,
+      "loss": 2.5723,
+      "step": 3713
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.699316628701594e-06,
+      "loss": 1.6322,
+      "step": 3714
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.687927107061504e-06,
+      "loss": 2.5691,
+      "step": 3715
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.676537585421414e-06,
+      "loss": 1.4523,
+      "step": 3716
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.665148063781322e-06,
+      "loss": 3.1713,
+      "step": 3717
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.65375854214123e-06,
+      "loss": 3.7495,
+      "step": 3718
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.642369020501138e-06,
+      "loss": 1.959,
+      "step": 3719
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.630979498861048e-06,
+      "loss": 1.8414,
+      "step": 3720
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.619589977220957e-06,
+      "loss": 3.0516,
+      "step": 3721
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.608200455580866e-06,
+      "loss": 2.0438,
+      "step": 3722
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.5968109339407745e-06,
+      "loss": 1.5647,
+      "step": 3723
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.5854214123006836e-06,
+      "loss": 3.6492,
+      "step": 3724
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.574031890660592e-06,
+      "loss": 1.7082,
+      "step": 3725
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.562642369020502e-06,
+      "loss": 3.2973,
+      "step": 3726
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.551252847380411e-06,
+      "loss": 2.6535,
+      "step": 3727
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.539863325740319e-06,
+      "loss": 2.7195,
+      "step": 3728
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.528473804100228e-06,
+      "loss": 1.3957,
+      "step": 3729
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.517084282460136e-06,
+      "loss": 4.1213,
+      "step": 3730
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.505694760820046e-06,
+      "loss": 3.2215,
+      "step": 3731
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.494305239179955e-06,
+      "loss": 2.5044,
+      "step": 3732
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.482915717539863e-06,
+      "loss": 1.9057,
+      "step": 3733
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.471526195899772e-06,
+      "loss": 2.0805,
+      "step": 3734
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.460136674259681e-06,
+      "loss": 2.3991,
+      "step": 3735
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.448747152619591e-06,
+      "loss": 2.3894,
+      "step": 3736
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.4373576309794996e-06,
+      "loss": 4.2237,
+      "step": 3737
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.425968109339408e-06,
+      "loss": 1.7175,
+      "step": 3738
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.414578587699317e-06,
+      "loss": 2.2061,
+      "step": 3739
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.403189066059225e-06,
+      "loss": 1.2353,
+      "step": 3740
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.391799544419136e-06,
+      "loss": 4.9974,
+      "step": 3741
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.380410022779044e-06,
+      "loss": 2.0135,
+      "step": 3742
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.369020501138953e-06,
+      "loss": 2.6636,
+      "step": 3743
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.357630979498861e-06,
+      "loss": 2.9362,
+      "step": 3744
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.34624145785877e-06,
+      "loss": 2.7473,
+      "step": 3745
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.3348519362186785e-06,
+      "loss": 2.7848,
+      "step": 3746
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.323462414578588e-06,
+      "loss": 1.3723,
+      "step": 3747
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.3120728929384974e-06,
+      "loss": 4.6248,
+      "step": 3748
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.300683371298406e-06,
+      "loss": 4.1979,
+      "step": 3749
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.289293849658315e-06,
+      "loss": 2.3771,
+      "step": 3750
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.277904328018223e-06,
+      "loss": 2.9984,
+      "step": 3751
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.266514806378133e-06,
+      "loss": 4.1165,
+      "step": 3752
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.255125284738042e-06,
+      "loss": 2.6354,
+      "step": 3753
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.24373576309795e-06,
+      "loss": 1.1716,
+      "step": 3754
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.232346241457859e-06,
+      "loss": 1.4915,
+      "step": 3755
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.220956719817767e-06,
+      "loss": 1.7115,
+      "step": 3756
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.209567198177677e-06,
+      "loss": 2.1622,
+      "step": 3757
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.198177676537586e-06,
+      "loss": 1.8595,
+      "step": 3758
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.1867881548974945e-06,
+      "loss": 4.097,
+      "step": 3759
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.1753986332574035e-06,
+      "loss": 1.5386,
+      "step": 3760
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.164009111617312e-06,
+      "loss": 1.605,
+      "step": 3761
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.152619589977222e-06,
+      "loss": 2.7848,
+      "step": 3762
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.141230068337131e-06,
+      "loss": 2.1106,
+      "step": 3763
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.129840546697039e-06,
+      "loss": 3.0203,
+      "step": 3764
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.118451025056948e-06,
+      "loss": 3.0344,
+      "step": 3765
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.107061503416856e-06,
+      "loss": 4.3169,
+      "step": 3766
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.095671981776766e-06,
+      "loss": 4.4794,
+      "step": 3767
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.084282460136675e-06,
+      "loss": 2.9672,
+      "step": 3768
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.072892938496583e-06,
+      "loss": 2.4343,
+      "step": 3769
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.061503416856492e-06,
+      "loss": 1.7275,
+      "step": 3770
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.0501138952164006e-06,
+      "loss": 2.7387,
+      "step": 3771
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.03872437357631e-06,
+      "loss": 2.2217,
+      "step": 3772
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.0273348519362195e-06,
+      "loss": 4.8548,
+      "step": 3773
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.015945330296128e-06,
+      "loss": 2.6517,
+      "step": 3774
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.004555808656037e-06,
+      "loss": 3.6235,
+      "step": 3775
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.993166287015945e-06,
+      "loss": 2.6832,
+      "step": 3776
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.981776765375854e-06,
+      "loss": 3.1014,
+      "step": 3777
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.970387243735764e-06,
+      "loss": 3.4453,
+      "step": 3778
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.958997722095673e-06,
+      "loss": 2.4436,
+      "step": 3779
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.947608200455581e-06,
+      "loss": 1.9919,
+      "step": 3780
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.936218678815489e-06,
+      "loss": 4.2956,
+      "step": 3781
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.9248291571753985e-06,
+      "loss": 2.1616,
+      "step": 3782
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.913439635535308e-06,
+      "loss": 2.6685,
+      "step": 3783
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.902050113895217e-06,
+      "loss": 4.2718,
+      "step": 3784
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.890660592255126e-06,
+      "loss": 4.044,
+      "step": 3785
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.879271070615035e-06,
+      "loss": 2.9188,
+      "step": 3786
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.867881548974943e-06,
+      "loss": 1.1585,
+      "step": 3787
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.856492027334853e-06,
+      "loss": 1.8719,
+      "step": 3788
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.845102505694762e-06,
+      "loss": 0.8228,
+      "step": 3789
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.83371298405467e-06,
+      "loss": 2.867,
+      "step": 3790
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.822323462414579e-06,
+      "loss": 2.7769,
+      "step": 3791
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.810933940774487e-06,
+      "loss": 3.0833,
+      "step": 3792
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.799544419134396e-06,
+      "loss": 2.2601,
+      "step": 3793
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.788154897494306e-06,
+      "loss": 3.1284,
+      "step": 3794
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.7767653758542144e-06,
+      "loss": 3.5313,
+      "step": 3795
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.7653758542141235e-06,
+      "loss": 4.7315,
+      "step": 3796
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.753986332574032e-06,
+      "loss": 1.8397,
+      "step": 3797
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.742596810933941e-06,
+      "loss": 3.7428,
+      "step": 3798
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.731207289293851e-06,
+      "loss": 1.4549,
+      "step": 3799
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.719817767653759e-06,
+      "loss": 3.2776,
+      "step": 3800
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.708428246013668e-06,
+      "loss": 2.5364,
+      "step": 3801
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.697038724373576e-06,
+      "loss": 2.4214,
+      "step": 3802
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.685649202733485e-06,
+      "loss": 2.1769,
+      "step": 3803
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.674259681093395e-06,
+      "loss": 1.2878,
+      "step": 3804
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.662870159453303e-06,
+      "loss": 2.6086,
+      "step": 3805
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.651480637813212e-06,
+      "loss": 3.7367,
+      "step": 3806
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.6400911161731205e-06,
+      "loss": 4.1788,
+      "step": 3807
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.62870159453303e-06,
+      "loss": 3.6217,
+      "step": 3808
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.6173120728929395e-06,
+      "loss": 3.1142,
+      "step": 3809
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.605922551252848e-06,
+      "loss": 4.4792,
+      "step": 3810
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.594533029612757e-06,
+      "loss": 1.9573,
+      "step": 3811
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.583143507972665e-06,
+      "loss": 4.5404,
+      "step": 3812
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.571753986332574e-06,
+      "loss": 2.0426,
+      "step": 3813
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.560364464692484e-06,
+      "loss": 1.2702,
+      "step": 3814
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.548974943052392e-06,
+      "loss": 2.8202,
+      "step": 3815
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.537585421412301e-06,
+      "loss": 2.2516,
+      "step": 3816
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.526195899772209e-06,
+      "loss": 3.7301,
+      "step": 3817
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.5148063781321184e-06,
+      "loss": 1.4664,
+      "step": 3818
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.503416856492027e-06,
+      "loss": 3.1836,
+      "step": 3819
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.492027334851937e-06,
+      "loss": 3.6821,
+      "step": 3820
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.480637813211846e-06,
+      "loss": 3.1033,
+      "step": 3821
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.469248291571755e-06,
+      "loss": 3.9781,
+      "step": 3822
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.457858769931663e-06,
+      "loss": 2.6483,
+      "step": 3823
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.446469248291571e-06,
+      "loss": 3.2832,
+      "step": 3824
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.435079726651482e-06,
+      "loss": 5.2922,
+      "step": 3825
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.42369020501139e-06,
+      "loss": 2.1662,
+      "step": 3826
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.412300683371299e-06,
+      "loss": 1.9987,
+      "step": 3827
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.400911161731207e-06,
+      "loss": 3.3791,
+      "step": 3828
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.389521640091116e-06,
+      "loss": 0.8129,
+      "step": 3829
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.378132118451026e-06,
+      "loss": 4.108,
+      "step": 3830
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.366742596810934e-06,
+      "loss": 1.9095,
+      "step": 3831
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.3553530751708435e-06,
+      "loss": 3.0671,
+      "step": 3832
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.343963553530752e-06,
+      "loss": 3.9178,
+      "step": 3833
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.332574031890661e-06,
+      "loss": 4.7403,
+      "step": 3834
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.321184510250571e-06,
+      "loss": 4.6629,
+      "step": 3835
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.309794988610479e-06,
+      "loss": 2.7364,
+      "step": 3836
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.298405466970388e-06,
+      "loss": 3.0953,
+      "step": 3837
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.287015945330296e-06,
+      "loss": 2.3709,
+      "step": 3838
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.275626423690205e-06,
+      "loss": 3.2119,
+      "step": 3839
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.264236902050115e-06,
+      "loss": 2.385,
+      "step": 3840
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.252847380410023e-06,
+      "loss": 3.0635,
+      "step": 3841
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.241457858769932e-06,
+      "loss": 3.6177,
+      "step": 3842
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.2300683371298405e-06,
+      "loss": 2.4466,
+      "step": 3843
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.21867881548975e-06,
+      "loss": 2.5154,
+      "step": 3844
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.207289293849659e-06,
+      "loss": 2.8114,
+      "step": 3845
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.195899772209568e-06,
+      "loss": 2.0387,
+      "step": 3846
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.184510250569477e-06,
+      "loss": 2.6524,
+      "step": 3847
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.173120728929385e-06,
+      "loss": 3.6145,
+      "step": 3848
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.161731207289294e-06,
+      "loss": 3.3727,
+      "step": 3849
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.150341685649203e-06,
+      "loss": 4.1801,
+      "step": 3850
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.138952164009112e-06,
+      "loss": 1.6868,
+      "step": 3851
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.127562642369021e-06,
+      "loss": 2.4848,
+      "step": 3852
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.116173120728929e-06,
+      "loss": 1.832,
+      "step": 3853
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.104783599088838e-06,
+      "loss": 1.7284,
+      "step": 3854
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.0933940774487474e-06,
+      "loss": 2.799,
+      "step": 3855
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.0820045558086565e-06,
+      "loss": 3.0289,
+      "step": 3856
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.0706150341685656e-06,
+      "loss": 2.1269,
+      "step": 3857
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.059225512528474e-06,
+      "loss": 3.254,
+      "step": 3858
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.047835990888383e-06,
+      "loss": 3.7651,
+      "step": 3859
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.036446469248292e-06,
+      "loss": 2.786,
+      "step": 3860
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.025056947608201e-06,
+      "loss": 1.8135,
+      "step": 3861
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.013667425968109e-06,
+      "loss": 3.7165,
+      "step": 3862
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.002277904328019e-06,
+      "loss": 2.5448,
+      "step": 3863
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.990888382687927e-06,
+      "loss": 4.1036,
+      "step": 3864
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.979498861047836e-06,
+      "loss": 1.8897,
+      "step": 3865
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.968109339407745e-06,
+      "loss": 2.2736,
+      "step": 3866
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9567198177676535e-06,
+      "loss": 1.8164,
+      "step": 3867
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9453302961275634e-06,
+      "loss": 3.2229,
+      "step": 3868
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.933940774487472e-06,
+      "loss": 2.1165,
+      "step": 3869
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.922551252847381e-06,
+      "loss": 1.7279,
+      "step": 3870
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.91116173120729e-06,
+      "loss": 2.5661,
+      "step": 3871
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.899772209567198e-06,
+      "loss": 2.2476,
+      "step": 3872
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.888382687927108e-06,
+      "loss": 2.5052,
+      "step": 3873
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.876993166287016e-06,
+      "loss": 3.644,
+      "step": 3874
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.865603644646925e-06,
+      "loss": 3.3664,
+      "step": 3875
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.854214123006834e-06,
+      "loss": 4.1445,
+      "step": 3876
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.842824601366742e-06,
+      "loss": 2.4038,
+      "step": 3877
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.831435079726652e-06,
+      "loss": 3.8708,
+      "step": 3878
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.8200455580865605e-06,
+      "loss": 1.3058,
+      "step": 3879
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.8086560364464695e-06,
+      "loss": 3.5952,
+      "step": 3880
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.797266514806379e-06,
+      "loss": 2.8244,
+      "step": 3881
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.785876993166287e-06,
+      "loss": 2.0345,
+      "step": 3882
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.774487471526197e-06,
+      "loss": 4.2203,
+      "step": 3883
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.763097949886105e-06,
+      "loss": 2.6414,
+      "step": 3884
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.751708428246014e-06,
+      "loss": 1.6332,
+      "step": 3885
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.740318906605923e-06,
+      "loss": 2.2773,
+      "step": 3886
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.728929384965832e-06,
+      "loss": 2.6544,
+      "step": 3887
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.71753986332574e-06,
+      "loss": 1.9221,
+      "step": 3888
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.706150341685649e-06,
+      "loss": 3.2166,
+      "step": 3889
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.694760820045558e-06,
+      "loss": 2.711,
+      "step": 3890
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.683371298405467e-06,
+      "loss": 1.8745,
+      "step": 3891
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.6719817767653765e-06,
+      "loss": 3.643,
+      "step": 3892
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.660592255125285e-06,
+      "loss": 2.8359,
+      "step": 3893
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.649202733485194e-06,
+      "loss": 3.2577,
+      "step": 3894
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.637813211845103e-06,
+      "loss": 2.9272,
+      "step": 3895
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.626423690205012e-06,
+      "loss": 3.8776,
+      "step": 3896
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.615034168564921e-06,
+      "loss": 2.1065,
+      "step": 3897
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.603644646924829e-06,
+      "loss": 2.1229,
+      "step": 3898
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.592255125284738e-06,
+      "loss": 2.8322,
+      "step": 3899
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.580865603644647e-06,
+      "loss": 3.8986,
+      "step": 3900
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.569476082004555e-06,
+      "loss": 2.8733,
+      "step": 3901
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.558086560364465e-06,
+      "loss": 2.624,
+      "step": 3902
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.5466970387243735e-06,
+      "loss": 1.6171,
+      "step": 3903
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.535307517084283e-06,
+      "loss": 1.2658,
+      "step": 3904
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.523917995444192e-06,
+      "loss": 2.8501,
+      "step": 3905
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.512528473804101e-06,
+      "loss": 2.8411,
+      "step": 3906
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.50113895216401e-06,
+      "loss": 2.8407,
+      "step": 3907
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.489749430523918e-06,
+      "loss": 2.9154,
+      "step": 3908
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.478359908883827e-06,
+      "loss": 3.34,
+      "step": 3909
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.466970387243736e-06,
+      "loss": 3.5427,
+      "step": 3910
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.455580865603645e-06,
+      "loss": 3.8142,
+      "step": 3911
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.444191343963554e-06,
+      "loss": 3.6374,
+      "step": 3912
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.432801822323462e-06,
+      "loss": 2.1306,
+      "step": 3913
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.421412300683371e-06,
+      "loss": 3.4901,
+      "step": 3914
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.4100227790432804e-06,
+      "loss": 2.4955,
+      "step": 3915
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.3986332574031895e-06,
+      "loss": 2.8929,
+      "step": 3916
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.3872437357630985e-06,
+      "loss": 0.982,
+      "step": 3917
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.375854214123007e-06,
+      "loss": 3.9745,
+      "step": 3918
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.364464692482916e-06,
+      "loss": 5.7595,
+      "step": 3919
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.353075170842825e-06,
+      "loss": 2.9188,
+      "step": 3920
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.341685649202734e-06,
+      "loss": 2.587,
+      "step": 3921
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.330296127562642e-06,
+      "loss": 2.4235,
+      "step": 3922
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.318906605922552e-06,
+      "loss": 1.7721,
+      "step": 3923
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.30751708428246e-06,
+      "loss": 3.721,
+      "step": 3924
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.296127562642369e-06,
+      "loss": 3.3653,
+      "step": 3925
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.284738041002278e-06,
+      "loss": 2.5387,
+      "step": 3926
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2733485193621865e-06,
+      "loss": 1.7016,
+      "step": 3927
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2619589977220964e-06,
+      "loss": 4.2017,
+      "step": 3928
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.250569476082005e-06,
+      "loss": 2.7381,
+      "step": 3929
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.239179954441914e-06,
+      "loss": 1.2712,
+      "step": 3930
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.227790432801823e-06,
+      "loss": 5.3078,
+      "step": 3931
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.216400911161731e-06,
+      "loss": 3.6997,
+      "step": 3932
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.205011389521641e-06,
+      "loss": 1.9819,
+      "step": 3933
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.193621867881549e-06,
+      "loss": 2.1012,
+      "step": 3934
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.182232346241458e-06,
+      "loss": 3.9829,
+      "step": 3935
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.170842824601367e-06,
+      "loss": 1.926,
+      "step": 3936
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.159453302961275e-06,
+      "loss": 1.5612,
+      "step": 3937
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.148063781321185e-06,
+      "loss": 2.3115,
+      "step": 3938
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.1366742596810935e-06,
+      "loss": 3.1799,
+      "step": 3939
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.1252847380410025e-06,
+      "loss": 3.1186,
+      "step": 3940
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.1138952164009116e-06,
+      "loss": 1.7677,
+      "step": 3941
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.10250569476082e-06,
+      "loss": 3.3657,
+      "step": 3942
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.09111617312073e-06,
+      "loss": 1.7963,
+      "step": 3943
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.079726651480638e-06,
+      "loss": 4.6658,
+      "step": 3944
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.068337129840547e-06,
+      "loss": 4.2724,
+      "step": 3945
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.056947608200456e-06,
+      "loss": 2.8562,
+      "step": 3946
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.045558086560365e-06,
+      "loss": 2.3867,
+      "step": 3947
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.034168564920273e-06,
+      "loss": 3.8212,
+      "step": 3948
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.022779043280182e-06,
+      "loss": 1.7755,
+      "step": 3949
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.011389521640091e-06,
+      "loss": 1.8629,
+      "step": 3950
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5e-06,
+      "loss": 2.2698,
+      "step": 3951
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.9886104783599095e-06,
+      "loss": 2.8359,
+      "step": 3952
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.977220956719818e-06,
+      "loss": 1.9824,
+      "step": 3953
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.965831435079727e-06,
+      "loss": 3.2601,
+      "step": 3954
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.954441913439636e-06,
+      "loss": 3.9076,
+      "step": 3955
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.943052391799545e-06,
+      "loss": 3.5578,
+      "step": 3956
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.931662870159454e-06,
+      "loss": 2.0512,
+      "step": 3957
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.920273348519362e-06,
+      "loss": 1.9631,
+      "step": 3958
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.908883826879271e-06,
+      "loss": 1.9619,
+      "step": 3959
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.89749430523918e-06,
+      "loss": 2.4924,
+      "step": 3960
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.886104783599088e-06,
+      "loss": 3.0433,
+      "step": 3961
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.874715261958998e-06,
+      "loss": 3.1123,
+      "step": 3962
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.8633257403189065e-06,
+      "loss": 3.0032,
+      "step": 3963
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.851936218678816e-06,
+      "loss": 1.657,
+      "step": 3964
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.840546697038725e-06,
+      "loss": 3.6754,
+      "step": 3965
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.829157175398634e-06,
+      "loss": 2.0528,
+      "step": 3966
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.817767653758543e-06,
+      "loss": 3.3224,
+      "step": 3967
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.806378132118451e-06,
+      "loss": 1.4369,
+      "step": 3968
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.794988610478361e-06,
+      "loss": 2.7068,
+      "step": 3969
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.783599088838269e-06,
+      "loss": 4.077,
+      "step": 3970
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.772209567198178e-06,
+      "loss": 1.1095,
+      "step": 3971
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.760820045558087e-06,
+      "loss": 1.9696,
+      "step": 3972
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.749430523917995e-06,
+      "loss": 2.7754,
+      "step": 3973
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.738041002277904e-06,
+      "loss": 1.6399,
+      "step": 3974
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.7266514806378134e-06,
+      "loss": 2.6345,
+      "step": 3975
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.7152619589977225e-06,
+      "loss": 3.2493,
+      "step": 3976
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.7038724373576315e-06,
+      "loss": 3.7921,
+      "step": 3977
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.69248291571754e-06,
+      "loss": 3.5882,
+      "step": 3978
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.681093394077449e-06,
+      "loss": 3.4359,
+      "step": 3979
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.669703872437358e-06,
+      "loss": 2.7959,
+      "step": 3980
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.658314350797267e-06,
+      "loss": 2.0996,
+      "step": 3981
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.646924829157175e-06,
+      "loss": 3.1002,
+      "step": 3982
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.635535307517085e-06,
+      "loss": 3.0376,
+      "step": 3983
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.624145785876993e-06,
+      "loss": 2.9241,
+      "step": 3984
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.612756264236902e-06,
+      "loss": 4.8721,
+      "step": 3985
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.601366742596811e-06,
+      "loss": 3.8612,
+      "step": 3986
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5899772209567195e-06,
+      "loss": 3.7822,
+      "step": 3987
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5785876993166294e-06,
+      "loss": 2.7455,
+      "step": 3988
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.567198177676538e-06,
+      "loss": 3.2109,
+      "step": 3989
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.555808656036447e-06,
+      "loss": 3.1663,
+      "step": 3990
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.544419134396356e-06,
+      "loss": 2.4408,
+      "step": 3991
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.533029612756264e-06,
+      "loss": 3.7674,
+      "step": 3992
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.521640091116174e-06,
+      "loss": 2.303,
+      "step": 3993
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.510250569476082e-06,
+      "loss": 2.3783,
+      "step": 3994
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.498861047835991e-06,
+      "loss": 2.9878,
+      "step": 3995
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.4874715261959e-06,
+      "loss": 3.19,
+      "step": 3996
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.476082004555808e-06,
+      "loss": 2.3281,
+      "step": 3997
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.464692482915718e-06,
+      "loss": 3.3754,
+      "step": 3998
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.4533029612756265e-06,
+      "loss": 2.6035,
+      "step": 3999
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.4419134396355355e-06,
+      "loss": 3.0137,
+      "step": 4000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.4305239179954446e-06,
+      "loss": 3.4405,
+      "step": 4001
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.419134396355353e-06,
+      "loss": 3.1801,
+      "step": 4002
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.407744874715263e-06,
+      "loss": 3.2459,
+      "step": 4003
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.396355353075171e-06,
+      "loss": 2.1256,
+      "step": 4004
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.38496583143508e-06,
+      "loss": 3.7366,
+      "step": 4005
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.373576309794989e-06,
+      "loss": 2.0286,
+      "step": 4006
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.362186788154898e-06,
+      "loss": 1.6524,
+      "step": 4007
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.350797266514806e-06,
+      "loss": 2.6632,
+      "step": 4008
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.339407744874715e-06,
+      "loss": 2.2429,
+      "step": 4009
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.328018223234624e-06,
+      "loss": 3.1723,
+      "step": 4010
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.316628701594533e-06,
+      "loss": 2.4762,
+      "step": 4011
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.3052391799544425e-06,
+      "loss": 1.9836,
+      "step": 4012
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.293849658314351e-06,
+      "loss": 3.795,
+      "step": 4013
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.28246013667426e-06,
+      "loss": 4.6828,
+      "step": 4014
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.271070615034169e-06,
+      "loss": 3.0646,
+      "step": 4015
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.259681093394078e-06,
+      "loss": 3.733,
+      "step": 4016
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.248291571753987e-06,
+      "loss": 2.2224,
+      "step": 4017
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.236902050113895e-06,
+      "loss": 3.0925,
+      "step": 4018
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.225512528473804e-06,
+      "loss": 3.1407,
+      "step": 4019
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.214123006833713e-06,
+      "loss": 2.313,
+      "step": 4020
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.202733485193621e-06,
+      "loss": 2.7047,
+      "step": 4021
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.191343963553531e-06,
+      "loss": 3.4231,
+      "step": 4022
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.1799544419134395e-06,
+      "loss": 1.3998,
+      "step": 4023
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.168564920273349e-06,
+      "loss": 2.4226,
+      "step": 4024
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.157175398633258e-06,
+      "loss": 2.8636,
+      "step": 4025
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.145785876993166e-06,
+      "loss": 1.6903,
+      "step": 4026
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.134396355353076e-06,
+      "loss": 3.146,
+      "step": 4027
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.123006833712984e-06,
+      "loss": 3.0669,
+      "step": 4028
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.111617312072894e-06,
+      "loss": 2.293,
+      "step": 4029
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.100227790432802e-06,
+      "loss": 1.7951,
+      "step": 4030
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.088838268792711e-06,
+      "loss": 3.2133,
+      "step": 4031
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.07744874715262e-06,
+      "loss": 3.144,
+      "step": 4032
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.066059225512528e-06,
+      "loss": 2.8026,
+      "step": 4033
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.054669703872437e-06,
+      "loss": 3.6565,
+      "step": 4034
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.0432801822323464e-06,
+      "loss": 0.5513,
+      "step": 4035
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.0318906605922555e-06,
+      "loss": 1.9905,
+      "step": 4036
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.0205011389521645e-06,
+      "loss": 2.5828,
+      "step": 4037
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.009111617312073e-06,
+      "loss": 1.5537,
+      "step": 4038
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.997722095671982e-06,
+      "loss": 4.9297,
+      "step": 4039
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.986332574031891e-06,
+      "loss": 3.4859,
+      "step": 4040
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.9749430523918e-06,
+      "loss": 2.3081,
+      "step": 4041
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.963553530751709e-06,
+      "loss": 3.5038,
+      "step": 4042
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.952164009111617e-06,
+      "loss": 3.0361,
+      "step": 4043
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.940774487471526e-06,
+      "loss": 3.2112,
+      "step": 4044
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.929384965831435e-06,
+      "loss": 3.7689,
+      "step": 4045
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.917995444191344e-06,
+      "loss": 1.3644,
+      "step": 4046
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.9066059225512525e-06,
+      "loss": 2.8666,
+      "step": 4047
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.8952164009111624e-06,
+      "loss": 3.6302,
+      "step": 4048
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.883826879271071e-06,
+      "loss": 2.4034,
+      "step": 4049
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.87243735763098e-06,
+      "loss": 3.1616,
+      "step": 4050
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.861047835990889e-06,
+      "loss": 2.1538,
+      "step": 4051
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.849658314350797e-06,
+      "loss": 1.8894,
+      "step": 4052
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.838268792710707e-06,
+      "loss": 1.6947,
+      "step": 4053
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.826879271070615e-06,
+      "loss": 0.7937,
+      "step": 4054
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.815489749430524e-06,
+      "loss": 2.8138,
+      "step": 4055
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.804100227790433e-06,
+      "loss": 3.112,
+      "step": 4056
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.7927107061503418e-06,
+      "loss": 2.8086,
+      "step": 4057
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.781321184510251e-06,
+      "loss": 1.9241,
+      "step": 4058
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.7699316628701595e-06,
+      "loss": 3.0276,
+      "step": 4059
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.758542141230068e-06,
+      "loss": 2.2349,
+      "step": 4060
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.7471526195899776e-06,
+      "loss": 2.3578,
+      "step": 4061
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.735763097949886e-06,
+      "loss": 2.728,
+      "step": 4062
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.7243735763097957e-06,
+      "loss": 1.9199,
+      "step": 4063
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.712984054669704e-06,
+      "loss": 2.7605,
+      "step": 4064
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.7015945330296125e-06,
+      "loss": 3.1785,
+      "step": 4065
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.690205011389522e-06,
+      "loss": 3.6496,
+      "step": 4066
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.6788154897494306e-06,
+      "loss": 1.8307,
+      "step": 4067
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6674259681093392e-06,
+      "loss": 2.6143,
+      "step": 4068
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6560364464692487e-06,
+      "loss": 3.8415,
+      "step": 4069
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6446469248291574e-06,
+      "loss": 1.2838,
+      "step": 4070
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6332574031890664e-06,
+      "loss": 2.5617,
+      "step": 4071
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.621867881548975e-06,
+      "loss": 2.7529,
+      "step": 4072
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6104783599088837e-06,
+      "loss": 1.4644,
+      "step": 4073
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.599088838268793e-06,
+      "loss": 2.2604,
+      "step": 4074
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.5876993166287018e-06,
+      "loss": 2.2165,
+      "step": 4075
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.576309794988611e-06,
+      "loss": 3.2851,
+      "step": 4076
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.5649202733485194e-06,
+      "loss": 2.3616,
+      "step": 4077
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.553530751708428e-06,
+      "loss": 2.6333,
+      "step": 4078
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.5421412300683376e-06,
+      "loss": 5.3412,
+      "step": 4079
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.530751708428246e-06,
+      "loss": 2.1651,
+      "step": 4080
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.519362186788155e-06,
+      "loss": 3.2542,
+      "step": 4081
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.507972665148064e-06,
+      "loss": 1.8553,
+      "step": 4082
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4965831435079725e-06,
+      "loss": 4.2835,
+      "step": 4083
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.485193621867882e-06,
+      "loss": 1.1565,
+      "step": 4084
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4738041002277906e-06,
+      "loss": 1.876,
+      "step": 4085
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4624145785876992e-06,
+      "loss": 2.9843,
+      "step": 4086
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4510250569476087e-06,
+      "loss": 2.7942,
+      "step": 4087
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4396355353075173e-06,
+      "loss": 2.3496,
+      "step": 4088
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4282460136674264e-06,
+      "loss": 2.2097,
+      "step": 4089
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.416856492027335e-06,
+      "loss": 3.0735,
+      "step": 4090
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4054669703872436e-06,
+      "loss": 3.099,
+      "step": 4091
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.394077448747153e-06,
+      "loss": 2.1831,
+      "step": 4092
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.3826879271070618e-06,
+      "loss": 3.4382,
+      "step": 4093
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.3712984054669704e-06,
+      "loss": 2.3859,
+      "step": 4094
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.3599088838268794e-06,
+      "loss": 3.151,
+      "step": 4095
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.348519362186788e-06,
+      "loss": 3.3021,
+      "step": 4096
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.3371298405466975e-06,
+      "loss": 3.7499,
+      "step": 4097
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.325740318906606e-06,
+      "loss": 2.5933,
+      "step": 4098
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.314350797266515e-06,
+      "loss": 4.4466,
+      "step": 4099
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.302961275626424e-06,
+      "loss": 2.7607,
+      "step": 4100
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.2915717539863325e-06,
+      "loss": 3.6197,
+      "step": 4101
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.280182232346242e-06,
+      "loss": 3.7392,
+      "step": 4102
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2687927107061506e-06,
+      "loss": 1.4053,
+      "step": 4103
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2574031890660592e-06,
+      "loss": 2.5039,
+      "step": 4104
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2460136674259687e-06,
+      "loss": 1.6102,
+      "step": 4105
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2346241457858773e-06,
+      "loss": 1.2294,
+      "step": 4106
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2232346241457855e-06,
+      "loss": 2.0802,
+      "step": 4107
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.211845102505695e-06,
+      "loss": 2.9158,
+      "step": 4108
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2004555808656036e-06,
+      "loss": 2.1256,
+      "step": 4109
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.189066059225513e-06,
+      "loss": 1.7126,
+      "step": 4110
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.1776765375854217e-06,
+      "loss": 2.5621,
+      "step": 4111
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1662870159453304e-06,
+      "loss": 2.9666,
+      "step": 4112
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1548974943052394e-06,
+      "loss": 3.261,
+      "step": 4113
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.143507972665148e-06,
+      "loss": 3.6821,
+      "step": 4114
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1321184510250575e-06,
+      "loss": 2.5871,
+      "step": 4115
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.120728929384966e-06,
+      "loss": 1.2435,
+      "step": 4116
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.109339407744875e-06,
+      "loss": 2.181,
+      "step": 4117
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.097949886104784e-06,
+      "loss": 2.7812,
+      "step": 4118
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.0865603644646925e-06,
+      "loss": 4.295,
+      "step": 4119
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.0751708428246015e-06,
+      "loss": 2.9569,
+      "step": 4120
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0637813211845106e-06,
+      "loss": 2.5986,
+      "step": 4121
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.052391799544419e-06,
+      "loss": 5.6403,
+      "step": 4122
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0410022779043282e-06,
+      "loss": 2.5646,
+      "step": 4123
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.029612756264237e-06,
+      "loss": 1.713,
+      "step": 4124
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.018223234624146e-06,
+      "loss": 3.5885,
+      "step": 4125
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.0068337129840546e-06,
+      "loss": 3.5023,
+      "step": 4126
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.9954441913439636e-06,
+      "loss": 3.811,
+      "step": 4127
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.9840546697038727e-06,
+      "loss": 2.9135,
+      "step": 4128
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.9726651480637817e-06,
+      "loss": 1.913,
+      "step": 4129
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.9612756264236903e-06,
+      "loss": 2.8536,
+      "step": 4130
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.949886104783599e-06,
+      "loss": 3.653,
+      "step": 4131
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.938496583143508e-06,
+      "loss": 1.9044,
+      "step": 4132
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.927107061503417e-06,
+      "loss": 4.0162,
+      "step": 4133
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.915717539863326e-06,
+      "loss": 2.3357,
+      "step": 4134
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.9043280182232348e-06,
+      "loss": 2.4934,
+      "step": 4135
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8929384965831434e-06,
+      "loss": 3.9347,
+      "step": 4136
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8815489749430524e-06,
+      "loss": 3.7257,
+      "step": 4137
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8701594533029615e-06,
+      "loss": 2.6763,
+      "step": 4138
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.85876993166287e-06,
+      "loss": 2.8249,
+      "step": 4139
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.847380410022779e-06,
+      "loss": 2.7748,
+      "step": 4140
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8359908883826882e-06,
+      "loss": 3.355,
+      "step": 4141
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.824601366742597e-06,
+      "loss": 2.8943,
+      "step": 4142
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.813211845102506e-06,
+      "loss": 4.4149,
+      "step": 4143
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8018223234624145e-06,
+      "loss": 2.5323,
+      "step": 4144
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.7904328018223236e-06,
+      "loss": 4.2345,
+      "step": 4145
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.7790432801822326e-06,
+      "loss": 2.2006,
+      "step": 4146
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.7676537585421417e-06,
+      "loss": 3.091,
+      "step": 4147
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.7562642369020503e-06,
+      "loss": 3.9505,
+      "step": 4148
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.744874715261959e-06,
+      "loss": 3.0677,
+      "step": 4149
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.733485193621868e-06,
+      "loss": 4.2254,
+      "step": 4150
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.722095671981777e-06,
+      "loss": 2.4756,
+      "step": 4151
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.7107061503416857e-06,
+      "loss": 1.6725,
+      "step": 4152
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6993166287015947e-06,
+      "loss": 2.6347,
+      "step": 4153
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6879271070615034e-06,
+      "loss": 1.7884,
+      "step": 4154
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6765375854214124e-06,
+      "loss": 3.8587,
+      "step": 4155
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.665148063781321e-06,
+      "loss": 2.1275,
+      "step": 4156
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.65375854214123e-06,
+      "loss": 1.5763,
+      "step": 4157
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.642369020501139e-06,
+      "loss": 2.0288,
+      "step": 4158
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6309794988610482e-06,
+      "loss": 1.9275,
+      "step": 4159
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.619589977220957e-06,
+      "loss": 2.8733,
+      "step": 4160
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6082004555808655e-06,
+      "loss": 1.3586,
+      "step": 4161
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.5968109339407745e-06,
+      "loss": 3.0881,
+      "step": 4162
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.5854214123006836e-06,
+      "loss": 3.7639,
+      "step": 4163
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.5740318906605926e-06,
+      "loss": 3.3061,
+      "step": 4164
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.5626423690205013e-06,
+      "loss": 2.7896,
+      "step": 4165
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.55125284738041e-06,
+      "loss": 2.4921,
+      "step": 4166
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.539863325740319e-06,
+      "loss": 2.5565,
+      "step": 4167
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.528473804100228e-06,
+      "loss": 1.0851,
+      "step": 4168
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.5170842824601366e-06,
+      "loss": 3.2264,
+      "step": 4169
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.5056947608200457e-06,
+      "loss": 3.239,
+      "step": 4170
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.4943052391799547e-06,
+      "loss": 2.662,
+      "step": 4171
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.4829157175398634e-06,
+      "loss": 4.8395,
+      "step": 4172
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.4715261958997724e-06,
+      "loss": 2.2737,
+      "step": 4173
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.460136674259681e-06,
+      "loss": 3.6158,
+      "step": 4174
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.44874715261959e-06,
+      "loss": 3.4485,
+      "step": 4175
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.437357630979499e-06,
+      "loss": 2.4193,
+      "step": 4176
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.425968109339408e-06,
+      "loss": 2.1871,
+      "step": 4177
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.414578587699317e-06,
+      "loss": 2.9182,
+      "step": 4178
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4031890660592255e-06,
+      "loss": 3.0201,
+      "step": 4179
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.3917995444191345e-06,
+      "loss": 2.5399,
+      "step": 4180
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.3804100227790436e-06,
+      "loss": 1.3748,
+      "step": 4181
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.369020501138952e-06,
+      "loss": 2.8851,
+      "step": 4182
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.3576309794988612e-06,
+      "loss": 4.0235,
+      "step": 4183
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.34624145785877e-06,
+      "loss": 2.7913,
+      "step": 4184
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.334851936218679e-06,
+      "loss": 2.2312,
+      "step": 4185
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.3234624145785876e-06,
+      "loss": 3.1382,
+      "step": 4186
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.3120728929384966e-06,
+      "loss": 2.7114,
+      "step": 4187
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.3006833712984057e-06,
+      "loss": 3.1657,
+      "step": 4188
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.2892938496583147e-06,
+      "loss": 1.1818,
+      "step": 4189
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.2779043280182233e-06,
+      "loss": 2.4375,
+      "step": 4190
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.266514806378132e-06,
+      "loss": 2.2039,
+      "step": 4191
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.255125284738041e-06,
+      "loss": 2.1475,
+      "step": 4192
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.24373576309795e-06,
+      "loss": 2.0388,
+      "step": 4193
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.232346241457859e-06,
+      "loss": 1.8244,
+      "step": 4194
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2209567198177678e-06,
+      "loss": 1.5431,
+      "step": 4195
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2095671981776764e-06,
+      "loss": 3.1759,
+      "step": 4196
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.1981776765375854e-06,
+      "loss": 3.1867,
+      "step": 4197
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.1867881548974945e-06,
+      "loss": 3.295,
+      "step": 4198
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.175398633257403e-06,
+      "loss": 4.1907,
+      "step": 4199
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.164009111617312e-06,
+      "loss": 1.866,
+      "step": 4200
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1526195899772212e-06,
+      "loss": 2.2811,
+      "step": 4201
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.14123006833713e-06,
+      "loss": 3.0037,
+      "step": 4202
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.129840546697039e-06,
+      "loss": 0.861,
+      "step": 4203
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1184510250569475e-06,
+      "loss": 2.5823,
+      "step": 4204
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1070615034168566e-06,
+      "loss": 2.4574,
+      "step": 4205
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.0956719817767656e-06,
+      "loss": 1.8622,
+      "step": 4206
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.0842824601366747e-06,
+      "loss": 2.032,
+      "step": 4207
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.072892938496583e-06,
+      "loss": 3.4848,
+      "step": 4208
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.061503416856492e-06,
+      "loss": 3.3984,
+      "step": 4209
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.050113895216401e-06,
+      "loss": 2.5084,
+      "step": 4210
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.03872437357631e-06,
+      "loss": 2.1374,
+      "step": 4211
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0273348519362187e-06,
+      "loss": 2.5441,
+      "step": 4212
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0159453302961277e-06,
+      "loss": 2.299,
+      "step": 4213
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0045558086560364e-06,
+      "loss": 1.3423,
+      "step": 4214
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.9931662870159454e-06,
+      "loss": 1.3905,
+      "step": 4215
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.9817767653758545e-06,
+      "loss": 2.1281,
+      "step": 4216
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.970387243735763e-06,
+      "loss": 2.1751,
+      "step": 4217
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.958997722095672e-06,
+      "loss": 3.4652,
+      "step": 4218
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.9476082004555812e-06,
+      "loss": 1.1185,
+      "step": 4219
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.93621867881549e-06,
+      "loss": 2.272,
+      "step": 4220
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9248291571753985e-06,
+      "loss": 3.9905,
+      "step": 4221
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9134396355353075e-06,
+      "loss": 2.0433,
+      "step": 4222
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9020501138952166e-06,
+      "loss": 3.3928,
+      "step": 4223
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.8906605922551254e-06,
+      "loss": 3.9195,
+      "step": 4224
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.879271070615034e-06,
+      "loss": 2.4448,
+      "step": 4225
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.867881548974943e-06,
+      "loss": 4.0929,
+      "step": 4226
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.856492027334852e-06,
+      "loss": 0.9881,
+      "step": 4227
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.845102505694761e-06,
+      "loss": 2.6893,
+      "step": 4228
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8337129840546696e-06,
+      "loss": 1.993,
+      "step": 4229
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8223234624145787e-06,
+      "loss": 3.1291,
+      "step": 4230
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8109339407744875e-06,
+      "loss": 2.9105,
+      "step": 4231
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7995444191343966e-06,
+      "loss": 1.0948,
+      "step": 4232
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7881548974943054e-06,
+      "loss": 3.0701,
+      "step": 4233
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.776765375854214e-06,
+      "loss": 2.3111,
+      "step": 4234
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.765375854214123e-06,
+      "loss": 1.5469,
+      "step": 4235
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.753986332574032e-06,
+      "loss": 3.6511,
+      "step": 4236
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.742596810933941e-06,
+      "loss": 3.2083,
+      "step": 4237
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7312072892938496e-06,
+      "loss": 3.7687,
+      "step": 4238
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7198177676537587e-06,
+      "loss": 3.9792,
+      "step": 4239
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7084282460136675e-06,
+      "loss": 2.7323,
+      "step": 4240
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.6970387243735766e-06,
+      "loss": 3.0827,
+      "step": 4241
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.6856492027334852e-06,
+      "loss": 2.8422,
+      "step": 4242
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.674259681093394e-06,
+      "loss": 2.9205,
+      "step": 4243
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.662870159453303e-06,
+      "loss": 2.6573,
+      "step": 4244
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.651480637813212e-06,
+      "loss": 0.9591,
+      "step": 4245
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.640091116173121e-06,
+      "loss": 2.4641,
+      "step": 4246
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.6287015945330296e-06,
+      "loss": 3.1802,
+      "step": 4247
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6173120728929387e-06,
+      "loss": 3.236,
+      "step": 4248
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6059225512528475e-06,
+      "loss": 2.9679,
+      "step": 4249
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.5945330296127566e-06,
+      "loss": 3.5177,
+      "step": 4250
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.5831435079726652e-06,
+      "loss": 2.9141,
+      "step": 4251
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.571753986332574e-06,
+      "loss": 1.6117,
+      "step": 4252
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.560364464692483e-06,
+      "loss": 1.5639,
+      "step": 4253
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.548974943052392e-06,
+      "loss": 4.4555,
+      "step": 4254
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5375854214123008e-06,
+      "loss": 2.4544,
+      "step": 4255
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5261958997722096e-06,
+      "loss": 3.2889,
+      "step": 4256
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5148063781321184e-06,
+      "loss": 3.9691,
+      "step": 4257
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5034168564920273e-06,
+      "loss": 4.3969,
+      "step": 4258
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.4920273348519363e-06,
+      "loss": 3.0245,
+      "step": 4259
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.4806378132118452e-06,
+      "loss": 2.4398,
+      "step": 4260
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.469248291571754e-06,
+      "loss": 4.6658,
+      "step": 4261
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.457858769931663e-06,
+      "loss": 3.7684,
+      "step": 4262
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4464692482915717e-06,
+      "loss": 1.8385,
+      "step": 4263
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4350797266514807e-06,
+      "loss": 2.1929,
+      "step": 4264
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4236902050113896e-06,
+      "loss": 2.989,
+      "step": 4265
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4123006833712984e-06,
+      "loss": 2.2061,
+      "step": 4266
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4009111617312073e-06,
+      "loss": 1.7827,
+      "step": 4267
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3895216400911163e-06,
+      "loss": 2.1989,
+      "step": 4268
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3781321184510252e-06,
+      "loss": 3.4512,
+      "step": 4269
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.366742596810934e-06,
+      "loss": 3.4809,
+      "step": 4270
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3553530751708428e-06,
+      "loss": 3.3056,
+      "step": 4271
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3439635535307517e-06,
+      "loss": 3.3166,
+      "step": 4272
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3325740318906605e-06,
+      "loss": 2.8599,
+      "step": 4273
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3211845102505696e-06,
+      "loss": 2.794,
+      "step": 4274
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3097949886104784e-06,
+      "loss": 1.5238,
+      "step": 4275
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.2984054669703873e-06,
+      "loss": 3.2195,
+      "step": 4276
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.2870159453302963e-06,
+      "loss": 5.2039,
+      "step": 4277
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.275626423690205e-06,
+      "loss": 2.0842,
+      "step": 4278
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.264236902050114e-06,
+      "loss": 2.2889,
+      "step": 4279
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2528473804100228e-06,
+      "loss": 2.5295,
+      "step": 4280
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2414578587699317e-06,
+      "loss": 3.063,
+      "step": 4281
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2300683371298405e-06,
+      "loss": 2.4605,
+      "step": 4282
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.2186788154897496e-06,
+      "loss": 1.124,
+      "step": 4283
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.2072892938496584e-06,
+      "loss": 3.3901,
+      "step": 4284
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1958997722095673e-06,
+      "loss": 3.0331,
+      "step": 4285
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.184510250569476e-06,
+      "loss": 2.2359,
+      "step": 4286
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.173120728929385e-06,
+      "loss": 1.6958,
+      "step": 4287
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1617312072892938e-06,
+      "loss": 2.2503,
+      "step": 4288
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1503416856492028e-06,
+      "loss": 1.6808,
+      "step": 4289
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1389521640091117e-06,
+      "loss": 2.6206,
+      "step": 4290
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1275626423690205e-06,
+      "loss": 3.2591,
+      "step": 4291
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1161731207289296e-06,
+      "loss": 3.5402,
+      "step": 4292
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1047835990888382e-06,
+      "loss": 3.0397,
+      "step": 4293
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.0933940774487472e-06,
+      "loss": 3.0436,
+      "step": 4294
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.082004555808656e-06,
+      "loss": 2.7244,
+      "step": 4295
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.070615034168565e-06,
+      "loss": 2.767,
+      "step": 4296
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0592255125284738e-06,
+      "loss": 3.9362,
+      "step": 4297
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0478359908883828e-06,
+      "loss": 3.1728,
+      "step": 4298
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0364464692482915e-06,
+      "loss": 1.9952,
+      "step": 4299
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0250569476082005e-06,
+      "loss": 3.0714,
+      "step": 4300
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0136674259681093e-06,
+      "loss": 1.7589,
+      "step": 4301
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0022779043280182e-06,
+      "loss": 2.7401,
+      "step": 4302
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 9.908883826879272e-07,
+      "loss": 4.0704,
+      "step": 4303
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 9.79498861047836e-07,
+      "loss": 2.7462,
+      "step": 4304
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.68109339407745e-07,
+      "loss": 1.1609,
+      "step": 4305
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.567198177676538e-07,
+      "loss": 3.3548,
+      "step": 4306
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.453302961275627e-07,
+      "loss": 2.6131,
+      "step": 4307
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.339407744874715e-07,
+      "loss": 3.1786,
+      "step": 4308
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.225512528473805e-07,
+      "loss": 2.5303,
+      "step": 4309
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.111617312072893e-07,
+      "loss": 2.2962,
+      "step": 4310
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 8.997722095671983e-07,
+      "loss": 3.7358,
+      "step": 4311
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 8.88382687927107e-07,
+      "loss": 2.4132,
+      "step": 4312
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 8.76993166287016e-07,
+      "loss": 2.3306,
+      "step": 4313
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.656036446469248e-07,
+      "loss": 4.2601,
+      "step": 4314
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.542141230068338e-07,
+      "loss": 3.1704,
+      "step": 4315
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.428246013667426e-07,
+      "loss": 1.9767,
+      "step": 4316
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.314350797266515e-07,
+      "loss": 3.7772,
+      "step": 4317
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.200455580865605e-07,
+      "loss": 1.1639,
+      "step": 4318
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.086560364464693e-07,
+      "loss": 1.2937,
+      "step": 4319
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.972665148063783e-07,
+      "loss": 1.4398,
+      "step": 4320
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.85876993166287e-07,
+      "loss": 2.4318,
+      "step": 4321
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.74487471526196e-07,
+      "loss": 3.6337,
+      "step": 4322
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.630979498861048e-07,
+      "loss": 2.2956,
+      "step": 4323
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.517084282460136e-07,
+      "loss": 2.3572,
+      "step": 4324
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.403189066059226e-07,
+      "loss": 2.3695,
+      "step": 4325
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.289293849658315e-07,
+      "loss": 1.8428,
+      "step": 4326
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.175398633257404e-07,
+      "loss": 2.913,
+      "step": 4327
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.061503416856492e-07,
+      "loss": 1.9168,
+      "step": 4328
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.947608200455582e-07,
+      "loss": 2.7317,
+      "step": 4329
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.83371298405467e-07,
+      "loss": 3.904,
+      "step": 4330
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.719817767653758e-07,
+      "loss": 2.4207,
+      "step": 4331
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.605922551252848e-07,
+      "loss": 2.7533,
+      "step": 4332
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.492027334851936e-07,
+      "loss": 2.2713,
+      "step": 4333
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.378132118451025e-07,
+      "loss": 2.0429,
+      "step": 4334
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.264236902050114e-07,
+      "loss": 2.7647,
+      "step": 4335
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.150341685649203e-07,
+      "loss": 2.9488,
+      "step": 4336
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.036446469248292e-07,
+      "loss": 2.5675,
+      "step": 4337
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.92255125284738e-07,
+      "loss": 2.9395,
+      "step": 4338
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.808656036446469e-07,
+      "loss": 2.7006,
+      "step": 4339
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.694760820045558e-07,
+      "loss": 3.2623,
+      "step": 4340
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.580865603644648e-07,
+      "loss": 2.4386,
+      "step": 4341
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.466970387243736e-07,
+      "loss": 3.3298,
+      "step": 4342
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.353075170842825e-07,
+      "loss": 2.8296,
+      "step": 4343
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.239179954441914e-07,
+      "loss": 1.527,
+      "step": 4344
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.125284738041003e-07,
+      "loss": 2.0225,
+      "step": 4345
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.011389521640091e-07,
+      "loss": 3.2317,
+      "step": 4346
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.89749430523918e-07,
+      "loss": 1.7161,
+      "step": 4347
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.783599088838269e-07,
+      "loss": 2.5795,
+      "step": 4348
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.669703872437358e-07,
+      "loss": 4.3288,
+      "step": 4349
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5558086560364467e-07,
+      "loss": 1.1628,
+      "step": 4350
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.441913439635535e-07,
+      "loss": 3.3934,
+      "step": 4351
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.328018223234624e-07,
+      "loss": 1.8901,
+      "step": 4352
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.214123006833713e-07,
+      "loss": 2.2452,
+      "step": 4353
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.1002277904328024e-07,
+      "loss": 2.8747,
+      "step": 4354
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.9863325740318914e-07,
+      "loss": 2.2138,
+      "step": 4355
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.87243735763098e-07,
+      "loss": 1.1285,
+      "step": 4356
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.758542141230068e-07,
+      "loss": 2.7676,
+      "step": 4357
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.6446469248291577e-07,
+      "loss": 1.9197,
+      "step": 4358
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.530751708428246e-07,
+      "loss": 3.5426,
+      "step": 4359
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.416856492027335e-07,
+      "loss": 2.1988,
+      "step": 4360
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.302961275626424e-07,
+      "loss": 1.8547,
+      "step": 4361
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.1890660592255124e-07,
+      "loss": 3.1477,
+      "step": 4362
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.0751708428246013e-07,
+      "loss": 3.493,
+      "step": 4363
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.96127562642369e-07,
+      "loss": 3.333,
+      "step": 4364
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.847380410022779e-07,
+      "loss": 3.0788,
+      "step": 4365
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.733485193621868e-07,
+      "loss": 1.8611,
+      "step": 4366
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.619589977220957e-07,
+      "loss": 2.5282,
+      "step": 4367
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.5056947608200455e-07,
+      "loss": 2.528,
+      "step": 4368
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.3917995444191344e-07,
+      "loss": 2.1447,
+      "step": 4369
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.2779043280182233e-07,
+      "loss": 2.4532,
+      "step": 4370
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.164009111617312e-07,
+      "loss": 2.5029,
+      "step": 4371
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.0501138952164012e-07,
+      "loss": 2.2652,
+      "step": 4372
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.93621867881549e-07,
+      "loss": 3.7848,
+      "step": 4373
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.8223234624145788e-07,
+      "loss": 1.9632,
+      "step": 4374
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.7084282460136675e-07,
+      "loss": 2.3705,
+      "step": 4375
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.5945330296127562e-07,
+      "loss": 3.7508,
+      "step": 4376
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.480637813211845e-07,
+      "loss": 2.5049,
+      "step": 4377
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.366742596810934e-07,
+      "loss": 2.2499,
+      "step": 4378
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2528473804100227e-07,
+      "loss": 4.5044,
+      "step": 4379
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1389521640091117e-07,
+      "loss": 3.3547,
+      "step": 4380
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.0250569476082006e-07,
+      "loss": 2.653,
+      "step": 4381
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.111617312072894e-08,
+      "loss": 1.7794,
+      "step": 4382
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.972665148063781e-08,
+      "loss": 4.054,
+      "step": 4383
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.83371298405467e-08,
+      "loss": 2.7965,
+      "step": 4384
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.6947608200455584e-08,
+      "loss": 3.4097,
+      "step": 4385
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.555808656036447e-08,
+      "loss": 2.7788,
+      "step": 4386
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.416856492027335e-08,
+      "loss": 2.4029,
+      "step": 4387
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.2779043280182235e-08,
+      "loss": 1.7554,
+      "step": 4388
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1389521640091118e-08,
+      "loss": 2.1503,
+      "step": 4389
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 1.4862,
+      "step": 4390
+    },
     {
       "epoch": 10.0,
-      "step": 70,
-      "total_flos": 256066191360000.0,
-      "train_loss": 2.381487710135324,
-      "train_runtime": 18.165,
-      "train_samples_per_second": 26.975,
-      "train_steps_per_second": 3.854
+      "step": 4390,
+      "total_flos": 481667817856500.0,
+      "train_loss": 3.5042898788946366,
+      "train_runtime": 613.1863,
+      "train_samples_per_second": 57.193,
+      "train_steps_per_second": 7.159
     }
   ],
-  "max_steps": 70,
+  "max_steps": 4390,
   "num_train_epochs": 10,
-  "total_flos": 256066191360000.0,
+  "total_flos": 481667817856500.0,
   "trial_name": null,
   "trial_params": null
 }