diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,58525 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9998846198223146,
+  "global_step": 9750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.8412,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4423,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.825938566552902e-08,
+      "loss": 2.1726,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3651877133105803e-07,
+      "loss": 2.6872,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0477815699658704e-07,
+      "loss": 2.1511,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7303754266211607e-07,
+      "loss": 2.6153,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.412969283276451e-07,
+      "loss": 2.3401,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0955631399317407e-07,
+      "loss": 2.5546,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.778156996587031e-07,
+      "loss": 2.4168,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.460750853242321e-07,
+      "loss": 2.6282,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.143344709897612e-07,
+      "loss": 2.2576,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.825938566552902e-07,
+      "loss": 2.443,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.508532423208192e-07,
+      "loss": 2.0398,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.508532423208192e-07,
+      "loss": 3.2916,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.191126279863481e-07,
+      "loss": 2.4862,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.873720136518772e-07,
+      "loss": 3.1544,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.556313993174062e-07,
+      "loss": 2.7106,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0238907849829352e-06,
+      "loss": 2.1121,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0921501706484643e-06,
+      "loss": 2.041,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1604095563139933e-06,
+      "loss": 1.8503,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2286689419795223e-06,
+      "loss": 1.7012,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2969283276450511e-06,
+      "loss": 1.5677,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3651877133105804e-06,
+      "loss": 1.9192,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4334470989761092e-06,
+      "loss": 1.6379,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5017064846416384e-06,
+      "loss": 1.5235,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5699658703071675e-06,
+      "loss": 1.4589,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6382252559726963e-06,
+      "loss": 1.6921,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7064846416382255e-06,
+      "loss": 1.6525,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7747440273037543e-06,
+      "loss": 1.9098,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8430034129692834e-06,
+      "loss": 1.6589,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9112627986348124e-06,
+      "loss": 1.6159,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9795221843003416e-06,
+      "loss": 1.69,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0477815699658705e-06,
+      "loss": 1.8627,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1160409556313997e-06,
+      "loss": 2.037,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1843003412969285e-06,
+      "loss": 1.4806,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2525597269624573e-06,
+      "loss": 1.6636,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3208191126279866e-06,
+      "loss": 1.4777,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3890784982935154e-06,
+      "loss": 1.4782,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4573378839590446e-06,
+      "loss": 1.8474,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5255972696245735e-06,
+      "loss": 1.8317,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5938566552901023e-06,
+      "loss": 1.4948,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.662116040955632e-06,
+      "loss": 1.7582,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7303754266211608e-06,
+      "loss": 1.6929,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7986348122866896e-06,
+      "loss": 1.8638,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8668941979522184e-06,
+      "loss": 1.479,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9351535836177476e-06,
+      "loss": 1.8665,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.003412969283277e-06,
+      "loss": 2.2169,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0716723549488057e-06,
+      "loss": 1.8436,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.139931740614335e-06,
+      "loss": 1.6936,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2081911262798638e-06,
+      "loss": 1.7007,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2764505119453926e-06,
+      "loss": 1.7064,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3447098976109214e-06,
+      "loss": 1.4948,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.412969283276451e-06,
+      "loss": 1.546,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48122866894198e-06,
+      "loss": 1.4594,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5494880546075087e-06,
+      "loss": 1.6538,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.617747440273038e-06,
+      "loss": 1.2752,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6860068259385667e-06,
+      "loss": 1.8364,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7542662116040956e-06,
+      "loss": 1.6131,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.822525597269625e-06,
+      "loss": 1.8046,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.890784982935154e-06,
+      "loss": 1.6027,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.959044368600683e-06,
+      "loss": 2.2379,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.027303754266212e-06,
+      "loss": 1.5847,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.095563139931741e-06,
+      "loss": 1.645,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.163822525597269e-06,
+      "loss": 1.5587,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.232081911262799e-06,
+      "loss": 1.3153,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.300341296928328e-06,
+      "loss": 1.5499,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.368600682593857e-06,
+      "loss": 1.5511,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.436860068259386e-06,
+      "loss": 1.6488,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.505119453924915e-06,
+      "loss": 1.4257,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.573378839590444e-06,
+      "loss": 1.259,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641638225255973e-06,
+      "loss": 1.3778,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.709897610921502e-06,
+      "loss": 1.5892,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.778156996587031e-06,
+      "loss": 1.4805,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.84641638225256e-06,
+      "loss": 1.5089,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914675767918089e-06,
+      "loss": 1.8413,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982935153583618e-06,
+      "loss": 1.7549,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.051194539249147e-06,
+      "loss": 1.5412,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.119453924914676e-06,
+      "loss": 1.6263,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1877133105802046e-06,
+      "loss": 1.4993,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.255972696245735e-06,
+      "loss": 1.7269,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.324232081911264e-06,
+      "loss": 1.6231,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.392491467576792e-06,
+      "loss": 1.5641,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4607508532423215e-06,
+      "loss": 1.7472,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.529010238907851e-06,
+      "loss": 1.4605,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.597269624573379e-06,
+      "loss": 2.2634,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.665529010238908e-06,
+      "loss": 1.7317,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.733788395904437e-06,
+      "loss": 1.612,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.802047781569966e-06,
+      "loss": 1.3336,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.870307167235495e-06,
+      "loss": 1.7836,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.938566552901024e-06,
+      "loss": 1.2789,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.006825938566554e-06,
+      "loss": 1.6814,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.075085324232083e-06,
+      "loss": 1.8984,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.143344709897611e-06,
+      "loss": 1.8542,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.211604095563141e-06,
+      "loss": 1.4444,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.27986348122867e-06,
+      "loss": 1.6082,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.348122866894198e-06,
+      "loss": 1.6632,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4163822525597275e-06,
+      "loss": 1.438,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.484641638225257e-06,
+      "loss": 1.8014,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.552901023890785e-06,
+      "loss": 1.6969,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.621160409556314e-06,
+      "loss": 1.4982,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.689419795221843e-06,
+      "loss": 1.3874,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.757679180887372e-06,
+      "loss": 2.3024,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.825938566552902e-06,
+      "loss": 1.9018,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.894197952218431e-06,
+      "loss": 1.4396,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.96245733788396e-06,
+      "loss": 1.5857,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.030716723549489e-06,
+      "loss": 1.5681,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.098976109215017e-06,
+      "loss": 1.3912,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.167235494880547e-06,
+      "loss": 1.7512,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.235494880546076e-06,
+      "loss": 1.3889,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.303754266211604e-06,
+      "loss": 1.4803,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3720136518771335e-06,
+      "loss": 1.4952,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.440273037542663e-06,
+      "loss": 1.8956,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.508532423208191e-06,
+      "loss": 1.7031,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.57679180887372e-06,
+      "loss": 1.6165,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.64505119453925e-06,
+      "loss": 1.6164,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.713310580204779e-06,
+      "loss": 1.6128,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.781569965870308e-06,
+      "loss": 1.5235,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.849829351535837e-06,
+      "loss": 1.4066,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.918088737201367e-06,
+      "loss": 1.4756,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.986348122866894e-06,
+      "loss": 1.3571,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.054607508532423e-06,
+      "loss": 1.6108,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.122866894197953e-06,
+      "loss": 1.6723,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.191126279863482e-06,
+      "loss": 1.1759,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.259385665529011e-06,
+      "loss": 1.483,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.327645051194539e-06,
+      "loss": 1.75,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.395904436860068e-06,
+      "loss": 1.5272,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.464163822525599e-06,
+      "loss": 2.1243,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.532423208191128e-06,
+      "loss": 1.4895,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.600682593856656e-06,
+      "loss": 1.5599,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.668941979522185e-06,
+      "loss": 2.6381,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.737201365187714e-06,
+      "loss": 1.3859,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.805460750853243e-06,
+      "loss": 1.6652,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.873720136518773e-06,
+      "loss": 1.7627,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.9419795221843e-06,
+      "loss": 1.5259,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.01023890784983e-06,
+      "loss": 1.6265,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.078498293515359e-06,
+      "loss": 1.6605,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.146757679180888e-06,
+      "loss": 1.7787,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.215017064846417e-06,
+      "loss": 1.5744,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.283276450511946e-06,
+      "loss": 1.3078,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.351535836177476e-06,
+      "loss": 1.4925,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.419795221843005e-06,
+      "loss": 2.6046,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.488054607508534e-06,
+      "loss": 1.6995,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.556313993174062e-06,
+      "loss": 1.6,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.62457337883959e-06,
+      "loss": 1.4351,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.69283276450512e-06,
+      "loss": 2.3479,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.76109215017065e-06,
+      "loss": 1.694,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.829351535836179e-06,
+      "loss": 1.4736,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.897610921501706e-06,
+      "loss": 1.4983,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965870307167235e-06,
+      "loss": 1.4491,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0034129692832766e-05,
+      "loss": 1.3217,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0102389078498294e-05,
+      "loss": 1.6762,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0170648464163823e-05,
+      "loss": 1.428,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0238907849829352e-05,
+      "loss": 1.4604,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0307167235494882e-05,
+      "loss": 1.5358,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0375426621160409e-05,
+      "loss": 1.4691,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.044368600682594e-05,
+      "loss": 1.5801,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.051194539249147e-05,
+      "loss": 1.4129,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0580204778156997e-05,
+      "loss": 1.3717,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0648464163822528e-05,
+      "loss": 1.5442,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0716723549488055e-05,
+      "loss": 1.558,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0784982935153585e-05,
+      "loss": 1.4089,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0853242320819112e-05,
+      "loss": 1.8121,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0921501706484643e-05,
+      "loss": 1.7298,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.098976109215017e-05,
+      "loss": 1.5517,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1058020477815702e-05,
+      "loss": 1.9167,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1126279863481229e-05,
+      "loss": 1.5759,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1194539249146758e-05,
+      "loss": 1.3289,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.126279863481229e-05,
+      "loss": 1.3922,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1331058020477817e-05,
+      "loss": 1.6326,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1399317406143346e-05,
+      "loss": 1.6648,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1467576791808874e-05,
+      "loss": 1.6214,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1535836177474405e-05,
+      "loss": 1.5009,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1604095563139932e-05,
+      "loss": 1.6154,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1672354948805463e-05,
+      "loss": 1.2757,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.174061433447099e-05,
+      "loss": 1.7028,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.180887372013652e-05,
+      "loss": 1.7266,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1877133105802047e-05,
+      "loss": 1.3768,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1945392491467578e-05,
+      "loss": 1.34,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2013651877133108e-05,
+      "loss": 1.5778,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2081911262798635e-05,
+      "loss": 1.4724,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2150170648464166e-05,
+      "loss": 1.6266,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2218430034129694e-05,
+      "loss": 1.3219,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2286689419795223e-05,
+      "loss": 1.343,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2354948805460752e-05,
+      "loss": 1.5996,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2423208191126281e-05,
+      "loss": 1.4521,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2491467576791809e-05,
+      "loss": 1.7369,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.255972696245734e-05,
+      "loss": 1.8441,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2627986348122867e-05,
+      "loss": 1.916,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2696245733788397e-05,
+      "loss": 1.8309,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2764505119453924e-05,
+      "loss": 1.5935,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2832764505119455e-05,
+      "loss": 1.4937,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2901023890784984e-05,
+      "loss": 1.7339,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2969283276450513e-05,
+      "loss": 1.6625,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3037542662116043e-05,
+      "loss": 1.6626,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.310580204778157e-05,
+      "loss": 1.4218,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3174061433447101e-05,
+      "loss": 1.6901,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3242320819112629e-05,
+      "loss": 1.8939,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3310580204778158e-05,
+      "loss": 1.4203,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3378839590443686e-05,
+      "loss": 1.4769,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3447098976109216e-05,
+      "loss": 1.5785,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3515358361774744e-05,
+      "loss": 1.2235,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3583617747440275e-05,
+      "loss": 1.6394,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3651877133105804e-05,
+      "loss": 1.7483,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3720136518771332e-05,
+      "loss": 1.6389,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3788395904436863e-05,
+      "loss": 1.5876,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.385665529010239e-05,
+      "loss": 1.5089,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.392491467576792e-05,
+      "loss": 1.4474,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3993174061433447e-05,
+      "loss": 1.6643,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4061433447098978e-05,
+      "loss": 1.3724,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4129692832764506e-05,
+      "loss": 1.6904,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4197952218430035e-05,
+      "loss": 1.5404,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4266211604095564e-05,
+      "loss": 1.9417,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4334470989761093e-05,
+      "loss": 1.4577,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4402730375426624e-05,
+      "loss": 1.9677,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4470989761092152e-05,
+      "loss": 1.6318,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4539249146757681e-05,
+      "loss": 1.9017,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4607508532423209e-05,
+      "loss": 1.5269,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.467576791808874e-05,
+      "loss": 1.5982,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4744027303754267e-05,
+      "loss": 1.329,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4812286689419796e-05,
+      "loss": 1.7279,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4880546075085325e-05,
+      "loss": 1.5186,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4948805460750855e-05,
+      "loss": 1.437,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5017064846416382e-05,
+      "loss": 1.6116,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5085324232081913e-05,
+      "loss": 1.2695,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.515358361774744e-05,
+      "loss": 1.4696,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.522184300341297e-05,
+      "loss": 1.5908,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.52901023890785e-05,
+      "loss": 1.2862,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5358361774744027e-05,
+      "loss": 1.4425,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5426621160409558e-05,
+      "loss": 1.3423,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5494880546075085e-05,
+      "loss": 1.5395,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5563139931740616e-05,
+      "loss": 1.3667,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5631399317406144e-05,
+      "loss": 1.5799,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5699658703071675e-05,
+      "loss": 1.4498,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5767918088737202e-05,
+      "loss": 1.4169,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5836177474402733e-05,
+      "loss": 1.4842,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.590443686006826e-05,
+      "loss": 1.7601,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5972696245733788e-05,
+      "loss": 1.8778,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.604095563139932e-05,
+      "loss": 1.323,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6109215017064847e-05,
+      "loss": 1.2963,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6177474402730378e-05,
+      "loss": 1.7007,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6245733788395905e-05,
+      "loss": 1.6089,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6313993174061436e-05,
+      "loss": 1.6133,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6382252559726964e-05,
+      "loss": 1.6082,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6450511945392495e-05,
+      "loss": 1.5821,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6518771331058022e-05,
+      "loss": 1.3171,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.658703071672355e-05,
+      "loss": 1.7381,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6655290102389077e-05,
+      "loss": 1.5334,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6723549488054608e-05,
+      "loss": 1.4736,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6791808873720136e-05,
+      "loss": 1.9966,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6860068259385667e-05,
+      "loss": 1.7247,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6928327645051198e-05,
+      "loss": 1.684,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6996587030716725e-05,
+      "loss": 2.133,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7064846416382256e-05,
+      "loss": 1.6815,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7133105802047784e-05,
+      "loss": 1.6114,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.720136518771331e-05,
+      "loss": 1.5072,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.726962457337884e-05,
+      "loss": 1.4935,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.733788395904437e-05,
+      "loss": 1.2276,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7406143344709897e-05,
+      "loss": 1.6363,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7474402730375428e-05,
+      "loss": 1.4623,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7542662116040956e-05,
+      "loss": 1.8082,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7610921501706487e-05,
+      "loss": 1.7799,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7679180887372018e-05,
+      "loss": 1.5904,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7747440273037545e-05,
+      "loss": 1.6091,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7815699658703073e-05,
+      "loss": 1.1618,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.78839590443686e-05,
+      "loss": 1.644,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795221843003413e-05,
+      "loss": 1.5779,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.802047781569966e-05,
+      "loss": 1.4141,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.808873720136519e-05,
+      "loss": 1.7799,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8156996587030717e-05,
+      "loss": 1.8154,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8225255972696248e-05,
+      "loss": 1.6624,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8293515358361776e-05,
+      "loss": 1.8761,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8361774744027307e-05,
+      "loss": 1.6507,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8430034129692834e-05,
+      "loss": 1.5478,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8498293515358362e-05,
+      "loss": 2.2087,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8566552901023893e-05,
+      "loss": 1.5804,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863481228668942e-05,
+      "loss": 1.775,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.870307167235495e-05,
+      "loss": 1.5949,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.877133105802048e-05,
+      "loss": 1.4485,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.883959044368601e-05,
+      "loss": 1.4081,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8907849829351537e-05,
+      "loss": 1.3659,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8976109215017068e-05,
+      "loss": 1.5709,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9044368600682596e-05,
+      "loss": 2.0675,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9112627986348123e-05,
+      "loss": 1.6122,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.918088737201365e-05,
+      "loss": 1.743,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.924914675767918e-05,
+      "loss": 1.6753,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9317406143344713e-05,
+      "loss": 1.9251,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.938566552901024e-05,
+      "loss": 1.6176,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945392491467577e-05,
+      "loss": 1.5742,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95221843003413e-05,
+      "loss": 1.264,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959044368600683e-05,
+      "loss": 1.4678,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658703071672357e-05,
+      "loss": 1.486,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9726962457337885e-05,
+      "loss": 1.5962,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795221843003412e-05,
+      "loss": 1.2002,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9863481228668943e-05,
+      "loss": 1.7388,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993174061433447e-05,
+      "loss": 1.5878,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.7033,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999448223797e-05,
+      "loss": 1.5834,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997792895246e-05,
+      "loss": 1.661,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995034014527e-05,
+      "loss": 1.5118,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991171581943e-05,
+      "loss": 2.3058,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998620559793e-05,
+      "loss": 1.6315,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999980136063026e-05,
+      "loss": 1.629,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999972962977905e-05,
+      "loss": 1.5923,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999964686343357e-05,
+      "loss": 1.59,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999553061603e-05,
+      "loss": 1.6337,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999944822429766e-05,
+      "loss": 1.6483,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993323515291e-05,
+      "loss": 1.3941,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999920544331012e-05,
+      "loss": 2.0347,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999906749965473e-05,
+      "loss": 2.0343,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999891852057814e-05,
+      "loss": 1.4051,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999875850609682e-05,
+      "loss": 1.729,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999985874562284e-05,
+      "loss": 1.2493,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999840537099178e-05,
+      "loss": 1.6045,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999821225040705e-05,
+      "loss": 1.4329,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999800809449548e-05,
+      "loss": 1.6395,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999779290327966e-05,
+      "loss": 1.5802,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999975666767833e-05,
+      "loss": 1.7261,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999732941503137e-05,
+      "loss": 1.5612,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999708111805004e-05,
+      "loss": 1.4742,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999682178586677e-05,
+      "loss": 1.554,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999655141851013e-05,
+      "loss": 1.6712,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999627001600995e-05,
+      "loss": 2.089,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959775783973e-05,
+      "loss": 1.356,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999567410570446e-05,
+      "loss": 1.4777,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999953595979649e-05,
+      "loss": 1.2841,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999503405521335e-05,
+      "loss": 1.5484,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999469747748572e-05,
+      "loss": 1.4988,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999434986481916e-05,
+      "loss": 1.5857,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99993991217252e-05,
+      "loss": 1.8556,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999936215348239e-05,
+      "loss": 1.4925,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999324081757557e-05,
+      "loss": 1.6744,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999284906554904e-05,
+      "loss": 1.4789,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999244627878758e-05,
+      "loss": 1.5377,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999203245733563e-05,
+      "loss": 1.5447,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999916076012388e-05,
+      "loss": 1.5472,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999117171054407e-05,
+      "loss": 1.4136,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999072478529945e-05,
+      "loss": 1.7653,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999026682555436e-05,
+      "loss": 1.591,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998979783135925e-05,
+      "loss": 1.5696,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999893178027659e-05,
+      "loss": 1.336,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999888267398273e-05,
+      "loss": 1.6644,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998832464259765e-05,
+      "loss": 1.4908,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998781151113233e-05,
+      "loss": 1.4784,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998728734548797e-05,
+      "loss": 1.5425,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999867521457224e-05,
+      "loss": 1.5464,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998620591189475e-05,
+      "loss": 1.5103,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999856486440652e-05,
+      "loss": 1.8024,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998508034229537e-05,
+      "loss": 1.3096,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999845010066479e-05,
+      "loss": 1.5652,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999839106371867e-05,
+      "loss": 1.4659,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998330923397694e-05,
+      "loss": 1.3484,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998269679708504e-05,
+      "loss": 1.7015,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999820733265785e-05,
+      "loss": 1.6008,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814388225262e-05,
+      "loss": 1.479,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998079328499814e-05,
+      "loss": 1.6091,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999801367140655e-05,
+      "loss": 1.4213,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997946910980085e-05,
+      "loss": 1.4633,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997879047227772e-05,
+      "loss": 1.9898,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997810080157114e-05,
+      "loss": 1.4637,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999774000977571e-05,
+      "loss": 1.6078,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997668836091303e-05,
+      "loss": 1.1996,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997596559111743e-05,
+      "loss": 1.4808,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997523178845006e-05,
+      "loss": 1.8181,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997448695299186e-05,
+      "loss": 1.4905,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999737310848251e-05,
+      "loss": 1.5813,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999729641840331e-05,
+      "loss": 1.7686,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997218625070062e-05,
+      "loss": 1.6451,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997139728491342e-05,
+      "loss": 1.7369,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997059728675854e-05,
+      "loss": 1.5529,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996978625632435e-05,
+      "loss": 1.4705,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999689641937003e-05,
+      "loss": 1.7105,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996813109897714e-05,
+      "loss": 1.2736,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996728697224675e-05,
+      "loss": 1.4745,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996643181360235e-05,
+      "loss": 1.599,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996556562313827e-05,
+      "loss": 1.0438,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996468840095012e-05,
+      "loss": 1.5942,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996380014713466e-05,
+      "loss": 1.7721,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996290086179e-05,
+      "loss": 1.4613,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999619905450153e-05,
+      "loss": 1.4039,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996106919691104e-05,
+      "loss": 1.6185,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999601368175789e-05,
+      "loss": 1.6324,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999591934071218e-05,
+      "loss": 1.9053,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999582389656438e-05,
+      "loss": 1.3566,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995727349325027e-05,
+      "loss": 1.5909,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995629699004776e-05,
+      "loss": 1.4584,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99955309456144e-05,
+      "loss": 1.8651,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995431089164797e-05,
+      "loss": 1.6737,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995330129666985e-05,
+      "loss": 1.5434,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995228067132112e-05,
+      "loss": 1.8111,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995124901571437e-05,
+      "loss": 1.6525,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995020632996347e-05,
+      "loss": 1.6112,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994915261418344e-05,
+      "loss": 1.5036,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999480878684906e-05,
+      "loss": 1.5021,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994701209300245e-05,
+      "loss": 1.6136,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994592528783768e-05,
+      "loss": 1.4963,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994482745311626e-05,
+      "loss": 1.3799,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994371858895932e-05,
+      "loss": 1.48,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994259869548923e-05,
+      "loss": 1.4943,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994146777282957e-05,
+      "loss": 1.5814,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994032582110514e-05,
+      "loss": 1.676,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993917284044203e-05,
+      "loss": 1.5806,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993800883096735e-05,
+      "loss": 1.5766,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999368337928097e-05,
+      "loss": 1.4309,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999368337928097e-05,
+      "loss": 1.524,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999356477260986e-05,
+      "loss": 1.3746,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999344506309651e-05,
+      "loss": 1.5771,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999332425075412e-05,
+      "loss": 1.4137,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999320233559602e-05,
+      "loss": 1.3426,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999307931763567e-05,
+      "loss": 2.0822,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992955196886646e-05,
+      "loss": 2.265,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999282997336264e-05,
+      "loss": 1.3914,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999270364707748e-05,
+      "loss": 1.3948,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992576218045098e-05,
+      "loss": 1.5026,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992447686279562e-05,
+      "loss": 1.6114,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999231805179505e-05,
+      "loss": 1.9273,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992187314605874e-05,
+      "loss": 1.768,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999205547472646e-05,
+      "loss": 1.241,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991922532171354e-05,
+      "loss": 1.7043,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999178848695523e-05,
+      "loss": 1.4158,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991653339092885e-05,
+      "loss": 2.0064,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991517088599223e-05,
+      "loss": 1.4805,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991379735489288e-05,
+      "loss": 1.9642,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991241279778233e-05,
+      "loss": 1.9955,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999110172148134e-05,
+      "loss": 1.6849,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999096106061401e-05,
+      "loss": 1.4449,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990819297191762e-05,
+      "loss": 1.649,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990676431230246e-05,
+      "loss": 1.4082,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990532462745228e-05,
+      "loss": 1.6755,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990387391752584e-05,
+      "loss": 1.7446,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999024121826834e-05,
+      "loss": 1.6249,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999009394230862e-05,
+      "loss": 1.5004,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989945563889673e-05,
+      "loss": 1.8699,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989796083027872e-05,
+      "loss": 1.6586,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989645499739724e-05,
+      "loss": 1.6217,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989493814041837e-05,
+      "loss": 1.7743,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989341025950956e-05,
+      "loss": 1.4678,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989187135483932e-05,
+      "loss": 1.5986,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989032142657762e-05,
+      "loss": 1.5511,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998887604748954e-05,
+      "loss": 1.2945,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988718849996495e-05,
+      "loss": 1.6456,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988560550195978e-05,
+      "loss": 1.4765,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998840114810545e-05,
+      "loss": 1.7208,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988240643742508e-05,
+      "loss": 1.5357,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988079037124866e-05,
+      "loss": 1.8481,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987916328270352e-05,
+      "loss": 1.5341,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987752517196926e-05,
+      "loss": 1.4246,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987587603922666e-05,
+      "loss": 1.4106,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987421588465764e-05,
+      "loss": 1.5021,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987254470844553e-05,
+      "loss": 1.4565,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987086251077464e-05,
+      "loss": 1.8795,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986916929183068e-05,
+      "loss": 1.8171,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986746505180047e-05,
+      "loss": 1.5041,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998657497908721e-05,
+      "loss": 1.4948,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986402350923484e-05,
+      "loss": 1.5116,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998622862070792e-05,
+      "loss": 1.7446,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986053788459694e-05,
+      "loss": 1.5529,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998587785419809e-05,
+      "loss": 1.5559,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985700817942533e-05,
+      "loss": 1.5746,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985522679712556e-05,
+      "loss": 1.6183,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985343439527816e-05,
+      "loss": 1.5039,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99851630974081e-05,
+      "loss": 2.0118,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984981653373298e-05,
+      "loss": 1.5081,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984799107443445e-05,
+      "loss": 1.973,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984615459638678e-05,
+      "loss": 2.2094,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984430709979266e-05,
+      "loss": 1.7333,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984244858485596e-05,
+      "loss": 1.6434,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998405790517818e-05,
+      "loss": 1.5258,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983869850077653e-05,
+      "loss": 1.4307,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998368069320476e-05,
+      "loss": 1.4171,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983490434580376e-05,
+      "loss": 1.7089,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99832990742255e-05,
+      "loss": 1.7096,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9983106612161253e-05,
+      "loss": 1.5599,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982913048408865e-05,
+      "loss": 1.5422,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982718382989708e-05,
+      "loss": 1.9057,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982522615925257e-05,
+      "loss": 1.4787,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982325747237113e-05,
+      "loss": 1.6398,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998212777694701e-05,
+      "loss": 2.298,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998192870507679e-05,
+      "loss": 1.7504,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981728531648424e-05,
+      "loss": 1.4923,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981527256684e-05,
+      "loss": 1.6372,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998132488020573e-05,
+      "loss": 1.8708,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981121402235945e-05,
+      "loss": 1.6254,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980916822797106e-05,
+      "loss": 1.3303,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980711141911788e-05,
+      "loss": 1.4385,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980504359602685e-05,
+      "loss": 1.8222,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998029647589262e-05,
+      "loss": 1.6983,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998008749080453e-05,
+      "loss": 1.6535,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979877404361483e-05,
+      "loss": 1.3782,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979666216586658e-05,
+      "loss": 1.4543,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979453927503366e-05,
+      "loss": 1.4867,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997924053713503e-05,
+      "loss": 1.8551,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99790260455052e-05,
+      "loss": 1.5396,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978810452637544e-05,
+      "loss": 1.4206,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997859375855586e-05,
+      "loss": 1.612,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978375963284055e-05,
+      "loss": 2.0936,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978157066846167e-05,
+      "loss": 1.6698,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977937069266355e-05,
+      "loss": 1.4183,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997771597056889e-05,
+      "loss": 1.6282,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977493770778173e-05,
+      "loss": 1.2097,
+      "step": 499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977270469918728e-05,
+      "loss": 1.3885,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99770460680152e-05,
+      "loss": 1.5186,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976820565092345e-05,
+      "loss": 1.5715,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976593961175054e-05,
+      "loss": 1.4919,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9976366256288332e-05,
+      "loss": 1.9499,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997613745045731e-05,
+      "loss": 1.2598,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975907543707236e-05,
+      "loss": 1.8084,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997567653606348e-05,
+      "loss": 1.6581,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975444427551534e-05,
+      "loss": 1.5358,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997521121819702e-05,
+      "loss": 1.4635,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974976908025664e-05,
+      "loss": 1.5068,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974741497063334e-05,
+      "loss": 1.6823,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974504985335996e-05,
+      "loss": 1.6135,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997426737286976e-05,
+      "loss": 1.66,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974028659690844e-05,
+      "loss": 1.5419,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973788845825592e-05,
+      "loss": 1.5038,
+      "step": 515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973547931300467e-05,
+      "loss": 1.2876,
+      "step": 516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997330591614206e-05,
+      "loss": 1.6949,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973062800377078e-05,
+      "loss": 1.4344,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972818584032343e-05,
+      "loss": 1.8135,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997257326713481e-05,
+      "loss": 2.073,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972326849711553e-05,
+      "loss": 1.4165,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972079331789765e-05,
+      "loss": 1.7525,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997183071339676e-05,
+      "loss": 1.6875,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997158099455997e-05,
+      "loss": 1.3993,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997133017530696e-05,
+      "loss": 1.5081,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971078255665405e-05,
+      "loss": 1.6138,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970825235663106e-05,
+      "loss": 1.6484,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970571115327987e-05,
+      "loss": 1.5746,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970315894688087e-05,
+      "loss": 1.4661,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997005957377158e-05,
+      "loss": 1.9931,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996980215260674e-05,
+      "loss": 1.4756,
+      "step": 531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969543631221983e-05,
+      "loss": 1.4312,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996928400964584e-05,
+      "loss": 2.1807,
+      "step": 533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969023287906957e-05,
+      "loss": 1.7448,
+      "step": 534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968761466034103e-05,
+      "loss": 1.5013,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968498544056182e-05,
+      "loss": 1.5416,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9968234522002197e-05,
+      "loss": 1.6837,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996796939990129e-05,
+      "loss": 1.8052,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996770317778272e-05,
+      "loss": 1.3089,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967435855675862e-05,
+      "loss": 1.3372,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996716743361022e-05,
+      "loss": 1.5434,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966897911615417e-05,
+      "loss": 1.6657,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996662728972119e-05,
+      "loss": 1.5276,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966355567957407e-05,
+      "loss": 1.8584,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966082746354055e-05,
+      "loss": 1.4268,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996580882494124e-05,
+      "loss": 2.0982,
+      "step": 546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965533803749192e-05,
+      "loss": 2.1043,
+      "step": 547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996525768280826e-05,
+      "loss": 1.4245,
+      "step": 548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996498046214891e-05,
+      "loss": 1.5443,
+      "step": 549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964702141801746e-05,
+      "loss": 1.3153,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964422721797474e-05,
+      "loss": 1.3179,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964142202166933e-05,
+      "loss": 2.051,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963860582941076e-05,
+      "loss": 1.5405,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996357786415099e-05,
+      "loss": 1.6562,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996329404582786e-05,
+      "loss": 1.3842,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996300912800302e-05,
+      "loss": 1.6904,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962723110707905e-05,
+      "loss": 1.3951,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996243599397408e-05,
+      "loss": 1.2291,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996214777783323e-05,
+      "loss": 1.5196,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961858462317164e-05,
+      "loss": 1.7897,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961568047457807e-05,
+      "loss": 1.6486,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961276533287206e-05,
+      "loss": 1.5862,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960983919837534e-05,
+      "loss": 1.3973,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960690207141083e-05,
+      "loss": 1.5312,
+      "step": 564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996039539523026e-05,
+      "loss": 1.963,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960099484137608e-05,
+      "loss": 1.5035,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959802473895778e-05,
+      "loss": 2.2408,
+      "step": 567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959504364537544e-05,
+      "loss": 1.2749,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9959205156095806e-05,
+      "loss": 1.6764,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958904848603584e-05,
+      "loss": 1.829,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958603442094014e-05,
+      "loss": 1.6057,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958300936600368e-05,
+      "loss": 1.732,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995799733215602e-05,
+      "loss": 1.3931,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957692628794477e-05,
+      "loss": 1.7691,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957386826549364e-05,
+      "loss": 1.4688,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957079925454427e-05,
+      "loss": 1.5976,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956771925543543e-05,
+      "loss": 1.4065,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956771925543543e-05,
+      "loss": 2.2596,
+      "step": 578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956462826850687e-05,
+      "loss": 2.1677,
+      "step": 579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995615262940998e-05,
+      "loss": 1.5793,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995584133325565e-05,
+      "loss": 1.6142,
+      "step": 581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955528938422048e-05,
+      "loss": 1.248,
+      "step": 582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955215444943655e-05,
+      "loss": 1.2069,
+      "step": 583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995490085285506e-05,
+      "loss": 1.3937,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954585162190986e-05,
+      "loss": 1.5343,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954268372986265e-05,
+      "loss": 1.4141,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995395048527586e-05,
+      "loss": 1.8331,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995363149909485e-05,
+      "loss": 1.6632,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995331141447844e-05,
+      "loss": 1.4762,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995299023146195e-05,
+      "loss": 1.8747,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952667950080823e-05,
+      "loss": 1.6779,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995234457037063e-05,
+      "loss": 2.1436,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952020092367052e-05,
+      "loss": 1.2505,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951694516105897e-05,
+      "loss": 1.5307,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951367841623097e-05,
+      "loss": 1.9421,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99510400689547e-05,
+      "loss": 1.4969,
+      "step": 596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950711198136882e-05,
+      "loss": 1.6654,
+      "step": 597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950381229205932e-05,
+      "loss": 1.7871,
+      "step": 598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995005016219826e-05,
+      "loss": 1.6526,
+      "step": 599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949717997150407e-05,
+      "loss": 1.5874,
+      "step": 600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994938473409903e-05,
+      "loss": 1.6862,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9949050373080898e-05,
+      "loss": 1.4911,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948714914132918e-05,
+      "loss": 1.3573,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948378357292104e-05,
+      "loss": 1.5713,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948040702595604e-05,
+      "loss": 1.6453,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994770195008067e-05,
+      "loss": 1.9384,
+      "step": 606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947362099784694e-05,
+      "loss": 1.6605,
+      "step": 607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947021151745175e-05,
+      "loss": 1.4581,
+      "step": 608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946679105999742e-05,
+      "loss": 1.4024,
+      "step": 609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946335962586142e-05,
+      "loss": 1.9073,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945991721542235e-05,
+      "loss": 1.3773,
+      "step": 611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945646382906018e-05,
+      "loss": 1.4953,
+      "step": 612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945299946715594e-05,
+      "loss": 1.291,
+      "step": 613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99449524130092e-05,
+      "loss": 1.7262,
+      "step": 614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994460378182519e-05,
+      "loss": 1.5309,
+      "step": 615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994425405320203e-05,
+      "loss": 1.3383,
+      "step": 616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943903227178318e-05,
+      "loss": 1.7479,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994355130379277e-05,
+      "loss": 1.7613,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943198283084225e-05,
+      "loss": 1.337,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942844165091636e-05,
+      "loss": 1.6086,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994248894985408e-05,
+      "loss": 1.5045,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942132637410766e-05,
+      "loss": 1.4088,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941775227801005e-05,
+      "loss": 1.6192,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941416721064243e-05,
+      "loss": 1.2725,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941057117240046e-05,
+      "loss": 1.3897,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940696416368094e-05,
+      "loss": 1.8737,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940334618488194e-05,
+      "loss": 1.4779,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939971723640274e-05,
+      "loss": 1.4105,
+      "step": 628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939607731864375e-05,
+      "loss": 1.418,
+      "step": 629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939242643200674e-05,
+      "loss": 1.4709,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938876457689456e-05,
+      "loss": 1.7243,
+      "step": 631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993850917537113e-05,
+      "loss": 1.4015,
+      "step": 632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993814079628623e-05,
+      "loss": 1.5421,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993777132047541e-05,
+      "loss": 1.5955,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9937400747979437e-05,
+      "loss": 1.6731,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9937029078839213e-05,
+      "loss": 1.31,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993665631309575e-05,
+      "loss": 1.3476,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936282450790187e-05,
+      "loss": 1.8276,
+      "step": 638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993590749196378e-05,
+      "loss": 1.1895,
+      "step": 639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935531436657906e-05,
+      "loss": 1.6257,
+      "step": 640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935154284914063e-05,
+      "loss": 1.3587,
+      "step": 641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934776036773883e-05,
+      "loss": 2.0791,
+      "step": 642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934396692279092e-05,
+      "loss": 1.7037,
+      "step": 643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934016251471563e-05,
+      "loss": 1.762,
+      "step": 644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993363471439328e-05,
+      "loss": 1.6573,
+      "step": 645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933252081086342e-05,
+      "loss": 1.6636,
+      "step": 646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932868351592978e-05,
+      "loss": 1.7634,
+      "step": 647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932483525955533e-05,
+      "loss": 1.6434,
+      "step": 648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993209760421648e-05,
+      "loss": 1.4759,
+      "step": 649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931710586418397e-05,
+      "loss": 1.5184,
+      "step": 650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931322472604e-05,
+      "loss": 1.4037,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993093326281612e-05,
+      "loss": 1.7192,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930542957097706e-05,
+      "loss": 1.7411,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930151555491834e-05,
+      "loss": 1.7491,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929759058041687e-05,
+      "loss": 1.8055,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929365464790594e-05,
+      "loss": 1.4622,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992897077578198e-05,
+      "loss": 1.625,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928574991059407e-05,
+      "loss": 1.5797,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928178110666545e-05,
+      "loss": 1.4542,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927780134647194e-05,
+      "loss": 1.6552,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927381063045275e-05,
+      "loss": 1.5097,
+      "step": 661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992698089590483e-05,
+      "loss": 1.6061,
+      "step": 662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926579633270013e-05,
+      "loss": 1.6021,
+      "step": 663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926177275185114e-05,
+      "loss": 1.7588,
+      "step": 664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925773821694523e-05,
+      "loss": 1.5519,
+      "step": 665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925369272842776e-05,
+      "loss": 1.7288,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992496362867451e-05,
+      "loss": 1.5182,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992455688923449e-05,
+      "loss": 1.4346,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9924149054567606e-05,
+      "loss": 1.7663,
+      "step": 669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992374012471886e-05,
+      "loss": 1.874,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923330099733378e-05,
+      "loss": 1.3484,
+      "step": 671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922918979656417e-05,
+      "loss": 1.3867,
+      "step": 672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992250676453334e-05,
+      "loss": 1.4803,
+      "step": 673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922093454409633e-05,
+      "loss": 1.4734,
+      "step": 674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992167904933092e-05,
+      "loss": 1.7392,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992126354934292e-05,
+      "loss": 1.7629,
+      "step": 676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920846954491494e-05,
+      "loss": 1.7587,
+      "step": 677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992042926482261e-05,
+      "loss": 1.2402,
+      "step": 678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920010480382362e-05,
+      "loss": 1.7698,
+      "step": 679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991959060121697e-05,
+      "loss": 1.9172,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919169627372766e-05,
+      "loss": 1.9026,
+      "step": 681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991874755889621e-05,
+      "loss": 1.5902,
+      "step": 682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918324395833876e-05,
+      "loss": 1.7575,
+      "step": 683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991790013823246e-05,
+      "loss": 1.3494,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917474786138792e-05,
+      "loss": 1.5937,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99170483395998e-05,
+      "loss": 1.4726,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916620798662548e-05,
+      "loss": 1.8618,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991619216337422e-05,
+      "loss": 1.3118,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915762433782116e-05,
+      "loss": 1.2244,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991533160993366e-05,
+      "loss": 2.0159,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991489969187639e-05,
+      "loss": 1.8483,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914466679657984e-05,
+      "loss": 1.4382,
+      "step": 692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914032573326218e-05,
+      "loss": 1.7297,
+      "step": 693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913597372928993e-05,
+      "loss": 1.5995,
+      "step": 694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913161078514344e-05,
+      "loss": 1.806,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912723690130417e-05,
+      "loss": 1.6206,
+      "step": 696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912285207825474e-05,
+      "loss": 1.7156,
+      "step": 697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911845631647916e-05,
+      "loss": 1.651,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991140496164624e-05,
+      "loss": 1.8355,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991096319786908e-05,
+      "loss": 1.3756,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991052034036519e-05,
+      "loss": 1.7723,
+      "step": 701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9910076389183442e-05,
+      "loss": 1.4516,
+      "step": 702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909631344372823e-05,
+      "loss": 1.5423,
+      "step": 703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909185205982453e-05,
+      "loss": 1.2375,
+      "step": 704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990873797406156e-05,
+      "loss": 1.71,
+      "step": 705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908289648659502e-05,
+      "loss": 1.6509,
+      "step": 706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907840229825747e-05,
+      "loss": 1.7626,
+      "step": 707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907389717609903e-05,
+      "loss": 1.6821,
+      "step": 708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906938112061674e-05,
+      "loss": 1.8302,
+      "step": 709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906485413230907e-05,
+      "loss": 1.6214,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906031621167553e-05,
+      "loss": 1.8574,
+      "step": 711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905576735921695e-05,
+      "loss": 1.1781,
+      "step": 712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990512075754353e-05,
+      "loss": 1.646,
+      "step": 713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904663686083373e-05,
+      "loss": 1.5215,
+      "step": 714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990420552159167e-05,
+      "loss": 1.5374,
+      "step": 715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990374626411898e-05,
+      "loss": 1.6062,
+      "step": 716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903285913715987e-05,
+      "loss": 1.6697,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902824470433487e-05,
+      "loss": 1.4987,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990236193432241e-05,
+      "loss": 1.4717,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901898305433797e-05,
+      "loss": 1.599,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901433583818807e-05,
+      "loss": 1.686,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990096776952873e-05,
+      "loss": 1.6576,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990050086261497e-05,
+      "loss": 1.9293,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990003286312905e-05,
+      "loss": 1.4186,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899563771122617e-05,
+      "loss": 1.4004,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989909358664744e-05,
+      "loss": 1.4242,
+      "step": 726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898622309755405e-05,
+      "loss": 1.2479,
+      "step": 727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898149940498523e-05,
+      "loss": 2.1678,
+      "step": 728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897676478928917e-05,
+      "loss": 1.5936,
+      "step": 729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989720192509884e-05,
+      "loss": 1.7643,
+      "step": 730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896726279060656e-05,
+      "loss": 1.4136,
+      "step": 731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989624954086686e-05,
+      "loss": 1.5237,
+      "step": 732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895771710570062e-05,
+      "loss": 1.693,
+      "step": 733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895292788222993e-05,
+      "loss": 1.4156,
+      "step": 734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894812773878507e-05,
+      "loss": 1.5377,
+      "step": 735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989433166758957e-05,
+      "loss": 1.6589,
+      "step": 736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893849469409278e-05,
+      "loss": 1.5881,
+      "step": 737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893366179390843e-05,
+      "loss": 1.8605,
+      "step": 738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98928817975876e-05,
+      "loss": 1.5213,
+      "step": 739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892396324053005e-05,
+      "loss": 1.4541,
+      "step": 740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989190975884063e-05,
+      "loss": 1.4941,
+      "step": 741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891422102004164e-05,
+      "loss": 1.7401,
+      "step": 742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890933353597438e-05,
+      "loss": 1.4275,
+      "step": 743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890443513674372e-05,
+      "loss": 1.5721,
+      "step": 744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889952582289028e-05,
+      "loss": 1.4532,
+      "step": 745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988946055949559e-05,
+      "loss": 1.5174,
+      "step": 746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888967445348343e-05,
+      "loss": 2.0316,
+      "step": 747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888473239901717e-05,
+      "loss": 1.6171,
+      "step": 748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988797794321024e-05,
+      "loss": 1.4597,
+      "step": 749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887481555328573e-05,
+      "loss": 1.5823,
+      "step": 750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98869840763115e-05,
+      "loss": 1.5435,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988648550621392e-05,
+      "loss": 1.3615,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988598584509084e-05,
+      "loss": 1.7165,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885485092997418e-05,
+      "loss": 1.7978,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884983249988907e-05,
+      "loss": 1.5621,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884480316120683e-05,
+      "loss": 1.6238,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883976291448254e-05,
+      "loss": 1.567,
+      "step": 757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883471176027237e-05,
+      "loss": 1.2839,
+      "step": 758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882964969913383e-05,
+      "loss": 1.6923,
+      "step": 759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882457673162546e-05,
+      "loss": 1.399,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988194928583071e-05,
+      "loss": 1.8525,
+      "step": 761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988143980797398e-05,
+      "loss": 1.4326,
+      "step": 762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988092923964858e-05,
+      "loss": 1.4711,
+      "step": 763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9880417580910852e-05,
+      "loss": 1.2785,
+      "step": 764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879904831817264e-05,
+      "loss": 1.7089,
+      "step": 765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879390992424395e-05,
+      "loss": 1.3768,
+      "step": 766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878876062788955e-05,
+      "loss": 1.5763,
+      "step": 767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878360042967762e-05,
+      "loss": 1.405,
+      "step": 768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877842933017772e-05,
+      "loss": 1.486,
+      "step": 769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987732473299604e-05,
+      "loss": 1.7129,
+      "step": 770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876805442959762e-05,
+      "loss": 1.7255,
+      "step": 771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876285062966238e-05,
+      "loss": 1.6232,
+      "step": 772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875763593072892e-05,
+      "loss": 2.0336,
+      "step": 773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987524103333728e-05,
+      "loss": 1.3588,
+      "step": 774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874717383817064e-05,
+      "loss": 1.4183,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874192644570028e-05,
+      "loss": 1.5181,
+      "step": 776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987366681565409e-05,
+      "loss": 1.5663,
+      "step": 777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873139897127266e-05,
+      "loss": 1.3815,
+      "step": 778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872611889047713e-05,
+      "loss": 1.3916,
+      "step": 779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872082791473695e-05,
+      "loss": 1.435,
+      "step": 780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871552604463602e-05,
+      "loss": 1.4591,
+      "step": 781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871021328075943e-05,
+      "loss": 1.4685,
+      "step": 782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870488962369344e-05,
+      "loss": 1.6692,
+      "step": 783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986995550740256e-05,
+      "loss": 2.2495,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869420963234458e-05,
+      "loss": 1.5355,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868885329924027e-05,
+      "loss": 1.5425,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986834860753038e-05,
+      "loss": 1.7291,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867810796112742e-05,
+      "loss": 1.8381,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867271895730467e-05,
+      "loss": 1.7239,
+      "step": 789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866731906443024e-05,
+      "loss": 1.8158,
+      "step": 790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866190828310005e-05,
+      "loss": 1.7101,
+      "step": 791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865648661391118e-05,
+      "loss": 1.5767,
+      "step": 792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865105405746198e-05,
+      "loss": 1.3931,
+      "step": 793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864561061435194e-05,
+      "loss": 1.491,
+      "step": 794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864015628518175e-05,
+      "loss": 1.5456,
+      "step": 795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863469107055334e-05,
+      "loss": 2.1687,
+      "step": 796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9862921497106987e-05,
+      "loss": 1.2406,
+      "step": 797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986237279873356e-05,
+      "loss": 2.1133,
+      "step": 798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9861823011995606e-05,
+      "loss": 1.6885,
+      "step": 799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98612721369538e-05,
+      "loss": 1.6701,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860720173668926e-05,
+      "loss": 1.5124,
+      "step": 801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860167122201904e-05,
+      "loss": 1.7562,
+      "step": 802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859612982613765e-05,
+      "loss": 1.767,
+      "step": 803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985905775496566e-05,
+      "loss": 1.3863,
+      "step": 804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858501439318858e-05,
+      "loss": 1.5171,
+      "step": 805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857944035734757e-05,
+      "loss": 1.5468,
+      "step": 806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857385544274864e-05,
+      "loss": 1.6217,
+      "step": 807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856825965000816e-05,
+      "loss": 1.6499,
+      "step": 808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856265297974362e-05,
+      "loss": 1.4462,
+      "step": 809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985570354325738e-05,
+      "loss": 1.5257,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855140700911854e-05,
+      "loss": 1.3318,
+      "step": 811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98545767709999e-05,
+      "loss": 1.3876,
+      "step": 812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985401175358376e-05,
+      "loss": 1.4457,
+      "step": 813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985344564872577e-05,
+      "loss": 1.4038,
+      "step": 814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852878456488418e-05,
+      "loss": 1.4218,
+      "step": 815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985231017693429e-05,
+      "loss": 1.6449,
+      "step": 816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851740810126093e-05,
+      "loss": 1.4397,
+      "step": 817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985117035612667e-05,
+      "loss": 1.9145,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850598814998968e-05,
+      "loss": 1.6752,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850026186806058e-05,
+      "loss": 1.9077,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849452471611135e-05,
+      "loss": 1.4852,
+      "step": 821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848877669477514e-05,
+      "loss": 1.3892,
+      "step": 822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848301780468623e-05,
+      "loss": 1.509,
+      "step": 823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847724804648017e-05,
+      "loss": 1.3761,
+      "step": 824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847146742079366e-05,
+      "loss": 2.5869,
+      "step": 825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846567592826464e-05,
+      "loss": 1.495,
+      "step": 826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845987356953223e-05,
+      "loss": 1.4178,
+      "step": 827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984540603452368e-05,
+      "loss": 1.5148,
+      "step": 828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844823625601973e-05,
+      "loss": 1.5193,
+      "step": 829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844240130252385e-05,
+      "loss": 1.5092,
+      "step": 830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984365554853931e-05,
+      "loss": 1.2061,
+      "step": 831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9843069880527254e-05,
+      "loss": 1.6058,
+      "step": 832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842483126280844e-05,
+      "loss": 1.46,
+      "step": 833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984189528586484e-05,
+      "loss": 1.5357,
+      "step": 834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841306359344115e-05,
+      "loss": 1.982,
+      "step": 835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840716346783652e-05,
+      "loss": 1.6231,
+      "step": 836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840125248248564e-05,
+      "loss": 1.3567,
+      "step": 837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839533063804084e-05,
+      "loss": 1.51,
+      "step": 838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838939793515563e-05,
+      "loss": 1.8142,
+      "step": 839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983834543744847e-05,
+      "loss": 1.4479,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837749995668397e-05,
+      "loss": 1.6978,
+      "step": 841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837153468241052e-05,
+      "loss": 1.5742,
+      "step": 842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836555855232266e-05,
+      "loss": 1.6474,
+      "step": 843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835957156707987e-05,
+      "loss": 1.6149,
+      "step": 844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835357372734286e-05,
+      "loss": 1.4583,
+      "step": 845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834756503377354e-05,
+      "loss": 1.4211,
+      "step": 846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98341545487035e-05,
+      "loss": 1.4809,
+      "step": 847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833551508779147e-05,
+      "loss": 1.4054,
+      "step": 848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983294738367085e-05,
+      "loss": 1.7427,
+      "step": 849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832342173445274e-05,
+      "loss": 2.0008,
+      "step": 850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831735878169212e-05,
+      "loss": 1.6367,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831128497909565e-05,
+      "loss": 1.6217,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830520032733365e-05,
+      "loss": 1.2211,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829910482707758e-05,
+      "loss": 1.3007,
+      "step": 854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982929984790001e-05,
+      "loss": 1.72,
+      "step": 855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828688128377512e-05,
+      "loss": 1.9176,
+      "step": 856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982807532420777e-05,
+      "loss": 1.5821,
+      "step": 857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98274614354584e-05,
+      "loss": 1.497,
+      "step": 858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982684646219716e-05,
+      "loss": 1.4581,
+      "step": 859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826230404491906e-05,
+      "loss": 1.7315,
+      "step": 860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825613262410632e-05,
+      "loss": 1.718,
+      "step": 861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982499503602144e-05,
+      "loss": 2.8577,
+      "step": 862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982437572539255e-05,
+      "loss": 1.6964,
+      "step": 863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9823755330592313e-05,
+      "loss": 1.5184,
+      "step": 864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9823133851689186e-05,
+      "loss": 1.2773,
+      "step": 865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982251128875176e-05,
+      "loss": 1.8845,
+      "step": 866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982188764184873e-05,
+      "loss": 1.4567,
+      "step": 867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821262911048928e-05,
+      "loss": 2.2927,
+      "step": 868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982063709642129e-05,
+      "loss": 1.2002,
+      "step": 869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820010198034874e-05,
+      "loss": 1.6114,
+      "step": 870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819382215958873e-05,
+      "loss": 1.37,
+      "step": 871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818753150262576e-05,
+      "loss": 1.9415,
+      "step": 872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818123001015412e-05,
+      "loss": 1.553,
+      "step": 873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817491768286917e-05,
+      "loss": 1.3806,
+      "step": 874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816859452146754e-05,
+      "loss": 1.6157,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98162260526647e-05,
+      "loss": 1.35,
+      "step": 876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815591569910654e-05,
+      "loss": 1.9254,
+      "step": 877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814956003954636e-05,
+      "loss": 1.7744,
+      "step": 878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814319354866787e-05,
+      "loss": 1.5809,
+      "step": 879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813681622717357e-05,
+      "loss": 1.5324,
+      "step": 880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813042807576726e-05,
+      "loss": 1.7413,
+      "step": 881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812402909515392e-05,
+      "loss": 1.5715,
+      "step": 882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981176192860397e-05,
+      "loss": 1.9367,
+      "step": 883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811119864913202e-05,
+      "loss": 1.4647,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810476718513933e-05,
+      "loss": 1.6779,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809832489477144e-05,
+      "loss": 1.4867,
+      "step": 886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980918717787392e-05,
+      "loss": 1.9358,
+      "step": 887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808540783775488e-05,
+      "loss": 1.3659,
+      "step": 888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807893307253177e-05,
+      "loss": 1.5247,
+      "step": 889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807244748378433e-05,
+      "loss": 1.7879,
+      "step": 890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806595107222832e-05,
+      "loss": 1.5668,
+      "step": 891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805944383858066e-05,
+      "loss": 1.5243,
+      "step": 892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805292578355943e-05,
+      "loss": 1.6486,
+      "step": 893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9804639690788393e-05,
+      "loss": 1.4305,
+      "step": 894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980398572122747e-05,
+      "loss": 1.5939,
+      "step": 895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9803330669745346e-05,
+      "loss": 1.4489,
+      "step": 896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9802674536414295e-05,
+      "loss": 1.5814,
+      "step": 897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980201732130674e-05,
+      "loss": 1.4971,
+      "step": 898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801359024495198e-05,
+      "loss": 1.3821,
+      "step": 899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800699646052323e-05,
+      "loss": 1.676,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800039186050872e-05,
+      "loss": 1.7135,
+      "step": 901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799377644563737e-05,
+      "loss": 1.4887,
+      "step": 902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798715021663923e-05,
+      "loss": 1.3025,
+      "step": 903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798051317424552e-05,
+      "loss": 1.6145,
+      "step": 904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797386531918862e-05,
+      "loss": 2.0085,
+      "step": 905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796720665220226e-05,
+      "loss": 1.4373,
+      "step": 906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979605371740212e-05,
+      "loss": 1.7,
+      "step": 907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795385688538144e-05,
+      "loss": 1.331,
+      "step": 908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794716578702023e-05,
+      "loss": 1.6752,
+      "step": 909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794046387967588e-05,
+      "loss": 1.5682,
+      "step": 910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979337511640881e-05,
+      "loss": 1.7072,
+      "step": 911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792702764099758e-05,
+      "loss": 1.7347,
+      "step": 912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792029331114634e-05,
+      "loss": 1.4593,
+      "step": 913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791354817527756e-05,
+      "loss": 1.209,
+      "step": 914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790679223413555e-05,
+      "loss": 1.5657,
+      "step": 915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979000254884659e-05,
+      "loss": 1.5237,
+      "step": 916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789324793901534e-05,
+      "loss": 1.5376,
+      "step": 917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788645958653182e-05,
+      "loss": 1.6251,
+      "step": 918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978796604317645e-05,
+      "loss": 1.5611,
+      "step": 919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787285047546363e-05,
+      "loss": 1.4204,
+      "step": 920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786602971838075e-05,
+      "loss": 1.3893,
+      "step": 921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978591981612686e-05,
+      "loss": 1.3754,
+      "step": 922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978523558048811e-05,
+      "loss": 2.2757,
+      "step": 923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784550264997327e-05,
+      "loss": 1.5594,
+      "step": 924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978386386973014e-05,
+      "loss": 1.7422,
+      "step": 925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97831763947623e-05,
+      "loss": 1.5653,
+      "step": 926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978248784016967e-05,
+      "loss": 1.5665,
+      "step": 927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978179820602824e-05,
+      "loss": 2.6619,
+      "step": 928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978110749241411e-05,
+      "loss": 1.5184,
+      "step": 929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978041569940351e-05,
+      "loss": 1.715,
+      "step": 930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977972282707277e-05,
+      "loss": 1.5634,
+      "step": 931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977902887549837e-05,
+      "loss": 1.4804,
+      "step": 932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977833384475688e-05,
+      "loss": 1.5132,
+      "step": 933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777637734925e-05,
+      "loss": 1.4768,
+      "step": 934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776940546079552e-05,
+      "loss": 1.5759,
+      "step": 935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776242278297476e-05,
+      "loss": 1.4935,
+      "step": 936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775542931655827e-05,
+      "loss": 1.7302,
+      "step": 937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774842506231783e-05,
+      "loss": 1.3504,
+      "step": 938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774141002102637e-05,
+      "loss": 1.4613,
+      "step": 939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977343841934581e-05,
+      "loss": 1.6983,
+      "step": 940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977273475803883e-05,
+      "loss": 1.4934,
+      "step": 941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977203001825935e-05,
+      "loss": 1.454,
+      "step": 942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977132420008514e-05,
+      "loss": 1.5769,
+      "step": 943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97706173035941e-05,
+      "loss": 1.476,
+      "step": 944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769909328864228e-05,
+      "loss": 1.9073,
+      "step": 945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769200275973658e-05,
+      "loss": 1.4637,
+      "step": 946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976849014500064e-05,
+      "loss": 1.4613,
+      "step": 947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767778936023537e-05,
+      "loss": 1.5522,
+      "step": 948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976706664912084e-05,
+      "loss": 1.4257,
+      "step": 949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766353284371145e-05,
+      "loss": 1.6014,
+      "step": 950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976563884185318e-05,
+      "loss": 1.5235,
+      "step": 951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976492332164579e-05,
+      "loss": 1.6117,
+      "step": 952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764206723827932e-05,
+      "loss": 1.3916,
+      "step": 953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976348904847869e-05,
+      "loss": 1.6441,
+      "step": 954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976277029567726e-05,
+      "loss": 1.7254,
+      "step": 955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762050465502963e-05,
+      "loss": 1.732,
+      "step": 956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761329558035236e-05,
+      "loss": 1.5403,
+      "step": 957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976060757335363e-05,
+      "loss": 1.6605,
+      "step": 958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9759884511537827e-05,
+      "loss": 1.6501,
+      "step": 959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9759160372667615e-05,
+      "loss": 1.58,
+      "step": 960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975843515682291e-05,
+      "loss": 1.8685,
+      "step": 961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975770886408374e-05,
+      "loss": 1.5215,
+      "step": 962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975698149453026e-05,
+      "loss": 1.4338,
+      "step": 963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756253048242735e-05,
+      "loss": 1.3755,
+      "step": 964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755523525301554e-05,
+      "loss": 1.5886,
+      "step": 965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754792925787225e-05,
+      "loss": 1.456,
+      "step": 966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975406124978037e-05,
+      "loss": 3.0461,
+      "step": 967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975332849736174e-05,
+      "loss": 1.4703,
+      "step": 968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752594668612187e-05,
+      "loss": 1.3527,
+      "step": 969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751859763612705e-05,
+      "loss": 1.8495,
+      "step": 970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975112378244439e-05,
+      "loss": 1.4265,
+      "step": 971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750386725188452e-05,
+      "loss": 1.4666,
+      "step": 972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749648591926244e-05,
+      "loss": 1.4805,
+      "step": 973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748909382739218e-05,
+      "loss": 1.7396,
+      "step": 974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748169097708944e-05,
+      "loss": 1.5162,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747427736917123e-05,
+      "loss": 1.616,
+      "step": 976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746685300445565e-05,
+      "loss": 1.6889,
+      "step": 977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745941788376203e-05,
+      "loss": 1.5825,
+      "step": 978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745197200791085e-05,
+      "loss": 1.5128,
+      "step": 979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974445153777238e-05,
+      "loss": 1.8438,
+      "step": 980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743704799402384e-05,
+      "loss": 1.2703,
+      "step": 981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974295698576349e-05,
+      "loss": 1.786,
+      "step": 982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742208096938236e-05,
+      "loss": 1.6064,
+      "step": 983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741458133009258e-05,
+      "loss": 1.4942,
+      "step": 984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974070709405932e-05,
+      "loss": 1.3738,
+      "step": 985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739954980171302e-05,
+      "loss": 1.2066,
+      "step": 986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739201791428207e-05,
+      "loss": 1.4877,
+      "step": 987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738447527913155e-05,
+      "loss": 1.5033,
+      "step": 988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737692189709373e-05,
+      "loss": 1.7519,
+      "step": 989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736935776900232e-05,
+      "loss": 1.9996,
+      "step": 990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736178289569186e-05,
+      "loss": 1.7117,
+      "step": 991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9735419727799846e-05,
+      "loss": 1.6263,
+      "step": 992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9734660091675917e-05,
+      "loss": 1.385,
+      "step": 993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9733899381281227e-05,
+      "loss": 1.9723,
+      "step": 994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9733137596699726e-05,
+      "loss": 2.0769,
+      "step": 995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732374738015477e-05,
+      "loss": 1.5765,
+      "step": 996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731610805312675e-05,
+      "loss": 1.2137,
+      "step": 997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973084579867561e-05,
+      "loss": 1.6302,
+      "step": 998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730079718188718e-05,
+      "loss": 1.7154,
+      "step": 999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729312563936532e-05,
+      "loss": 1.7901,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728544336003713e-05,
+      "loss": 1.5987,
+      "step": 1001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972777503447504e-05,
+      "loss": 1.5091,
+      "step": 1002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972700465943541e-05,
+      "loss": 1.3889,
+      "step": 1003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972623321096984e-05,
+      "loss": 1.5654,
+      "step": 1004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725460689163455e-05,
+      "loss": 1.7335,
+      "step": 1005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724687094101517e-05,
+      "loss": 1.3067,
+      "step": 1006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723912425869388e-05,
+      "loss": 1.5246,
+      "step": 1007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723136684552563e-05,
+      "loss": 1.5542,
+      "step": 1008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722359870236646e-05,
+      "loss": 1.616,
+      "step": 1009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721581983007362e-05,
+      "loss": 1.4896,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720803022950555e-05,
+      "loss": 1.6223,
+      "step": 1011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972002299015219e-05,
+      "loss": 1.4502,
+      "step": 1012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719241884698346e-05,
+      "loss": 1.8113,
+      "step": 1013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718459706675223e-05,
+      "loss": 1.4859,
+      "step": 1014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717676456169133e-05,
+      "loss": 1.4961,
+      "step": 1015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971689213326652e-05,
+      "loss": 1.3994,
+      "step": 1016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716106738053934e-05,
+      "loss": 1.645,
+      "step": 1017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715320270618047e-05,
+      "loss": 1.9651,
+      "step": 1018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971453273104565e-05,
+      "loss": 1.3647,
+      "step": 1019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713744119423654e-05,
+      "loss": 1.676,
+      "step": 1020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712954435839085e-05,
+      "loss": 1.5601,
+      "step": 1021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971216368037909e-05,
+      "loss": 1.4136,
+      "step": 1022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971137185313093e-05,
+      "loss": 1.6789,
+      "step": 1023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.971057895418199e-05,
+      "loss": 1.7855,
+      "step": 1024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970978498361977e-05,
+      "loss": 1.6091,
+      "step": 1025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9708989941531887e-05,
+      "loss": 1.3707,
+      "step": 1026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9708193828006083e-05,
+      "loss": 2.5062,
+      "step": 1027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970739664313021e-05,
+      "loss": 1.5835,
+      "step": 1028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970659838699224e-05,
+      "loss": 1.6969,
+      "step": 1029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705799059680265e-05,
+      "loss": 1.5362,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704998661282494e-05,
+      "loss": 1.7993,
+      "step": 1031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704197191887263e-05,
+      "loss": 1.5039,
+      "step": 1032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970339465158301e-05,
+      "loss": 1.6583,
+      "step": 1033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702591040458304e-05,
+      "loss": 1.437,
+      "step": 1034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701786358601824e-05,
+      "loss": 1.4915,
+      "step": 1035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700980606102372e-05,
+      "loss": 1.6814,
+      "step": 1036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970017378304887e-05,
+      "loss": 1.4936,
+      "step": 1037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969936588953035e-05,
+      "loss": 1.6603,
+      "step": 1038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698556925635973e-05,
+      "loss": 1.4571,
+      "step": 1039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969774689145501e-05,
+      "loss": 1.0641,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969693578707685e-05,
+      "loss": 1.6064,
+      "step": 1041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696123612591004e-05,
+      "loss": 1.4078,
+      "step": 1042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695310368087103e-05,
+      "loss": 2.2056,
+      "step": 1043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969449605365489e-05,
+      "loss": 1.4134,
+      "step": 1044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969368066938423e-05,
+      "loss": 1.395,
+      "step": 1045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96928642153651e-05,
+      "loss": 1.5353,
+      "step": 1046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969204669168761e-05,
+      "loss": 1.3326,
+      "step": 1047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691228098441968e-05,
+      "loss": 1.5765,
+      "step": 1048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690408435718514e-05,
+      "loss": 1.4238,
+      "step": 1049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96895877036077e-05,
+      "loss": 1.7715,
+      "step": 1050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688765902200106e-05,
+      "loss": 1.9262,
+      "step": 1051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687943031586414e-05,
+      "loss": 1.77,
+      "step": 1052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687119091857435e-05,
+      "loss": 1.8629,
+      "step": 1053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686294083104095e-05,
+      "loss": 1.5802,
+      "step": 1054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685468005417435e-05,
+      "loss": 1.8795,
+      "step": 1055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968464085888862e-05,
+      "loss": 1.6065,
+      "step": 1056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9683812643608934e-05,
+      "loss": 1.2943,
+      "step": 1057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9682983359669764e-05,
+      "loss": 1.1357,
+      "step": 1058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968215300716264e-05,
+      "loss": 1.5997,
+      "step": 1059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968132158617918e-05,
+      "loss": 1.6536,
+      "step": 1060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968048909681115e-05,
+      "loss": 1.565,
+      "step": 1061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679655539150407e-05,
+      "loss": 1.5521,
+      "step": 1062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967882091328895e-05,
+      "loss": 1.4013,
+      "step": 1063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967798521931888e-05,
+      "loss": 1.5993,
+      "step": 1064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677148457332414e-05,
+      "loss": 1.6593,
+      "step": 1065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96763106274219e-05,
+      "loss": 1.9984,
+      "step": 1066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675471729679797e-05,
+      "loss": 1.6479,
+      "step": 1067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674631764198676e-05,
+      "loss": 1.3075,
+      "step": 1068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967379073107124e-05,
+      "loss": 1.3119,
+      "step": 1069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672948630390296e-05,
+      "loss": 1.448,
+      "step": 1070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672105462248778e-05,
+      "loss": 1.5184,
+      "step": 1071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671261226739725e-05,
+      "loss": 1.4492,
+      "step": 1072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967041592395631e-05,
+      "loss": 1.8916,
+      "step": 1073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966956955399182e-05,
+      "loss": 1.4273,
+      "step": 1074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668722116939648e-05,
+      "loss": 1.8058,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966787361289332e-05,
+      "loss": 1.5747,
+      "step": 1076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966702404194647e-05,
+      "loss": 1.4415,
+      "step": 1077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966617340419285e-05,
+      "loss": 1.721,
+      "step": 1078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966532169972634e-05,
+      "loss": 1.7611,
+      "step": 1079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966446892864092e-05,
+      "loss": 2.0739,
+      "step": 1080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663615091030706e-05,
+      "loss": 2.24,
+      "step": 1081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662760186989914e-05,
+      "loss": 1.1867,
+      "step": 1082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96619042166129e-05,
+      "loss": 1.5673,
+      "step": 1083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661047179994117e-05,
+      "loss": 1.2588,
+      "step": 1084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660189077228146e-05,
+      "loss": 1.4905,
+      "step": 1085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659329908409676e-05,
+      "loss": 1.427,
+      "step": 1086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658469673633533e-05,
+      "loss": 1.6082,
+      "step": 1087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657608372994636e-05,
+      "loss": 1.4785,
+      "step": 1088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9656746006588046e-05,
+      "loss": 1.7323,
+      "step": 1089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965588257450892e-05,
+      "loss": 1.4739,
+      "step": 1090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9655018076852546e-05,
+      "loss": 1.4823,
+      "step": 1091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965415251371433e-05,
+      "loss": 1.4584,
+      "step": 1092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653285885189782e-05,
+      "loss": 1.9687,
+      "step": 1093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652418191374546e-05,
+      "loss": 1.5625,
+      "step": 1094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651549432364374e-05,
+      "loss": 1.5861,
+      "step": 1095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965067960825514e-05,
+      "loss": 1.5171,
+      "step": 1096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649808719142832e-05,
+      "loss": 1.1143,
+      "step": 1097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964893676512356e-05,
+      "loss": 1.4926,
+      "step": 1098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964806374629354e-05,
+      "loss": 1.712,
+      "step": 1099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964718966274913e-05,
+      "loss": 2.0818,
+      "step": 1100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646314514586776e-05,
+      "loss": 1.6519,
+      "step": 1101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645438301903058e-05,
+      "loss": 1.6655,
+      "step": 1102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644561024794672e-05,
+      "loss": 1.3422,
+      "step": 1103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964368268335843e-05,
+      "loss": 1.6276,
+      "step": 1104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642803277691264e-05,
+      "loss": 1.2556,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964192280789022e-05,
+      "loss": 1.4743,
+      "step": 1106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641041274052458e-05,
+      "loss": 2.1234,
+      "step": 1107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640158676275266e-05,
+      "loss": 1.3514,
+      "step": 1108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963927501465604e-05,
+      "loss": 1.6848,
+      "step": 1109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638390289292296e-05,
+      "loss": 1.5636,
+      "step": 1110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963750450028167e-05,
+      "loss": 1.5743,
+      "step": 1111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963661764772191e-05,
+      "loss": 1.6523,
+      "step": 1112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963572973171089e-05,
+      "loss": 1.5836,
+      "step": 1113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634840752346597e-05,
+      "loss": 1.4872,
+      "step": 1114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633950709727126e-05,
+      "loss": 1.6318,
+      "step": 1115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633059603950705e-05,
+      "loss": 1.2803,
+      "step": 1116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632167435115673e-05,
+      "loss": 1.5388,
+      "step": 1117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963127420332048e-05,
+      "loss": 1.5516,
+      "step": 1118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96303799086637e-05,
+      "loss": 1.3612,
+      "step": 1119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962948455124403e-05,
+      "loss": 1.2257,
+      "step": 1120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628588131160268e-05,
+      "loss": 1.3626,
+      "step": 1121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9627690648511343e-05,
+      "loss": 1.5637,
+      "step": 1122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96267921033963e-05,
+      "loss": 1.3794,
+      "step": 1123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962589249591429e-05,
+      "loss": 1.6158,
+      "step": 1124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9624991826164598e-05,
+      "loss": 1.5757,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962409009424661e-05,
+      "loss": 1.3344,
+      "step": 1126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623187300259847e-05,
+      "loss": 1.5996,
+      "step": 1127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962228344430393e-05,
+      "loss": 1.6873,
+      "step": 1128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96213785264786e-05,
+      "loss": 1.5641,
+      "step": 1129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962047254688373e-05,
+      "loss": 1.4052,
+      "step": 1130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619565505619288e-05,
+      "loss": 1.7827,
+      "step": 1131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961865740278538e-05,
+      "loss": 1.7826,
+      "step": 1132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617748238482216e-05,
+      "loss": 1.7037,
+      "step": 1133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616838012810132e-05,
+      "loss": 1.6531,
+      "step": 1134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615926725869564e-05,
+      "loss": 1.3687,
+      "step": 1135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961501437776109e-05,
+      "loss": 1.7377,
+      "step": 1136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961410096858539e-05,
+      "loss": 1.4679,
+      "step": 1137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961318649844326e-05,
+      "loss": 1.5847,
+      "step": 1138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612270967435614e-05,
+      "loss": 1.5051,
+      "step": 1139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961135437566349e-05,
+      "loss": 1.4465,
+      "step": 1140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961043672322804e-05,
+      "loss": 1.5313,
+      "step": 1141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960951801023053e-05,
+      "loss": 2.0989,
+      "step": 1142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608598236772345e-05,
+      "loss": 1.5488,
+      "step": 1143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607677402954985e-05,
+      "loss": 1.6308,
+      "step": 1144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960675550888007e-05,
+      "loss": 1.4625,
+      "step": 1145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960583255464934e-05,
+      "loss": 1.6936,
+      "step": 1146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960490854036464e-05,
+      "loss": 1.2786,
+      "step": 1147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603983466127944e-05,
+      "loss": 1.7147,
+      "step": 1148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960305733204134e-05,
+      "loss": 1.5988,
+      "step": 1149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602130138207027e-05,
+      "loss": 1.4609,
+      "step": 1150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601201884727334e-05,
+      "loss": 1.7103,
+      "step": 1151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600272571704687e-05,
+      "loss": 1.9191,
+      "step": 1152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599342199241653e-05,
+      "loss": 1.518,
+      "step": 1153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9598410767440896e-05,
+      "loss": 1.6047,
+      "step": 1154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9597478276405207e-05,
+      "loss": 1.5822,
+      "step": 1155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959654472623749e-05,
+      "loss": 1.2884,
+      "step": 1156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9595610117040766e-05,
+      "loss": 1.7547,
+      "step": 1157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594674448918175e-05,
+      "loss": 1.7084,
+      "step": 1158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593737721972977e-05,
+      "loss": 1.5788,
+      "step": 1159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592799936308536e-05,
+      "loss": 1.693,
+      "step": 1160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591861092028352e-05,
+      "loss": 1.6187,
+      "step": 1161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590921189236024e-05,
+      "loss": 1.4663,
+      "step": 1162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589980228035275e-05,
+      "loss": 1.4422,
+      "step": 1163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958903820852995e-05,
+      "loss": 1.6765,
+      "step": 1164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588095130824004e-05,
+      "loss": 1.4054,
+      "step": 1165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587150995021507e-05,
+      "loss": 1.4088,
+      "step": 1166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586205801226655e-05,
+      "loss": 1.8127,
+      "step": 1167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585259549543746e-05,
+      "loss": 1.6792,
+      "step": 1168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584312240077216e-05,
+      "loss": 1.1408,
+      "step": 1169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583363872931595e-05,
+      "loss": 1.8863,
+      "step": 1170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958241444821155e-05,
+      "loss": 1.2526,
+      "step": 1171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581463966021846e-05,
+      "loss": 1.3491,
+      "step": 1172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580512426467377e-05,
+      "loss": 1.6633,
+      "step": 1173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579559829653152e-05,
+      "loss": 1.6026,
+      "step": 1174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578606175684295e-05,
+      "loss": 1.8926,
+      "step": 1175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577651464666042e-05,
+      "loss": 1.4405,
+      "step": 1176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576695696703757e-05,
+      "loss": 1.61,
+      "step": 1177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575738871902912e-05,
+      "loss": 1.7417,
+      "step": 1178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574780990369092e-05,
+      "loss": 1.8399,
+      "step": 1179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573822052208013e-05,
+      "loss": 1.3756,
+      "step": 1180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572862057525493e-05,
+      "loss": 1.7429,
+      "step": 1181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571901006427473e-05,
+      "loss": 1.576,
+      "step": 1182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570938899020014e-05,
+      "loss": 1.5815,
+      "step": 1183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569975735409284e-05,
+      "loss": 1.9448,
+      "step": 1184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956901151570158e-05,
+      "loss": 1.3852,
+      "step": 1185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95680462400033e-05,
+      "loss": 1.5793,
+      "step": 1186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9567079908420973e-05,
+      "loss": 1.7824,
+      "step": 1187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9566112521061237e-05,
+      "loss": 1.5838,
+      "step": 1188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956514407803085e-05,
+      "loss": 1.5861,
+      "step": 1189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9564174579436686e-05,
+      "loss": 2.0089,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9563204025385727e-05,
+      "loss": 1.8724,
+      "step": 1191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562232415985085e-05,
+      "loss": 1.778,
+      "step": 1192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561259751341985e-05,
+      "loss": 1.8669,
+      "step": 1193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560286031563756e-05,
+      "loss": 2.0309,
+      "step": 1194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955931125675786e-05,
+      "loss": 1.2761,
+      "step": 1195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558335427031868e-05,
+      "loss": 2.1443,
+      "step": 1196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557358542493468e-05,
+      "loss": 1.6923,
+      "step": 1197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955638060325046e-05,
+      "loss": 1.9606,
+      "step": 1198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555401609410768e-05,
+      "loss": 1.6625,
+      "step": 1199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955442156108243e-05,
+      "loss": 1.5586,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95534404583736e-05,
+      "loss": 1.5777,
+      "step": 1201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552458301392545e-05,
+      "loss": 1.9367,
+      "step": 1202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551475090247653e-05,
+      "loss": 1.5591,
+      "step": 1203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955049082504743e-05,
+      "loss": 1.4638,
+      "step": 1204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549505505900487e-05,
+      "loss": 1.5068,
+      "step": 1205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954851913291556e-05,
+      "loss": 1.5622,
+      "step": 1206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954753170620151e-05,
+      "loss": 1.1746,
+      "step": 1207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546543225867293e-05,
+      "loss": 1.7821,
+      "step": 1208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545553692021997e-05,
+      "loss": 1.0941,
+      "step": 1209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544563104774826e-05,
+      "loss": 1.2762,
+      "step": 1210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543571464235093e-05,
+      "loss": 1.514,
+      "step": 1211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542578770512236e-05,
+      "loss": 1.7265,
+      "step": 1212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541585023715793e-05,
+      "loss": 1.6298,
+      "step": 1213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540590223955437e-05,
+      "loss": 1.4474,
+      "step": 1214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953959437134095e-05,
+      "loss": 1.5375,
+      "step": 1215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953859746598223e-05,
+      "loss": 1.68,
+      "step": 1216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537599507989285e-05,
+      "loss": 1.7475,
+      "step": 1217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953660049747225e-05,
+      "loss": 1.6126,
+      "step": 1218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9535600434541366e-05,
+      "loss": 1.6455,
+      "step": 1219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9534599319307e-05,
+      "loss": 1.5731,
+      "step": 1220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9533597151879632e-05,
+      "loss": 1.3296,
+      "step": 1221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953259393236985e-05,
+      "loss": 1.4569,
+      "step": 1222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9531589660888368e-05,
+      "loss": 1.9065,
+      "step": 1223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530584337546013e-05,
+      "loss": 1.7631,
+      "step": 1224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529577962453726e-05,
+      "loss": 1.6137,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528570535722568e-05,
+      "loss": 1.4588,
+      "step": 1226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952756205746371e-05,
+      "loss": 1.4943,
+      "step": 1227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952655252778845e-05,
+      "loss": 1.8234,
+      "step": 1228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525541946808187e-05,
+      "loss": 1.2081,
+      "step": 1229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524530314634452e-05,
+      "loss": 1.5031,
+      "step": 1230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523517631378874e-05,
+      "loss": 2.3358,
+      "step": 1231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522503897153215e-05,
+      "loss": 1.4265,
+      "step": 1232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521489112069346e-05,
+      "loss": 1.2834,
+      "step": 1233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952047327623925e-05,
+      "loss": 1.745,
+      "step": 1234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951945638977503e-05,
+      "loss": 1.6436,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951843845278891e-05,
+      "loss": 1.6233,
+      "step": 1236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951741946539322e-05,
+      "loss": 1.554,
+      "step": 1237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951639942770041e-05,
+      "loss": 1.5111,
+      "step": 1238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515378339823048e-05,
+      "loss": 1.613,
+      "step": 1239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514356201873817e-05,
+      "loss": 1.6253,
+      "step": 1240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513333013965515e-05,
+      "loss": 1.5315,
+      "step": 1241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512308776211057e-05,
+      "loss": 1.6687,
+      "step": 1242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511283488723473e-05,
+      "loss": 1.8454,
+      "step": 1243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510257151615907e-05,
+      "loss": 1.8468,
+      "step": 1244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950922976500162e-05,
+      "loss": 1.8495,
+      "step": 1245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508201328993994e-05,
+      "loss": 1.5716,
+      "step": 1246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507171843706516e-05,
+      "loss": 1.3678,
+      "step": 1247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506141309252798e-05,
+      "loss": 1.5236,
+      "step": 1248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950510972574657e-05,
+      "loss": 1.5548,
+      "step": 1249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504077093301664e-05,
+      "loss": 1.6543,
+      "step": 1250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503043412032042e-05,
+      "loss": 1.3293,
+      "step": 1251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9502008682051775e-05,
+      "loss": 1.9636,
+      "step": 1252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.950097290347505e-05,
+      "loss": 1.6849,
+      "step": 1253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9499936076416173e-05,
+      "loss": 1.8355,
+      "step": 1254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9498898200989556e-05,
+      "loss": 1.6558,
+      "step": 1255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497859277309744e-05,
+      "loss": 1.4871,
+      "step": 1256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496819305491382e-05,
+      "loss": 1.9324,
+      "step": 1257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495778285649238e-05,
+      "loss": 1.5627,
+      "step": 1258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494736217898192e-05,
+      "loss": 1.3645,
+      "step": 1259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493693102353242e-05,
+      "loss": 1.2276,
+      "step": 1260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492648939129503e-05,
+      "loss": 1.3715,
+      "step": 1261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491603728342205e-05,
+      "loss": 1.5702,
+      "step": 1262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949055747010669e-05,
+      "loss": 1.4332,
+      "step": 1263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489510164538418e-05,
+      "loss": 1.789,
+      "step": 1264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488461811752963e-05,
+      "loss": 1.2764,
+      "step": 1265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948741241186602e-05,
+      "loss": 1.7632,
+      "step": 1266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486361964993395e-05,
+      "loss": 1.5738,
+      "step": 1267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485310471251007e-05,
+      "loss": 1.5522,
+      "step": 1268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94842579307549e-05,
+      "loss": 1.646,
+      "step": 1269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483204343621227e-05,
+      "loss": 1.2073,
+      "step": 1270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482149709966247e-05,
+      "loss": 1.9703,
+      "step": 1271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481094029906354e-05,
+      "loss": 1.4531,
+      "step": 1272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480037303558046e-05,
+      "loss": 1.7103,
+      "step": 1273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478979531037935e-05,
+      "loss": 1.3165,
+      "step": 1274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477920712462756e-05,
+      "loss": 1.3876,
+      "step": 1275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476860847949353e-05,
+      "loss": 1.7891,
+      "step": 1276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475799937614686e-05,
+      "loss": 1.6432,
+      "step": 1277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474737981575833e-05,
+      "loss": 1.6436,
+      "step": 1278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947367497994999e-05,
+      "loss": 1.2197,
+      "step": 1279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947261093285446e-05,
+      "loss": 1.8621,
+      "step": 1280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947154584040667e-05,
+      "loss": 1.409,
+      "step": 1281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470479702724153e-05,
+      "loss": 1.3794,
+      "step": 1282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469412519924567e-05,
+      "loss": 1.85,
+      "step": 1283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9468344292125683e-05,
+      "loss": 1.5193,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9467275019445387e-05,
+      "loss": 1.4031,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9466204702001667e-05,
+      "loss": 1.5341,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9465133339912653e-05,
+      "loss": 1.6988,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9464060933296564e-05,
+      "loss": 1.3939,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462987482271752e-05,
+      "loss": 1.3957,
+      "step": 1289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461912986956678e-05,
+      "loss": 1.9203,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460837447469914e-05,
+      "loss": 1.5189,
+      "step": 1291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459760863930154e-05,
+      "loss": 1.5333,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458683236456207e-05,
+      "loss": 1.5602,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457604565166992e-05,
+      "loss": 1.8659,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456524850181546e-05,
+      "loss": 1.913,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945544409161902e-05,
+      "loss": 1.456,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454362289598685e-05,
+      "loss": 1.8041,
+      "step": 1297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453279444239922e-05,
+      "loss": 1.6039,
+      "step": 1298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452195555662226e-05,
+      "loss": 1.4683,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451110623985213e-05,
+      "loss": 1.5392,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945002464932861e-05,
+      "loss": 1.6594,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944893763181226e-05,
+      "loss": 1.5885,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447849571556123e-05,
+      "loss": 1.6409,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446760468680262e-05,
+      "loss": 1.5856,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944567032330488e-05,
+      "loss": 1.2933,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444579135550275e-05,
+      "loss": 1.7758,
+      "step": 1306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443486905536862e-05,
+      "loss": 1.4447,
+      "step": 1307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442393633385176e-05,
+      "loss": 1.393,
+      "step": 1308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441299319215866e-05,
+      "loss": 1.6759,
+      "step": 1309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944020396314969e-05,
+      "loss": 1.4087,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439107565307537e-05,
+      "loss": 2.1283,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438010125810393e-05,
+      "loss": 1.9173,
+      "step": 1312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436911644779366e-05,
+      "loss": 1.2146,
+      "step": 1313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435812122335682e-05,
+      "loss": 1.5997,
+      "step": 1314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434711558600677e-05,
+      "loss": 1.4679,
+      "step": 1315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433609953695806e-05,
+      "loss": 1.3706,
+      "step": 1316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9432507307742635e-05,
+      "loss": 1.4724,
+      "step": 1317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9431403620862845e-05,
+      "loss": 1.6537,
+      "step": 1318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.943029889317824e-05,
+      "loss": 1.2515,
+      "step": 1319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9429193124810726e-05,
+      "loss": 1.8474,
+      "step": 1320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9428086315882333e-05,
+      "loss": 1.482,
+      "step": 1321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426978466515202e-05,
+      "loss": 1.5052,
+      "step": 1322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942586957683159e-05,
+      "loss": 2.047,
+      "step": 1323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424759646953874e-05,
+      "loss": 1.2067,
+      "step": 1324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423648677004533e-05,
+      "loss": 1.4932,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422536667106172e-05,
+      "loss": 1.4932,
+      "step": 1326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942142361738151e-05,
+      "loss": 1.6003,
+      "step": 1327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420309527953372e-05,
+      "loss": 1.427,
+      "step": 1328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419194398944704e-05,
+      "loss": 1.7069,
+      "step": 1329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418078230478572e-05,
+      "loss": 1.2928,
+      "step": 1330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416961022678146e-05,
+      "loss": 1.7936,
+      "step": 1331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415842775666718e-05,
+      "loss": 1.6358,
+      "step": 1332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941472348956769e-05,
+      "loss": 1.6361,
+      "step": 1333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413603164504588e-05,
+      "loss": 1.4203,
+      "step": 1334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412481800601036e-05,
+      "loss": 2.7013,
+      "step": 1335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411359397980788e-05,
+      "loss": 1.6016,
+      "step": 1336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410235956767707e-05,
+      "loss": 1.7697,
+      "step": 1337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409111477085768e-05,
+      "loss": 1.9262,
+      "step": 1338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407985959059065e-05,
+      "loss": 1.7238,
+      "step": 1339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406859402811807e-05,
+      "loss": 1.6324,
+      "step": 1340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940573180846831e-05,
+      "loss": 1.5553,
+      "step": 1341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404603176153016e-05,
+      "loss": 1.7421,
+      "step": 1342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940347350599047e-05,
+      "loss": 1.4479,
+      "step": 1343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402342798105343e-05,
+      "loss": 1.4682,
+      "step": 1344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940121105262241e-05,
+      "loss": 1.3522,
+      "step": 1345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400078269666563e-05,
+      "loss": 1.5069,
+      "step": 1346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398944449362815e-05,
+      "loss": 1.6031,
+      "step": 1347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397809591836286e-05,
+      "loss": 1.3678,
+      "step": 1348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939667369721222e-05,
+      "loss": 1.4889,
+      "step": 1349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939553676561596e-05,
+      "loss": 2.1843,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9394398797172976e-05,
+      "loss": 1.8268,
+      "step": 1351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939325979200885e-05,
+      "loss": 1.5486,
+      "step": 1352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9392119750249278e-05,
+      "loss": 1.4774,
+      "step": 1353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390978672020068e-05,
+      "loss": 1.7902,
+      "step": 1354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389836557447145e-05,
+      "loss": 1.1299,
+      "step": 1355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388693406656544e-05,
+      "loss": 1.479,
+      "step": 1356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938754921977442e-05,
+      "loss": 1.2697,
+      "step": 1357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386403996927038e-05,
+      "loss": 1.6257,
+      "step": 1358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385257738240786e-05,
+      "loss": 1.9502,
+      "step": 1359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938411044384215e-05,
+      "loss": 1.7573,
+      "step": 1360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382962113857752e-05,
+      "loss": 1.7891,
+      "step": 1361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93818127484143e-05,
+      "loss": 1.3793,
+      "step": 1362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380662347638647e-05,
+      "loss": 1.5344,
+      "step": 1363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379510911657743e-05,
+      "loss": 1.6544,
+      "step": 1364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378358440598647e-05,
+      "loss": 1.4653,
+      "step": 1365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377204934588547e-05,
+      "loss": 1.5614,
+      "step": 1366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376050393754737e-05,
+      "loss": 1.3953,
+      "step": 1367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374894818224628e-05,
+      "loss": 1.6852,
+      "step": 1368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937373820812574e-05,
+      "loss": 1.6513,
+      "step": 1369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372580563585718e-05,
+      "loss": 1.4892,
+      "step": 1370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371421884732306e-05,
+      "loss": 1.557,
+      "step": 1371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370262171693378e-05,
+      "loss": 1.5442,
+      "step": 1372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369101424596905e-05,
+      "loss": 1.3792,
+      "step": 1373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936793964357099e-05,
+      "loss": 1.4564,
+      "step": 1374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366776828743836e-05,
+      "loss": 1.8641,
+      "step": 1375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936561298024377e-05,
+      "loss": 1.1728,
+      "step": 1376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936444809819923e-05,
+      "loss": 1.3855,
+      "step": 1377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363282182738755e-05,
+      "loss": 2.2012,
+      "step": 1378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362115233991025e-05,
+      "loss": 1.3072,
+      "step": 1379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360947252084813e-05,
+      "loss": 1.3182,
+      "step": 1380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359778237149012e-05,
+      "loss": 1.8531,
+      "step": 1381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9358608189312626e-05,
+      "loss": 1.4491,
+      "step": 1382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9357437108704776e-05,
+      "loss": 1.6338,
+      "step": 1383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9356264995454707e-05,
+      "loss": 1.5352,
+      "step": 1384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9355091849691752e-05,
+      "loss": 1.5187,
+      "step": 1385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353917671545387e-05,
+      "loss": 1.7736,
+      "step": 1386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935274246114518e-05,
+      "loss": 1.4288,
+      "step": 1387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351566218620826e-05,
+      "loss": 1.5607,
+      "step": 1388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935038894410213e-05,
+      "loss": 1.433,
+      "step": 1389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934921063771901e-05,
+      "loss": 1.378,
+      "step": 1390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348031299601496e-05,
+      "loss": 1.4436,
+      "step": 1391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346850929879738e-05,
+      "loss": 1.4356,
+      "step": 1392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934566952868399e-05,
+      "loss": 1.2784,
+      "step": 1393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934448709614463e-05,
+      "loss": 1.5865,
+      "step": 1394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343303632392145e-05,
+      "loss": 1.6043,
+      "step": 1395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934211913755714e-05,
+      "loss": 1.4179,
+      "step": 1396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340933611770322e-05,
+      "loss": 1.4867,
+      "step": 1397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933974705516253e-05,
+      "loss": 1.3738,
+      "step": 1398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338559467864696e-05,
+      "loss": 1.3986,
+      "step": 1399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337370850007883e-05,
+      "loss": 1.5827,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336181201723262e-05,
+      "loss": 1.2667,
+      "step": 1401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334990523142113e-05,
+      "loss": 1.5857,
+      "step": 1402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333798814395835e-05,
+      "loss": 1.4963,
+      "step": 1403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933260607561594e-05,
+      "loss": 2.3949,
+      "step": 1404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331412306934056e-05,
+      "loss": 1.4026,
+      "step": 1405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330217508481915e-05,
+      "loss": 1.7998,
+      "step": 1406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329021680391373e-05,
+      "loss": 1.4854,
+      "step": 1407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93278248227944e-05,
+      "loss": 1.3448,
+      "step": 1408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326626935823064e-05,
+      "loss": 1.7777,
+      "step": 1409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932542801960957e-05,
+      "loss": 1.691,
+      "step": 1410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324228074286223e-05,
+      "loss": 1.8785,
+      "step": 1411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323027099985438e-05,
+      "loss": 1.527,
+      "step": 1412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932182509683975e-05,
+      "loss": 1.5334,
+      "step": 1413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9320622064981802e-05,
+      "loss": 1.9911,
+      "step": 1414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931941800454437e-05,
+      "loss": 2.3023,
+      "step": 1415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931821291566031e-05,
+      "loss": 1.1791,
+      "step": 1416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9317006798462622e-05,
+      "loss": 1.4171,
+      "step": 1417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9315799653084404e-05,
+      "loss": 1.3817,
+      "step": 1418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314591479658873e-05,
+      "loss": 1.8823,
+      "step": 1419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313382278319354e-05,
+      "loss": 1.8818,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312172049199287e-05,
+      "loss": 1.3905,
+      "step": 1421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931096079243223e-05,
+      "loss": 1.676,
+      "step": 1422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309748508151857e-05,
+      "loss": 1.3387,
+      "step": 1423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308535196491937e-05,
+      "loss": 1.4387,
+      "step": 1424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307320857586377e-05,
+      "loss": 1.7474,
+      "step": 1425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930610549156918e-05,
+      "loss": 1.4893,
+      "step": 1426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304889098574468e-05,
+      "loss": 1.4952,
+      "step": 1427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930367167873648e-05,
+      "loss": 1.4663,
+      "step": 1428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930245323218956e-05,
+      "loss": 1.6334,
+      "step": 1429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301233759068174e-05,
+      "loss": 1.852,
+      "step": 1430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930001325950689e-05,
+      "loss": 1.6705,
+      "step": 1431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298791733640407e-05,
+      "loss": 1.4602,
+      "step": 1432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929756918160352e-05,
+      "loss": 1.7435,
+      "step": 1433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296345603531148e-05,
+      "loss": 1.574,
+      "step": 1434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295120999558318e-05,
+      "loss": 1.2814,
+      "step": 1435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293895369820166e-05,
+      "loss": 1.3552,
+      "step": 1436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292668714451954e-05,
+      "loss": 1.7019,
+      "step": 1437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291441033589046e-05,
+      "loss": 1.6545,
+      "step": 1438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290212327366924e-05,
+      "loss": 1.3458,
+      "step": 1439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288982595921186e-05,
+      "loss": 1.9166,
+      "step": 1440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928775183938753e-05,
+      "loss": 1.715,
+      "step": 1441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286520057901784e-05,
+      "loss": 1.2953,
+      "step": 1442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928528725159988e-05,
+      "loss": 2.1198,
+      "step": 1443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284053420617865e-05,
+      "loss": 1.7125,
+      "step": 1444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282818565091895e-05,
+      "loss": 1.5466,
+      "step": 1445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281582685158245e-05,
+      "loss": 1.5016,
+      "step": 1446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9280345780953304e-05,
+      "loss": 1.6314,
+      "step": 1447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927910785261357e-05,
+      "loss": 1.4749,
+      "step": 1448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927786890027565e-05,
+      "loss": 1.5185,
+      "step": 1449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927662892407627e-05,
+      "loss": 1.6101,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275387924152274e-05,
+      "loss": 2.0385,
+      "step": 1451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274145900640608e-05,
+      "loss": 1.6083,
+      "step": 1452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272902853678336e-05,
+      "loss": 1.8493,
+      "step": 1453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271658783402635e-05,
+      "loss": 1.4683,
+      "step": 1454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270413689950796e-05,
+      "loss": 1.3983,
+      "step": 1455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926916757346022e-05,
+      "loss": 1.5149,
+      "step": 1456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926792043406842e-05,
+      "loss": 1.4105,
+      "step": 1457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266672271913032e-05,
+      "loss": 2.2278,
+      "step": 1458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265423087131792e-05,
+      "loss": 2.0177,
+      "step": 1459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264172879862553e-05,
+      "loss": 1.3989,
+      "step": 1460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262921650243286e-05,
+      "loss": 2.0277,
+      "step": 1461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261669398412064e-05,
+      "loss": 1.6796,
+      "step": 1462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926041612450709e-05,
+      "loss": 1.4095,
+      "step": 1463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925916182866666e-05,
+      "loss": 1.4654,
+      "step": 1464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257906511029194e-05,
+      "loss": 1.2926,
+      "step": 1465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256650171733228e-05,
+      "loss": 1.5152,
+      "step": 1466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92553928109174e-05,
+      "loss": 1.5577,
+      "step": 1467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254134428720467e-05,
+      "loss": 1.3837,
+      "step": 1468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252875025281302e-05,
+      "loss": 1.4735,
+      "step": 1469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251614600738882e-05,
+      "loss": 1.4362,
+      "step": 1470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250353155232306e-05,
+      "loss": 1.6517,
+      "step": 1471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249090688900777e-05,
+      "loss": 1.2006,
+      "step": 1472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247827201883614e-05,
+      "loss": 1.4126,
+      "step": 1473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246562694320258e-05,
+      "loss": 1.5349,
+      "step": 1474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245297166350242e-05,
+      "loss": 1.6034,
+      "step": 1475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244030618113233e-05,
+      "loss": 1.7065,
+      "step": 1476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242763049748996e-05,
+      "loss": 1.5156,
+      "step": 1477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241494461397416e-05,
+      "loss": 1.6283,
+      "step": 1478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.924022485319849e-05,
+      "loss": 2.0029,
+      "step": 1479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9238954225292323e-05,
+      "loss": 1.554,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9237682577819135e-05,
+      "loss": 1.6694,
+      "step": 1481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923640991091926e-05,
+      "loss": 1.5614,
+      "step": 1482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9235136224733145e-05,
+      "loss": 1.9843,
+      "step": 1483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233861519401348e-05,
+      "loss": 1.6994,
+      "step": 1484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232585795064537e-05,
+      "loss": 1.448,
+      "step": 1485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231309051863497e-05,
+      "loss": 1.7599,
+      "step": 1486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230031289939122e-05,
+      "loss": 1.3568,
+      "step": 1487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228752509432418e-05,
+      "loss": 1.675,
+      "step": 1488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922747271048451e-05,
+      "loss": 1.8124,
+      "step": 1489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226191893236627e-05,
+      "loss": 1.3684,
+      "step": 1490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224910057830114e-05,
+      "loss": 1.4869,
+      "step": 1491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922362720440643e-05,
+      "loss": 1.6966,
+      "step": 1492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922234333310714e-05,
+      "loss": 2.1281,
+      "step": 1493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221058444073933e-05,
+      "loss": 1.5291,
+      "step": 1494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219772537448597e-05,
+      "loss": 2.516,
+      "step": 1495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218485613373042e-05,
+      "loss": 1.7147,
+      "step": 1496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217197671989287e-05,
+      "loss": 1.8074,
+      "step": 1497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215908713439462e-05,
+      "loss": 1.5204,
+      "step": 1498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921461873786581e-05,
+      "loss": 1.6076,
+      "step": 1499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921332774541069e-05,
+      "loss": 1.3137,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921203573621656e-05,
+      "loss": 1.5474,
+      "step": 1501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921074271042601e-05,
+      "loss": 1.5661,
+      "step": 1502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209448668181728e-05,
+      "loss": 1.5007,
+      "step": 1503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208153609626524e-05,
+      "loss": 1.4773,
+      "step": 1504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206857534903305e-05,
+      "loss": 1.6471,
+      "step": 1505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205560444155104e-05,
+      "loss": 1.5024,
+      "step": 1506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204262337525067e-05,
+      "loss": 1.6285,
+      "step": 1507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202963215156436e-05,
+      "loss": 1.6574,
+      "step": 1508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201663077192587e-05,
+      "loss": 1.5574,
+      "step": 1509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920036192377699e-05,
+      "loss": 1.3173,
+      "step": 1510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199059755053235e-05,
+      "loss": 1.8451,
+      "step": 1511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9197756571165024e-05,
+      "loss": 1.5271,
+      "step": 1512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919645237225617e-05,
+      "loss": 1.2631,
+      "step": 1513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9195147158470605e-05,
+      "loss": 1.5358,
+      "step": 1514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193840929952352e-05,
+      "loss": 1.5679,
+      "step": 1515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9192533686845573e-05,
+      "loss": 1.552,
+      "step": 1516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919122542929452e-05,
+      "loss": 1.6163,
+      "step": 1517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918991615744357e-05,
+      "loss": 1.6288,
+      "step": 1518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188605871437215e-05,
+      "loss": 1.3906,
+      "step": 1519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918729457142004e-05,
+      "loss": 1.3406,
+      "step": 1520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185982257536756e-05,
+      "loss": 1.672,
+      "step": 1521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184668929932193e-05,
+      "loss": 1.5774,
+      "step": 1522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183354588751274e-05,
+      "loss": 1.3945,
+      "step": 1523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182039234139047e-05,
+      "loss": 1.3141,
+      "step": 1524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180722866240667e-05,
+      "loss": 2.1583,
+      "step": 1525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179405485201406e-05,
+      "loss": 1.8481,
+      "step": 1526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178087091166637e-05,
+      "loss": 1.5741,
+      "step": 1527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176767684281857e-05,
+      "loss": 1.8283,
+      "step": 1528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917544726469267e-05,
+      "loss": 1.3477,
+      "step": 1529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917412583254479e-05,
+      "loss": 1.9788,
+      "step": 1530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917280338798404e-05,
+      "loss": 1.5168,
+      "step": 1531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171479931156363e-05,
+      "loss": 1.6432,
+      "step": 1532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917015546220781e-05,
+      "loss": 1.6146,
+      "step": 1533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916882998128454e-05,
+      "loss": 1.4188,
+      "step": 1534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916750348853283e-05,
+      "loss": 1.7988,
+      "step": 1535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166175984099058e-05,
+      "loss": 1.7452,
+      "step": 1536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916484746812973e-05,
+      "loss": 1.6629,
+      "step": 1537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163517940771455e-05,
+      "loss": 1.2385,
+      "step": 1538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162187402170947e-05,
+      "loss": 2.0478,
+      "step": 1539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160855852475037e-05,
+      "loss": 1.4833,
+      "step": 1540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915952329183068e-05,
+      "loss": 1.6484,
+      "step": 1541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158189720384915e-05,
+      "loss": 1.4908,
+      "step": 1542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915685513828492e-05,
+      "loss": 1.6597,
+      "step": 1543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9155519545677973e-05,
+      "loss": 1.584,
+      "step": 1544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9154182942711456e-05,
+      "loss": 1.6221,
+      "step": 1545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.915284532953288e-05,
+      "loss": 1.7204,
+      "step": 1546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9151506706289845e-05,
+      "loss": 1.6577,
+      "step": 1547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.915016707313009e-05,
+      "loss": 1.5673,
+      "step": 1548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914882643020144e-05,
+      "loss": 1.4592,
+      "step": 1549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147484777651844e-05,
+      "loss": 1.5522,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914614211562936e-05,
+      "loss": 1.6295,
+      "step": 1551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914479844428216e-05,
+      "loss": 1.9419,
+      "step": 1552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143453763758525e-05,
+      "loss": 1.5917,
+      "step": 1553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914210807420685e-05,
+      "loss": 1.5507,
+      "step": 1554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140761375775628e-05,
+      "loss": 1.47,
+      "step": 1555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139413668613487e-05,
+      "loss": 1.6408,
+      "step": 1556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138064952869152e-05,
+      "loss": 1.8284,
+      "step": 1557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136715228691452e-05,
+      "loss": 1.6223,
+      "step": 1558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135364496229344e-05,
+      "loss": 1.56,
+      "step": 1559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134012755631885e-05,
+      "loss": 1.3805,
+      "step": 1560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132660007048244e-05,
+      "loss": 1.323,
+      "step": 1561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131306250627712e-05,
+      "loss": 2.3587,
+      "step": 1562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912995148651968e-05,
+      "loss": 1.4096,
+      "step": 1563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128595714873647e-05,
+      "loss": 1.5971,
+      "step": 1564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127238935839235e-05,
+      "loss": 1.5911,
+      "step": 1565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125881149566175e-05,
+      "loss": 1.5496,
+      "step": 1566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91245223562043e-05,
+      "loss": 1.4526,
+      "step": 1567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123162555903562e-05,
+      "loss": 1.5556,
+      "step": 1568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121801748814022e-05,
+      "loss": 1.4438,
+      "step": 1569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912043993508585e-05,
+      "loss": 1.6516,
+      "step": 1570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119077114869335e-05,
+      "loss": 1.6378,
+      "step": 1571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117713288314864e-05,
+      "loss": 2.125,
+      "step": 1572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116348455572947e-05,
+      "loss": 1.4848,
+      "step": 1573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114982616794203e-05,
+      "loss": 1.5383,
+      "step": 1574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113615772129353e-05,
+      "loss": 2.3136,
+      "step": 1575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911224792172924e-05,
+      "loss": 1.8092,
+      "step": 1576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911087906574481e-05,
+      "loss": 1.5053,
+      "step": 1577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9109509204327127e-05,
+      "loss": 1.4599,
+      "step": 1578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910813833762736e-05,
+      "loss": 1.8798,
+      "step": 1579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9106766465796792e-05,
+      "loss": 1.1541,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910539358898682e-05,
+      "loss": 1.5757,
+      "step": 1581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910401970734894e-05,
+      "loss": 1.3689,
+      "step": 1582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102644821034775e-05,
+      "loss": 1.3111,
+      "step": 1583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101268930196047e-05,
+      "loss": 1.561,
+      "step": 1584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909989203498459e-05,
+      "loss": 1.6702,
+      "step": 1585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098514135552358e-05,
+      "loss": 1.4787,
+      "step": 1586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097135232051404e-05,
+      "loss": 1.3234,
+      "step": 1587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095755324633903e-05,
+      "loss": 1.4781,
+      "step": 1588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909437441345213e-05,
+      "loss": 1.6842,
+      "step": 1589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092992498658477e-05,
+      "loss": 1.6463,
+      "step": 1590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091609580405444e-05,
+      "loss": 1.6124,
+      "step": 1591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909022565884565e-05,
+      "loss": 1.629,
+      "step": 1592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908884073413181e-05,
+      "loss": 1.5879,
+      "step": 1593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908745480641676e-05,
+      "loss": 1.5884,
+      "step": 1594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086067875853446e-05,
+      "loss": 1.4609,
+      "step": 1595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084679942594923e-05,
+      "loss": 1.9016,
+      "step": 1596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083291006794356e-05,
+      "loss": 1.6295,
+      "step": 1597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908190106860502e-05,
+      "loss": 1.5315,
+      "step": 1598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080510128180305e-05,
+      "loss": 1.5215,
+      "step": 1599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079118185673706e-05,
+      "loss": 1.2947,
+      "step": 1600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077725241238833e-05,
+      "loss": 1.5168,
+      "step": 1601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076331295029403e-05,
+      "loss": 1.0744,
+      "step": 1602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074936347199247e-05,
+      "loss": 1.7033,
+      "step": 1603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90735403979023e-05,
+      "loss": 1.3722,
+      "step": 1604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907214344729262e-05,
+      "loss": 1.5678,
+      "step": 1605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070745495524364e-05,
+      "loss": 1.7181,
+      "step": 1606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069346542751804e-05,
+      "loss": 1.5389,
+      "step": 1607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906794658912932e-05,
+      "loss": 1.6688,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9066545634811405e-05,
+      "loss": 1.8995,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9065143679952662e-05,
+      "loss": 1.6109,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90637407247078e-05,
+      "loss": 1.4811,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.906233676923165e-05,
+      "loss": 1.9096,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9060931813679145e-05,
+      "loss": 1.7099,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059525858205323e-05,
+      "loss": 1.7149,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058118902965344e-05,
+      "loss": 1.7345,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905671094811447e-05,
+      "loss": 1.4202,
+      "step": 1616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055301993808076e-05,
+      "loss": 1.32,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905389204020165e-05,
+      "loss": 1.4481,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052481087450787e-05,
+      "loss": 1.8634,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051069135711192e-05,
+      "loss": 1.7365,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904965618513868e-05,
+      "loss": 1.5189,
+      "step": 1621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048242235889182e-05,
+      "loss": 1.4445,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904682728811873e-05,
+      "loss": 1.4661,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045411341983473e-05,
+      "loss": 1.7433,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043994397639667e-05,
+      "loss": 1.6107,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042576455243682e-05,
+      "loss": 1.7083,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041157514951995e-05,
+      "loss": 1.7732,
+      "step": 1627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903973757692119e-05,
+      "loss": 1.7316,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038316641307964e-05,
+      "loss": 1.7448,
+      "step": 1629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903689470826913e-05,
+      "loss": 1.8408,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035471777961603e-05,
+      "loss": 2.2279,
+      "step": 1631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903404785054241e-05,
+      "loss": 1.4841,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032622926168693e-05,
+      "loss": 1.5453,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903119700499769e-05,
+      "loss": 1.9068,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029770087186775e-05,
+      "loss": 1.396,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90283421728934e-05,
+      "loss": 1.5065,
+      "step": 1636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026913262275148e-05,
+      "loss": 1.4878,
+      "step": 1637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025483355489712e-05,
+      "loss": 1.49,
+      "step": 1638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024052452694883e-05,
+      "loss": 1.3552,
+      "step": 1639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022620554048572e-05,
+      "loss": 1.3484,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021187659708797e-05,
+      "loss": 1.4657,
+      "step": 1641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901975376983368e-05,
+      "loss": 1.7373,
+      "step": 1642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9018318884581467e-05,
+      "loss": 1.3988,
+      "step": 1643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90168830041105e-05,
+      "loss": 1.8962,
+      "step": 1644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9015446128579236e-05,
+      "loss": 1.4746,
+      "step": 1645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901400825814624e-05,
+      "loss": 1.6057,
+      "step": 1646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012569392970192e-05,
+      "loss": 1.3555,
+      "step": 1647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011129533209877e-05,
+      "loss": 1.7476,
+      "step": 1648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.6335,
+      "step": 1649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900824683057214e-05,
+      "loss": 1.527,
+      "step": 1650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006803988012844e-05,
+      "loss": 1.3583,
+      "step": 1651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005360151505517e-05,
+      "loss": 1.6639,
+      "step": 1652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003915321209504e-05,
+      "loss": 2.1558,
+      "step": 1653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002469497284246e-05,
+      "loss": 1.5709,
+      "step": 1654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001022679889298e-05,
+      "loss": 1.5204,
+      "step": 1655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999574869184325e-05,
+      "loss": 1.515,
+      "step": 1656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998126065329098e-05,
+      "loss": 1.4371,
+      "step": 1657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89966762684835e-05,
+      "loss": 1.6056,
+      "step": 1658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899522547880752e-05,
+      "loss": 1.5137,
+      "step": 1659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899377369646127e-05,
+      "loss": 1.7344,
+      "step": 1660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992320921604958e-05,
+      "loss": 1.3216,
+      "step": 1661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89908671543989e-05,
+      "loss": 1.3844,
+      "step": 1662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989412395003536e-05,
+      "loss": 1.5481,
+      "step": 1663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987956643579396e-05,
+      "loss": 1.4334,
+      "step": 1664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986499900287138e-05,
+      "loss": 1.3839,
+      "step": 1665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898504216528752e-05,
+      "loss": 1.6818,
+      "step": 1666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983583438741405e-05,
+      "loss": 1.6754,
+      "step": 1667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982123720809777e-05,
+      "loss": 1.1591,
+      "step": 1668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980663011653718e-05,
+      "loss": 1.4324,
+      "step": 1669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979201311434434e-05,
+      "loss": 1.6138,
+      "step": 1670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977738620313225e-05,
+      "loss": 1.8071,
+      "step": 1671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976274938451504e-05,
+      "loss": 1.5225,
+      "step": 1672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974810266010803e-05,
+      "loss": 1.689,
+      "step": 1673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8973344603152748e-05,
+      "loss": 1.3264,
+      "step": 1674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8971877950039093e-05,
+      "loss": 1.1842,
+      "step": 1675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.897041030683168e-05,
+      "loss": 1.4855,
+      "step": 1676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896894167369248e-05,
+      "loss": 1.6533,
+      "step": 1677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8967472050783556e-05,
+      "loss": 1.5815,
+      "step": 1678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966001438267095e-05,
+      "loss": 1.6933,
+      "step": 1679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896452983630539e-05,
+      "loss": 1.1758,
+      "step": 1680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896305724506083e-05,
+      "loss": 1.512,
+      "step": 1681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961583664695927e-05,
+      "loss": 1.5902,
+      "step": 1682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960109095373304e-05,
+      "loss": 1.7243,
+      "step": 1683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895863353725568e-05,
+      "loss": 1.2169,
+      "step": 1684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957156990505893e-05,
+      "loss": 1.7476,
+      "step": 1685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955679455286892e-05,
+      "loss": 1.4245,
+      "step": 1686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954200931761723e-05,
+      "loss": 2.0149,
+      "step": 1687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952721420093552e-05,
+      "loss": 1.8302,
+      "step": 1688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951240920445654e-05,
+      "loss": 1.6787,
+      "step": 1689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949759432981407e-05,
+      "loss": 1.378,
+      "step": 1690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89482769578643e-05,
+      "loss": 1.4325,
+      "step": 1691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946793495257932e-05,
+      "loss": 1.3033,
+      "step": 1692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945309045326018e-05,
+      "loss": 1.3612,
+      "step": 1693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943823608232363e-05,
+      "loss": 1.3283,
+      "step": 1694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89423371841409e-05,
+      "loss": 1.6835,
+      "step": 1695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940849773215663e-05,
+      "loss": 2.0652,
+      "step": 1696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939361375620796e-05,
+      "loss": 1.5973,
+      "step": 1697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893787199152055e-05,
+      "loss": 1.9069,
+      "step": 1698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893638162107929e-05,
+      "loss": 1.5943,
+      "step": 1699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934890264461477e-05,
+      "loss": 1.7458,
+      "step": 1700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89333979218317e-05,
+      "loss": 1.8188,
+      "step": 1701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893190459335464e-05,
+      "loss": 1.3395,
+      "step": 1702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930410279195102e-05,
+      "loss": 1.3544,
+      "step": 1703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892891497951798e-05,
+      "loss": 1.6085,
+      "step": 1704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927418694488294e-05,
+      "loss": 1.3014,
+      "step": 1705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925921424271175e-05,
+      "loss": 1.7642,
+      "step": 1706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.892442316903184e-05,
+      "loss": 1.83,
+      "step": 1707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8922923928935634e-05,
+      "loss": 1.4835,
+      "step": 1708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.892142370414801e-05,
+      "loss": 1.5493,
+      "step": 1709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8919922494834525e-05,
+      "loss": 1.4616,
+      "step": 1710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8918420301160846e-05,
+      "loss": 1.8131,
+      "step": 1711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891691712329274e-05,
+      "loss": 2.0573,
+      "step": 1712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89154129613961e-05,
+      "loss": 1.6795,
+      "step": 1713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913907815636913e-05,
+      "loss": 1.6199,
+      "step": 1714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912401686181284e-05,
+      "loss": 1.1215,
+      "step": 1715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910894573195414e-05,
+      "loss": 1.4813,
+      "step": 1716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890938647684563e-05,
+      "loss": 1.2701,
+      "step": 1717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907877397298354e-05,
+      "loss": 1.4827,
+      "step": 1718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906367334720125e-05,
+      "loss": 1.4758,
+      "step": 1719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904856289277576e-05,
+      "loss": 1.6295,
+      "step": 1720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903344261137472e-05,
+      "loss": 1.5598,
+      "step": 1721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901831250466664e-05,
+      "loss": 1.5122,
+      "step": 1722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900317257432123e-05,
+      "loss": 1.4574,
+      "step": 1723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898802282200927e-05,
+      "loss": 1.5032,
+      "step": 1724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889728632494026e-05,
+      "loss": 1.3871,
+      "step": 1725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889576938581742e-05,
+      "loss": 1.2646,
+      "step": 1726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894251464999803e-05,
+      "loss": 1.2745,
+      "step": 1727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892732562654924e-05,
+      "loss": 1.7651,
+      "step": 1728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891212678950398e-05,
+      "loss": 1.585,
+      "step": 1729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888969181405396e-05,
+      "loss": 1.5478,
+      "step": 1730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888169968133436e-05,
+      "loss": 1.4887,
+      "step": 1731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886647141356773e-05,
+      "loss": 1.4894,
+      "step": 1732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885123333892025e-05,
+      "loss": 1.6564,
+      "step": 1733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883598545907352e-05,
+      "loss": 1.7786,
+      "step": 1734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882072777571018e-05,
+      "loss": 1.4343,
+      "step": 1735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880546029051406e-05,
+      "loss": 1.5202,
+      "step": 1736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879018300516992e-05,
+      "loss": 1.4474,
+      "step": 1737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877489592136376e-05,
+      "loss": 1.3003,
+      "step": 1738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.887595990407826e-05,
+      "loss": 1.821,
+      "step": 1739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.887442923651145e-05,
+      "loss": 1.1164,
+      "step": 1740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.887289758960486e-05,
+      "loss": 1.57,
+      "step": 1741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.887136496352752e-05,
+      "loss": 2.2742,
+      "step": 1742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8869831358448563e-05,
+      "loss": 1.4277,
+      "step": 1743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886829677453723e-05,
+      "loss": 1.4188,
+      "step": 1744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866761211962867e-05,
+      "loss": 1.4695,
+      "step": 1745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886522467089494e-05,
+      "loss": 2.8682,
+      "step": 1746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863687151503e-05,
+      "loss": 1.4597,
+      "step": 1747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862148653956732e-05,
+      "loss": 1.4434,
+      "step": 1748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860609178425916e-05,
+      "loss": 1.506,
+      "step": 1749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859068725080436e-05,
+      "loss": 1.4487,
+      "step": 1750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857527294090294e-05,
+      "loss": 1.5457,
+      "step": 1751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855984885625594e-05,
+      "loss": 1.4723,
+      "step": 1752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854441499856547e-05,
+      "loss": 1.7459,
+      "step": 1753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852897136953473e-05,
+      "loss": 1.8103,
+      "step": 1754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851351797086806e-05,
+      "loss": 1.5361,
+      "step": 1755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849805480427076e-05,
+      "loss": 1.2077,
+      "step": 1756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848258187144932e-05,
+      "loss": 1.2206,
+      "step": 1757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846709917411124e-05,
+      "loss": 1.7083,
+      "step": 1758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884516067139651e-05,
+      "loss": 1.3251,
+      "step": 1759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884361044927206e-05,
+      "loss": 1.514,
+      "step": 1760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842059251208845e-05,
+      "loss": 1.7913,
+      "step": 1761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840507077378055e-05,
+      "loss": 1.7735,
+      "step": 1762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838953927950972e-05,
+      "loss": 1.6698,
+      "step": 1763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837399803099003e-05,
+      "loss": 1.3884,
+      "step": 1764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883584470299365e-05,
+      "loss": 1.8865,
+      "step": 1765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834288627806523e-05,
+      "loss": 1.7432,
+      "step": 1766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832731577709347e-05,
+      "loss": 1.4997,
+      "step": 1767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831173552873946e-05,
+      "loss": 1.7636,
+      "step": 1768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829614553472265e-05,
+      "loss": 1.6197,
+      "step": 1769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882805457967634e-05,
+      "loss": 1.4538,
+      "step": 1770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826493631658325e-05,
+      "loss": 1.5192,
+      "step": 1771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8824931709590476e-05,
+      "loss": 2.1425,
+      "step": 1772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8823368813645168e-05,
+      "loss": 1.6491,
+      "step": 1773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8821804943994863e-05,
+      "loss": 1.3372,
+      "step": 1774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.882024010081215e-05,
+      "loss": 1.6565,
+      "step": 1775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8818674284269714e-05,
+      "loss": 1.4145,
+      "step": 1776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817107494540355e-05,
+      "loss": 1.3743,
+      "step": 1777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881553973179697e-05,
+      "loss": 1.3483,
+      "step": 1778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813970996212577e-05,
+      "loss": 1.6333,
+      "step": 1779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812401287960295e-05,
+      "loss": 1.6274,
+      "step": 1780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881083060721334e-05,
+      "loss": 1.3214,
+      "step": 1781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809258954145052e-05,
+      "loss": 1.5823,
+      "step": 1782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807686328928868e-05,
+      "loss": 1.7054,
+      "step": 1783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806112731738338e-05,
+      "loss": 1.9244,
+      "step": 1784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804538162747118e-05,
+      "loss": 1.6637,
+      "step": 1785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802962622128965e-05,
+      "loss": 1.6645,
+      "step": 1786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880138611005775e-05,
+      "loss": 2.3935,
+      "step": 1787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879980862670745e-05,
+      "loss": 1.5259,
+      "step": 1788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798230172252154e-05,
+      "loss": 1.4805,
+      "step": 1789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796650746866042e-05,
+      "loss": 1.5325,
+      "step": 1790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795070350723418e-05,
+      "loss": 1.4494,
+      "step": 1791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879348898399869e-05,
+      "loss": 1.6062,
+      "step": 1792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879190664686636e-05,
+      "loss": 1.2909,
+      "step": 1793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879032333950106e-05,
+      "loss": 1.861,
+      "step": 1794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878873906207751e-05,
+      "loss": 1.3972,
+      "step": 1795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787153814770536e-05,
+      "loss": 1.4824,
+      "step": 1796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878556759775509e-05,
+      "loss": 1.5304,
+      "step": 1797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783980411206216e-05,
+      "loss": 1.0918,
+      "step": 1798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782392255299067e-05,
+      "loss": 1.3588,
+      "step": 1799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780803130208903e-05,
+      "loss": 1.2624,
+      "step": 1800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779213036111095e-05,
+      "loss": 1.4789,
+      "step": 1801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777621973181116e-05,
+      "loss": 1.5706,
+      "step": 1802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877602994159455e-05,
+      "loss": 2.0191,
+      "step": 1803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8774436941527085e-05,
+      "loss": 1.2837,
+      "step": 1804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.877284297315452e-05,
+      "loss": 1.487,
+      "step": 1805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.877124803665275e-05,
+      "loss": 1.742,
+      "step": 1806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876965213219779e-05,
+      "loss": 1.4792,
+      "step": 1807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876805525996576e-05,
+      "loss": 1.74,
+      "step": 1808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766457420132878e-05,
+      "loss": 1.3144,
+      "step": 1809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764858612875475e-05,
+      "loss": 1.4213,
+      "step": 1810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763258838369984e-05,
+      "loss": 1.8669,
+      "step": 1811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761658096792957e-05,
+      "loss": 1.9616,
+      "step": 1812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760056388321035e-05,
+      "loss": 1.3535,
+      "step": 1813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758453713130985e-05,
+      "loss": 1.8241,
+      "step": 1814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756850071399663e-05,
+      "loss": 1.561,
+      "step": 1815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755245463304043e-05,
+      "loss": 1.5955,
+      "step": 1816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753639889021197e-05,
+      "loss": 1.284,
+      "step": 1817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752033348728312e-05,
+      "loss": 1.5924,
+      "step": 1818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875042584260268e-05,
+      "loss": 1.4337,
+      "step": 1819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8748817370821696e-05,
+      "loss": 1.5544,
+      "step": 1820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874720793356286e-05,
+      "loss": 1.7034,
+      "step": 1821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745597531003786e-05,
+      "loss": 1.6603,
+      "step": 1822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743986163322193e-05,
+      "loss": 1.6001,
+      "step": 1823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87423738306959e-05,
+      "loss": 1.5105,
+      "step": 1824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740760533302832e-05,
+      "loss": 1.4312,
+      "step": 1825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739146271321032e-05,
+      "loss": 1.3551,
+      "step": 1826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873753104492864e-05,
+      "loss": 1.5061,
+      "step": 1827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87359148543039e-05,
+      "loss": 1.4794,
+      "step": 1828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734297699625175e-05,
+      "loss": 1.4991,
+      "step": 1829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732679581070924e-05,
+      "loss": 1.7791,
+      "step": 1830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873106049881971e-05,
+      "loss": 1.7988,
+      "step": 1831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729440453050215e-05,
+      "loss": 1.4285,
+      "step": 1832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727819443941215e-05,
+      "loss": 1.2694,
+      "step": 1833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726197471671597e-05,
+      "loss": 1.5291,
+      "step": 1834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872457453642035e-05,
+      "loss": 1.314,
+      "step": 1835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872295063836658e-05,
+      "loss": 1.6545,
+      "step": 1836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8721325777689495e-05,
+      "loss": 1.9922,
+      "step": 1837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8719699954568398e-05,
+      "loss": 1.3772,
+      "step": 1838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8718073169182714e-05,
+      "loss": 1.322,
+      "step": 1839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8716445421711964e-05,
+      "loss": 1.4872,
+      "step": 1840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871481671233578e-05,
+      "loss": 1.5956,
+      "step": 1841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713187041233896e-05,
+      "loss": 1.3769,
+      "step": 1842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711556408586154e-05,
+      "loss": 1.659,
+      "step": 1843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870992481457251e-05,
+      "loss": 1.7424,
+      "step": 1844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708292259373015e-05,
+      "loss": 1.6205,
+      "step": 1845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870665874316783e-05,
+      "loss": 1.7963,
+      "step": 1846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870502426613722e-05,
+      "loss": 1.3326,
+      "step": 1847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870338882846156e-05,
+      "loss": 1.4499,
+      "step": 1848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701752430321326e-05,
+      "loss": 1.7981,
+      "step": 1849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700115071897107e-05,
+      "loss": 1.6811,
+      "step": 1850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698476753369594e-05,
+      "loss": 1.7179,
+      "step": 1851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869683747491958e-05,
+      "loss": 1.2076,
+      "step": 1852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695197236727977e-05,
+      "loss": 1.39,
+      "step": 1853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693556038975784e-05,
+      "loss": 1.5157,
+      "step": 1854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691913881844123e-05,
+      "loss": 1.7462,
+      "step": 1855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690270765514206e-05,
+      "loss": 1.0688,
+      "step": 1856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868862669016737e-05,
+      "loss": 1.7076,
+      "step": 1857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686981655985036e-05,
+      "loss": 1.6428,
+      "step": 1858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685335663148753e-05,
+      "loss": 1.5707,
+      "step": 1859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868368871184016e-05,
+      "loss": 1.9593,
+      "step": 1860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682040802241006e-05,
+      "loss": 1.206,
+      "step": 1861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868039193453315e-05,
+      "loss": 1.2297,
+      "step": 1862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678742108898546e-05,
+      "loss": 1.9058,
+      "step": 1863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677091325519268e-05,
+      "loss": 1.6828,
+      "step": 1864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675439584577485e-05,
+      "loss": 1.3329,
+      "step": 1865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673786886255478e-05,
+      "loss": 1.7874,
+      "step": 1866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672133230735628e-05,
+      "loss": 1.6938,
+      "step": 1867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670478618200424e-05,
+      "loss": 1.3251,
+      "step": 1868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8668823048832465e-05,
+      "loss": 2.0868,
+      "step": 1869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866716652281445e-05,
+      "loss": 1.2611,
+      "step": 1870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8665509040329188e-05,
+      "loss": 1.6831,
+      "step": 1871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8663850601559583e-05,
+      "loss": 1.5638,
+      "step": 1872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8662191206688658e-05,
+      "loss": 1.5091,
+      "step": 1873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660530855899533e-05,
+      "loss": 1.3574,
+      "step": 1874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658869549375446e-05,
+      "loss": 1.8061,
+      "step": 1875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657207287299716e-05,
+      "loss": 1.7571,
+      "step": 1876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655544069855795e-05,
+      "loss": 1.5167,
+      "step": 1877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653879897227222e-05,
+      "loss": 1.7702,
+      "step": 1878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652214769597648e-05,
+      "loss": 1.5328,
+      "step": 1879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650548687150825e-05,
+      "loss": 1.509,
+      "step": 1880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864888165007062e-05,
+      "loss": 1.6223,
+      "step": 1881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647213658540997e-05,
+      "loss": 1.6678,
+      "step": 1882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645544712746028e-05,
+      "loss": 1.4738,
+      "step": 1883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643874812869884e-05,
+      "loss": 1.2563,
+      "step": 1884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642203959096857e-05,
+      "loss": 1.7536,
+      "step": 1885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640532151611328e-05,
+      "loss": 1.4901,
+      "step": 1886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863885939059779e-05,
+      "loss": 1.4428,
+      "step": 1887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637185676240846e-05,
+      "loss": 1.5398,
+      "step": 1888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635511008725196e-05,
+      "loss": 1.848,
+      "step": 1889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633835388235647e-05,
+      "loss": 1.3914,
+      "step": 1890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863215881495711e-05,
+      "loss": 1.7035,
+      "step": 1891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863048128907461e-05,
+      "loss": 1.4948,
+      "step": 1892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862880281077327e-05,
+      "loss": 1.2603,
+      "step": 1893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627123380238314e-05,
+      "loss": 1.59,
+      "step": 1894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625442997655078e-05,
+      "loss": 1.5596,
+      "step": 1895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623761663209003e-05,
+      "loss": 1.9927,
+      "step": 1896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622079377085633e-05,
+      "loss": 1.5113,
+      "step": 1897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862039613947061e-05,
+      "loss": 1.2221,
+      "step": 1898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86187119505497e-05,
+      "loss": 1.4902,
+      "step": 1899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617026810508757e-05,
+      "loss": 1.3083,
+      "step": 1900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861534071953374e-05,
+      "loss": 1.4553,
+      "step": 1901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8613653677810722e-05,
+      "loss": 1.5976,
+      "step": 1902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.861196568552588e-05,
+      "loss": 1.7511,
+      "step": 1903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.861027674286549e-05,
+      "loss": 1.5562,
+      "step": 1904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860858685001593e-05,
+      "loss": 1.317,
+      "step": 1905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.86068960071637e-05,
+      "loss": 1.8389,
+      "step": 1906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605204214495383e-05,
+      "loss": 1.6435,
+      "step": 1907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603511472197686e-05,
+      "loss": 1.5765,
+      "step": 1908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601817780457403e-05,
+      "loss": 1.6683,
+      "step": 1909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860012313946145e-05,
+      "loss": 1.2737,
+      "step": 1910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598427549396834e-05,
+      "loss": 1.5071,
+      "step": 1911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596731010450675e-05,
+      "loss": 1.2254,
+      "step": 1912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595033522810193e-05,
+      "loss": 2.2498,
+      "step": 1913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593335086662715e-05,
+      "loss": 1.6454,
+      "step": 1914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591635702195672e-05,
+      "loss": 1.4751,
+      "step": 1915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589935369596605e-05,
+      "loss": 1.8188,
+      "step": 1916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588234089053147e-05,
+      "loss": 1.503,
+      "step": 1917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858653186075305e-05,
+      "loss": 2.6311,
+      "step": 1918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584828684884158e-05,
+      "loss": 1.7633,
+      "step": 1919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858312456163443e-05,
+      "loss": 1.1448,
+      "step": 1920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581419491191923e-05,
+      "loss": 1.419,
+      "step": 1921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579713473744795e-05,
+      "loss": 1.7426,
+      "step": 1922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578006509481326e-05,
+      "loss": 1.3172,
+      "step": 1923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576298598589878e-05,
+      "loss": 1.7333,
+      "step": 1924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574589741258932e-05,
+      "loss": 1.5997,
+      "step": 1925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572879937677072e-05,
+      "loss": 1.6555,
+      "step": 1926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571169188032977e-05,
+      "loss": 1.7533,
+      "step": 1927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569457492515447e-05,
+      "loss": 1.6961,
+      "step": 1928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567744851313363e-05,
+      "loss": 1.3448,
+      "step": 1929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566031264615735e-05,
+      "loss": 1.519,
+      "step": 1930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564316732611666e-05,
+      "loss": 1.3126,
+      "step": 1931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562601255490355e-05,
+      "loss": 1.7984,
+      "step": 1932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856088483344112e-05,
+      "loss": 1.4886,
+      "step": 1933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855916746665338e-05,
+      "loss": 1.7853,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855744915531665e-05,
+      "loss": 1.7263,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855572989962056e-05,
+      "loss": 1.3762,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8554009699754833e-05,
+      "loss": 1.3325,
+      "step": 1937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85522885559093e-05,
+      "loss": 1.6237,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855056646827391e-05,
+      "loss": 1.232,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548843437038693e-05,
+      "loss": 1.5007,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547119462393798e-05,
+      "loss": 1.4254,
+      "step": 1941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854539454452948e-05,
+      "loss": 1.762,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543668683636086e-05,
+      "loss": 2.5998,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854194187990407e-05,
+      "loss": 1.8936,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540214133524005e-05,
+      "loss": 1.5969,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538485444686552e-05,
+      "loss": 1.6511,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853675581358248e-05,
+      "loss": 1.4844,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853502524040266e-05,
+      "loss": 1.3945,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853329372533808e-05,
+      "loss": 1.0225,
+      "step": 1949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853156126857981e-05,
+      "loss": 1.5295,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529827870319042e-05,
+      "loss": 1.7506,
+      "step": 1951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852809353074706e-05,
+      "loss": 1.6661,
+      "step": 1952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526358250055267e-05,
+      "loss": 1.4726,
+      "step": 1953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524622028435156e-05,
+      "loss": 1.3266,
+      "step": 1954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522884866078322e-05,
+      "loss": 1.3171,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521146763176476e-05,
+      "loss": 1.577,
+      "step": 1956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519407719921427e-05,
+      "loss": 1.0684,
+      "step": 1957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851766773650509e-05,
+      "loss": 1.3083,
+      "step": 1958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851592681311947e-05,
+      "loss": 1.516,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85141849499567e-05,
+      "loss": 1.1285,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512442147208997e-05,
+      "loss": 1.2898,
+      "step": 1961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851069840506869e-05,
+      "loss": 1.5638,
+      "step": 1962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850895372372821e-05,
+      "loss": 1.45,
+      "step": 1963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507208103380093e-05,
+      "loss": 1.4037,
+      "step": 1964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505461544216975e-05,
+      "loss": 1.4389,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503714046431597e-05,
+      "loss": 1.629,
+      "step": 1966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8501965610216807e-05,
+      "loss": 1.2649,
+      "step": 1967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8500216235765553e-05,
+      "loss": 1.2848,
+      "step": 1968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849846592327089e-05,
+      "loss": 1.4967,
+      "step": 1969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849671467292597e-05,
+      "loss": 1.5109,
+      "step": 1970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849496248492406e-05,
+      "loss": 1.6047,
+      "step": 1971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493209359458515e-05,
+      "loss": 1.6202,
+      "step": 1972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491455296722805e-05,
+      "loss": 1.6241,
+      "step": 1973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489700296910496e-05,
+      "loss": 1.672,
+      "step": 1974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8487944360215272e-05,
+      "loss": 1.2775,
+      "step": 1975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84861874868309e-05,
+      "loss": 1.3758,
+      "step": 1976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484429676951264e-05,
+      "loss": 1.548,
+      "step": 1977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482670930770343e-05,
+      "loss": 1.5915,
+      "step": 1978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480911248482232e-05,
+      "loss": 1.4495,
+      "step": 1979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479150630281113e-05,
+      "loss": 1.357,
+      "step": 1980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477389076361288e-05,
+      "loss": 1.8813,
+      "step": 1981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475626586917146e-05,
+      "loss": 1.1999,
+      "step": 1982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847386316214319e-05,
+      "loss": 1.2867,
+      "step": 1983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472098802234022e-05,
+      "loss": 1.6045,
+      "step": 1984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470333507384352e-05,
+      "loss": 1.7438,
+      "step": 1985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468567277788983e-05,
+      "loss": 2.0491,
+      "step": 1986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466800113642833e-05,
+      "loss": 1.5627,
+      "step": 1987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846503201514092e-05,
+      "loss": 1.697,
+      "step": 1988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463262982478353e-05,
+      "loss": 1.2955,
+      "step": 1989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461493015850365e-05,
+      "loss": 1.9148,
+      "step": 1990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459722115452274e-05,
+      "loss": 1.5762,
+      "step": 1991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8457950281479513e-05,
+      "loss": 1.4507,
+      "step": 1992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456177514127608e-05,
+      "loss": 1.7179,
+      "step": 1993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454403813592197e-05,
+      "loss": 1.4229,
+      "step": 1994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845262918006902e-05,
+      "loss": 1.8271,
+      "step": 1995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845085361375391e-05,
+      "loss": 1.669,
+      "step": 1996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449077114842814e-05,
+      "loss": 1.5558,
+      "step": 1997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447299683531773e-05,
+      "loss": 1.6695,
+      "step": 1998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8445521320016948e-05,
+      "loss": 1.3792,
+      "step": 1999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8443742024494576e-05,
+      "loss": 1.6087,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.844196179716102e-05,
+      "loss": 1.6724,
+      "step": 2001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8440180638212738e-05,
+      "loss": 1.591,
+      "step": 2002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8438398547846284e-05,
+      "loss": 2.1578,
+      "step": 2003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843661552625833e-05,
+      "loss": 1.2421,
+      "step": 2004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434831573645637e-05,
+      "loss": 1.4634,
+      "step": 2005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843304669020507e-05,
+      "loss": 1.8798,
+      "step": 2006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431260876133606e-05,
+      "loss": 1.3466,
+      "step": 2007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429474131628313e-05,
+      "loss": 1.5065,
+      "step": 2008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427686456886375e-05,
+      "loss": 1.7185,
+      "step": 2009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425897852105063e-05,
+      "loss": 1.4421,
+      "step": 2010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842410831748177e-05,
+      "loss": 1.3336,
+      "step": 2011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422317853213968e-05,
+      "loss": 1.3148,
+      "step": 2012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420526459499252e-05,
+      "loss": 1.7375,
+      "step": 2013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418734136535308e-05,
+      "loss": 1.655,
+      "step": 2014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841694088451993e-05,
+      "loss": 1.3521,
+      "step": 2015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415146703651013e-05,
+      "loss": 1.773,
+      "step": 2016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413351594126556e-05,
+      "loss": 1.4285,
+      "step": 2017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411555556144657e-05,
+      "loss": 1.609,
+      "step": 2018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409758589903516e-05,
+      "loss": 1.4941,
+      "step": 2019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407960695601442e-05,
+      "loss": 1.5165,
+      "step": 2020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406161873436838e-05,
+      "loss": 1.5415,
+      "step": 2021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404362123608215e-05,
+      "loss": 1.6257,
+      "step": 2022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840256144631418e-05,
+      "loss": 1.1234,
+      "step": 2023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400759841753458e-05,
+      "loss": 1.5425,
+      "step": 2024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839895731012486e-05,
+      "loss": 1.4011,
+      "step": 2025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397153851627304e-05,
+      "loss": 1.1432,
+      "step": 2026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395349466459808e-05,
+      "loss": 1.8103,
+      "step": 2027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393544154821502e-05,
+      "loss": 1.4907,
+      "step": 2028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391737916911603e-05,
+      "loss": 1.5623,
+      "step": 2029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389930752929454e-05,
+      "loss": 1.5488,
+      "step": 2030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838812266307447e-05,
+      "loss": 1.5865,
+      "step": 2031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8386313647546184e-05,
+      "loss": 1.4598,
+      "step": 2032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.838450370654424e-05,
+      "loss": 2.0351,
+      "step": 2033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8382692840268367e-05,
+      "loss": 1.4384,
+      "step": 2034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8380881048918406e-05,
+      "loss": 1.559,
+      "step": 2035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8379068332694302e-05,
+      "loss": 1.5689,
+      "step": 2036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837725469179609e-05,
+      "loss": 1.412,
+      "step": 2037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375440126423916e-05,
+      "loss": 1.6255,
+      "step": 2038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837362463677803e-05,
+      "loss": 1.2596,
+      "step": 2039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371808223058782e-05,
+      "loss": 1.2049,
+      "step": 2040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369990885466618e-05,
+      "loss": 1.4408,
+      "step": 2041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836817262420209e-05,
+      "loss": 1.5093,
+      "step": 2042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366353439465862e-05,
+      "loss": 1.162,
+      "step": 2043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364533331458684e-05,
+      "loss": 1.3243,
+      "step": 2044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362712300381413e-05,
+      "loss": 1.3157,
+      "step": 2045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836089034643501e-05,
+      "loss": 1.4213,
+      "step": 2046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359067469820538e-05,
+      "loss": 1.9855,
+      "step": 2047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357243670739162e-05,
+      "loss": 1.7844,
+      "step": 2048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835541894939215e-05,
+      "loss": 1.4386,
+      "step": 2049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353593305980865e-05,
+      "loss": 1.2283,
+      "step": 2050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351766740706775e-05,
+      "loss": 1.7914,
+      "step": 2051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349939253771456e-05,
+      "loss": 1.4204,
+      "step": 2052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348110845376576e-05,
+      "loss": 1.4154,
+      "step": 2053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346281515723917e-05,
+      "loss": 1.3637,
+      "step": 2054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344451265015348e-05,
+      "loss": 1.4671,
+      "step": 2055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834262009345285e-05,
+      "loss": 1.8153,
+      "step": 2056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340788001238497e-05,
+      "loss": 2.3963,
+      "step": 2057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833895498857448e-05,
+      "loss": 1.6959,
+      "step": 2058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337121055663077e-05,
+      "loss": 1.7278,
+      "step": 2059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833528620270667e-05,
+      "loss": 1.6524,
+      "step": 2060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333450429907744e-05,
+      "loss": 1.4056,
+      "step": 2061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331613737468888e-05,
+      "loss": 1.645,
+      "step": 2062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329776125592787e-05,
+      "loss": 1.9722,
+      "step": 2063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.832793759448224e-05,
+      "loss": 2.2177,
+      "step": 2064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.832609814434013e-05,
+      "loss": 1.5492,
+      "step": 2065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8324257775369453e-05,
+      "loss": 1.4049,
+      "step": 2066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8322416487773308e-05,
+      "loss": 1.6944,
+      "step": 2067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8320574281754882e-05,
+      "loss": 1.4164,
+      "step": 2068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831873115751748e-05,
+      "loss": 1.3587,
+      "step": 2069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316887115264493e-05,
+      "loss": 1.5485,
+      "step": 2070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315042155199426e-05,
+      "loss": 1.4702,
+      "step": 2071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831319627752588e-05,
+      "loss": 1.698,
+      "step": 2072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311349482447554e-05,
+      "loss": 1.4361,
+      "step": 2073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309501770168252e-05,
+      "loss": 1.2077,
+      "step": 2074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830765314089188e-05,
+      "loss": 1.6498,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305803594822448e-05,
+      "loss": 1.0593,
+      "step": 2076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303953132164056e-05,
+      "loss": 1.4479,
+      "step": 2077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8302101753120915e-05,
+      "loss": 1.6001,
+      "step": 2078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300249457897336e-05,
+      "loss": 1.6324,
+      "step": 2079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298396246697726e-05,
+      "loss": 1.4535,
+      "step": 2080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82965421197266e-05,
+      "loss": 1.7727,
+      "step": 2081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829468707718857e-05,
+      "loss": 1.4498,
+      "step": 2082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829283111928835e-05,
+      "loss": 1.3908,
+      "step": 2083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290974246230754e-05,
+      "loss": 2.2027,
+      "step": 2084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289116458220695e-05,
+      "loss": 1.5277,
+      "step": 2085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287257755463193e-05,
+      "loss": 1.5966,
+      "step": 2086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8285398138163363e-05,
+      "loss": 1.7928,
+      "step": 2087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283537606526426e-05,
+      "loss": 1.6345,
+      "step": 2088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82816761607577e-05,
+      "loss": 1.304,
+      "step": 2089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827981380106261e-05,
+      "loss": 1.6006,
+      "step": 2090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277950527646672e-05,
+      "loss": 1.5536,
+      "step": 2091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276086340715504e-05,
+      "loss": 1.4818,
+      "step": 2092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827422124047484e-05,
+      "loss": 2.2493,
+      "step": 2093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827235522713049e-05,
+      "loss": 1.3966,
+      "step": 2094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827048830088839e-05,
+      "loss": 1.5543,
+      "step": 2095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268620461954562e-05,
+      "loss": 1.5299,
+      "step": 2096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.826675171053513e-05,
+      "loss": 1.5072,
+      "step": 2097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.826488204683632e-05,
+      "loss": 1.5182,
+      "step": 2098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8263011471064463e-05,
+      "loss": 1.6404,
+      "step": 2099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.826113998342598e-05,
+      "loss": 1.7675,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8259267584127408e-05,
+      "loss": 1.5176,
+      "step": 2101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257394273375367e-05,
+      "loss": 1.3994,
+      "step": 2102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255520051376593e-05,
+      "loss": 1.3838,
+      "step": 2103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253644918337917e-05,
+      "loss": 1.1293,
+      "step": 2104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251768874466264e-05,
+      "loss": 1.6578,
+      "step": 2105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824989191996867e-05,
+      "loss": 1.2947,
+      "step": 2106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248014055052266e-05,
+      "loss": 1.361,
+      "step": 2107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246135279924284e-05,
+      "loss": 1.5232,
+      "step": 2108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244255594792056e-05,
+      "loss": 1.4284,
+      "step": 2109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242374999863016e-05,
+      "loss": 1.3255,
+      "step": 2110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240493495344695e-05,
+      "loss": 1.4287,
+      "step": 2111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823861108144473e-05,
+      "loss": 1.5315,
+      "step": 2112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236727758370853e-05,
+      "loss": 1.4512,
+      "step": 2113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82348435263309e-05,
+      "loss": 1.6981,
+      "step": 2114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232958385532807e-05,
+      "loss": 1.4094,
+      "step": 2115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231072336184604e-05,
+      "loss": 1.2058,
+      "step": 2116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229185378494433e-05,
+      "loss": 1.8717,
+      "step": 2117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822729751267053e-05,
+      "loss": 1.4618,
+      "step": 2118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225408738921225e-05,
+      "loss": 1.6898,
+      "step": 2119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223519057454954e-05,
+      "loss": 1.7458,
+      "step": 2120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221628468480257e-05,
+      "loss": 0.9804,
+      "step": 2121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821973697220577e-05,
+      "loss": 1.5375,
+      "step": 2122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821784456884023e-05,
+      "loss": 1.5453,
+      "step": 2123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821595125859247e-05,
+      "loss": 1.7344,
+      "step": 2124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214057041671434e-05,
+      "loss": 1.5604,
+      "step": 2125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821216191828615e-05,
+      "loss": 1.4394,
+      "step": 2126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821026588864576e-05,
+      "loss": 1.4313,
+      "step": 2127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82083689529595e-05,
+      "loss": 1.5591,
+      "step": 2128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8206471111436705e-05,
+      "loss": 1.4729,
+      "step": 2129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8204572364286816e-05,
+      "loss": 1.4929,
+      "step": 2130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8202672711719364e-05,
+      "loss": 1.5492,
+      "step": 2131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.820077215394399e-05,
+      "loss": 1.5555,
+      "step": 2132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8198870691170426e-05,
+      "loss": 1.5244,
+      "step": 2133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196968323608515e-05,
+      "loss": 1.9479,
+      "step": 2134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195065051468187e-05,
+      "loss": 1.5725,
+      "step": 2135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193160874959486e-05,
+      "loss": 1.7137,
+      "step": 2136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191255794292536e-05,
+      "loss": 1.8261,
+      "step": 2137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189349809677584e-05,
+      "loss": 1.4036,
+      "step": 2138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818744292132496e-05,
+      "loss": 2.2164,
+      "step": 2139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185535129445096e-05,
+      "loss": 1.3958,
+      "step": 2140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818362643424853e-05,
+      "loss": 1.3843,
+      "step": 2141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181716835945902e-05,
+      "loss": 1.577,
+      "step": 2142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179806334747937e-05,
+      "loss": 1.7531,
+      "step": 2143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177894930865476e-05,
+      "loss": 1.5387,
+      "step": 2144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817598262450945e-05,
+      "loss": 1.6174,
+      "step": 2145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174069415890888e-05,
+      "loss": 1.895,
+      "step": 2146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8172155305220932e-05,
+      "loss": 1.7028,
+      "step": 2147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170240292710802e-05,
+      "loss": 1.4975,
+      "step": 2148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168324378571837e-05,
+      "loss": 1.7326,
+      "step": 2149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816640756301547e-05,
+      "loss": 1.4365,
+      "step": 2150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164489846253225e-05,
+      "loss": 1.4767,
+      "step": 2151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162571228496738e-05,
+      "loss": 1.372,
+      "step": 2152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160651709957737e-05,
+      "loss": 1.3464,
+      "step": 2153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158731290848052e-05,
+      "loss": 1.9772,
+      "step": 2154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156809971379607e-05,
+      "loss": 1.6334,
+      "step": 2155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815488775176443e-05,
+      "loss": 1.4865,
+      "step": 2156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8152964632214654e-05,
+      "loss": 1.5482,
+      "step": 2157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81510406129425e-05,
+      "loss": 1.2191,
+      "step": 2158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149115694160295e-05,
+      "loss": 1.3604,
+      "step": 2159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147189876080463e-05,
+      "loss": 1.9548,
+      "step": 2160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145263158915533e-05,
+      "loss": 1.6634,
+      "step": 2161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8143335542878122e-05,
+      "loss": 1.8381,
+      "step": 2162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8141407028180953e-05,
+      "loss": 1.4093,
+      "step": 2163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8139477615036852e-05,
+      "loss": 1.638,
+      "step": 2164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813754730365874e-05,
+      "loss": 1.4911,
+      "step": 2165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8135616094259632e-05,
+      "loss": 1.701,
+      "step": 2166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813368398705265e-05,
+      "loss": 1.5904,
+      "step": 2167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813175098225101e-05,
+      "loss": 1.6363,
+      "step": 2168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129817080068033e-05,
+      "loss": 1.6992,
+      "step": 2169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127882280717134e-05,
+      "loss": 1.3693,
+      "step": 2170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812594658441183e-05,
+      "loss": 1.5495,
+      "step": 2171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124009991365724e-05,
+      "loss": 1.4891,
+      "step": 2172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122072501792547e-05,
+      "loss": 1.6408,
+      "step": 2173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120134115906097e-05,
+      "loss": 1.7455,
+      "step": 2174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811819483392029e-05,
+      "loss": 1.3985,
+      "step": 2175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116254656049138e-05,
+      "loss": 1.6443,
+      "step": 2176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811431358250675e-05,
+      "loss": 1.53,
+      "step": 2177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112371613507326e-05,
+      "loss": 1.5545,
+      "step": 2178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110428749265184e-05,
+      "loss": 1.322,
+      "step": 2179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810848498999472e-05,
+      "loss": 1.4949,
+      "step": 2180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810654033591044e-05,
+      "loss": 0.9921,
+      "step": 2181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810459478722695e-05,
+      "loss": 1.5488,
+      "step": 2182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810264834415895e-05,
+      "loss": 1.6472,
+      "step": 2183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810070100692124e-05,
+      "loss": 1.2899,
+      "step": 2184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098752775728713e-05,
+      "loss": 1.4689,
+      "step": 2185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809680365079638e-05,
+      "loss": 1.1893,
+      "step": 2186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094853632339324e-05,
+      "loss": 1.2808,
+      "step": 2187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092902720572744e-05,
+      "loss": 1.7498,
+      "step": 2188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090950915711937e-05,
+      "loss": 1.7422,
+      "step": 2189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088998217972294e-05,
+      "loss": 1.3694,
+      "step": 2190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80870446275693e-05,
+      "loss": 1.2861,
+      "step": 2191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808509014471855e-05,
+      "loss": 1.4878,
+      "step": 2192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083134769635728e-05,
+      "loss": 1.5116,
+      "step": 2193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.808117850253662e-05,
+      "loss": 1.7038,
+      "step": 2194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8079221343637115e-05,
+      "loss": 1.3315,
+      "step": 2195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807726329315319e-05,
+      "loss": 1.3985,
+      "step": 2196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8075304351300925e-05,
+      "loss": 1.7286,
+      "step": 2197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8073344518296505e-05,
+      "loss": 1.4898,
+      "step": 2198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8071383794356207e-05,
+      "loss": 1.4858,
+      "step": 2199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069422179696404e-05,
+      "loss": 1.8138,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067459674533573e-05,
+      "loss": 1.5997,
+      "step": 2201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065496279084286e-05,
+      "loss": 1.5977,
+      "step": 2202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8063531993565208e-05,
+      "loss": 1.3579,
+      "step": 2203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806156681819312e-05,
+      "loss": 1.7031,
+      "step": 2204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059600753184884e-05,
+      "loss": 1.9744,
+      "step": 2205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805763379875746e-05,
+      "loss": 1.77,
+      "step": 2206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055665955127922e-05,
+      "loss": 1.2791,
+      "step": 2207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053697222513422e-05,
+      "loss": 1.5984,
+      "step": 2208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8051727601131228e-05,
+      "loss": 1.5585,
+      "step": 2209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049757091198693e-05,
+      "loss": 1.4195,
+      "step": 2210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047785692933276e-05,
+      "loss": 1.3537,
+      "step": 2211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804581340655253e-05,
+      "loss": 1.9591,
+      "step": 2212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043840232274104e-05,
+      "loss": 1.4549,
+      "step": 2213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8041866170315755e-05,
+      "loss": 1.5264,
+      "step": 2214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039891220895322e-05,
+      "loss": 1.5849,
+      "step": 2215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803791538423076e-05,
+      "loss": 1.6481,
+      "step": 2216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035938660540106e-05,
+      "loss": 1.7192,
+      "step": 2217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033961050041508e-05,
+      "loss": 1.5667,
+      "step": 2218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80319825529532e-05,
+      "loss": 1.5612,
+      "step": 2219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030003169493526e-05,
+      "loss": 1.5655,
+      "step": 2220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028022899880913e-05,
+      "loss": 1.403,
+      "step": 2221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80260417443339e-05,
+      "loss": 1.7977,
+      "step": 2222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802405970307112e-05,
+      "loss": 1.4814,
+      "step": 2223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022076776311296e-05,
+      "loss": 1.5618,
+      "step": 2224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020092964273256e-05,
+      "loss": 1.4977,
+      "step": 2225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801810826717592e-05,
+      "loss": 1.4148,
+      "step": 2226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.801612268523832e-05,
+      "loss": 1.3347,
+      "step": 2227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8014136218679566e-05,
+      "loss": 1.2294,
+      "step": 2228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8012148867718882e-05,
+      "loss": 1.2196,
+      "step": 2229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.801016063257558e-05,
+      "loss": 1.4354,
+      "step": 2230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8008171513469067e-05,
+      "loss": 1.8515,
+      "step": 2231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800618151061886e-05,
+      "loss": 1.4837,
+      "step": 2232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800419062424456e-05,
+      "loss": 1.772,
+      "step": 2233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800219885456588e-05,
+      "loss": 1.8034,
+      "step": 2234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000206201802615e-05,
+      "loss": 1.4907,
+      "step": 2235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998212666174668e-05,
+      "loss": 1.4284,
+      "step": 2236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996218247902035e-05,
+      "loss": 1.5515,
+      "step": 2237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799422294720481e-05,
+      "loss": 1.6565,
+      "step": 2238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992226764303183e-05,
+      "loss": 1.5884,
+      "step": 2239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990229699417448e-05,
+      "loss": 1.3281,
+      "step": 2240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988231752767992e-05,
+      "loss": 1.6042,
+      "step": 2241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798623292457529e-05,
+      "loss": 1.305,
+      "step": 2242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984233215059934e-05,
+      "loss": 1.2187,
+      "step": 2243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79822326244426e-05,
+      "loss": 1.3644,
+      "step": 2244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980231152944053e-05,
+      "loss": 1.8716,
+      "step": 2245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978228800785182e-05,
+      "loss": 1.9237,
+      "step": 2246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976225568186944e-05,
+      "loss": 1.662,
+      "step": 2247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974221455370417e-05,
+      "loss": 1.6639,
+      "step": 2248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797221646255676e-05,
+      "loss": 1.5141,
+      "step": 2249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970210589967227e-05,
+      "loss": 1.258,
+      "step": 2250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968203837823192e-05,
+      "loss": 1.923,
+      "step": 2251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79661962063461e-05,
+      "loss": 1.5974,
+      "step": 2252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964187695757502e-05,
+      "loss": 2.4358,
+      "step": 2253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7962178306279055e-05,
+      "loss": 1.6058,
+      "step": 2254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960168038132505e-05,
+      "loss": 1.8952,
+      "step": 2255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795815689153969e-05,
+      "loss": 1.4601,
+      "step": 2256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956144866722556e-05,
+      "loss": 1.3203,
+      "step": 2257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954131963903134e-05,
+      "loss": 1.4708,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7952118183303567e-05,
+      "loss": 1.4724,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.795010352514608e-05,
+      "loss": 1.4074,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7948087989653004e-05,
+      "loss": 1.5995,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7946071577046764e-05,
+      "loss": 2.1482,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7944054287549878e-05,
+      "loss": 1.5809,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794203612138497e-05,
+      "loss": 1.5139,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940017078774746e-05,
+      "loss": 1.4488,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793799715994203e-05,
+      "loss": 1.4208,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935976365109723e-05,
+      "loss": 1.3525,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933954694500832e-05,
+      "loss": 1.3858,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793193214833846e-05,
+      "loss": 1.9707,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929908726845804e-05,
+      "loss": 1.54,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792788443024616e-05,
+      "loss": 1.7625,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925859258762916e-05,
+      "loss": 1.6027,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923833212619567e-05,
+      "loss": 1.6959,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921806292039696e-05,
+      "loss": 1.3635,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791977849724698e-05,
+      "loss": 1.3129,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917749828465203e-05,
+      "loss": 1.6304,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791572028591823e-05,
+      "loss": 1.3751,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913689869830044e-05,
+      "loss": 1.426,
+      "step": 2278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911658580424704e-05,
+      "loss": 1.4175,
+      "step": 2279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909626417926375e-05,
+      "loss": 1.5905,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907593382559316e-05,
+      "loss": 1.5811,
+      "step": 2281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905559474547886e-05,
+      "loss": 1.8327,
+      "step": 2282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903524694116532e-05,
+      "loss": 1.5883,
+      "step": 2283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901489041489806e-05,
+      "loss": 1.6536,
+      "step": 2284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899452516892355e-05,
+      "loss": 1.607,
+      "step": 2285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897415120548916e-05,
+      "loss": 1.491,
+      "step": 2286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895376852684333e-05,
+      "loss": 1.8526,
+      "step": 2287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893337713523534e-05,
+      "loss": 1.8381,
+      "step": 2288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7891297703291545e-05,
+      "loss": 1.5402,
+      "step": 2289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78892568222135e-05,
+      "loss": 1.609,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788721507051462e-05,
+      "loss": 1.1791,
+      "step": 2291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7885172448420218e-05,
+      "loss": 1.2473,
+      "step": 2292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7883128956155705e-05,
+      "loss": 1.5373,
+      "step": 2293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7881084593946604e-05,
+      "loss": 1.3992,
+      "step": 2294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7879039362018513e-05,
+      "loss": 1.4398,
+      "step": 2295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7876993260597133e-05,
+      "loss": 1.6613,
+      "step": 2296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874946289908264e-05,
+      "loss": 1.4097,
+      "step": 2297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78728984501778e-05,
+      "loss": 1.2209,
+      "step": 2298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787084974163173e-05,
+      "loss": 1.3632,
+      "step": 2299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786880016449614e-05,
+      "loss": 1.6848,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866749718997212e-05,
+      "loss": 1.4704,
+      "step": 2301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864698405361223e-05,
+      "loss": 1.1937,
+      "step": 2302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786264622381455e-05,
+      "loss": 1.471,
+      "step": 2303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786059317458366e-05,
+      "loss": 1.6681,
+      "step": 2304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785853925789511e-05,
+      "loss": 1.1775,
+      "step": 2305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785648447397557e-05,
+      "loss": 1.3885,
+      "step": 2306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785442882305179e-05,
+      "loss": 1.4268,
+      "step": 2307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852372305350626e-05,
+      "loss": 1.5101,
+      "step": 2308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850314921099025e-05,
+      "loss": 1.3023,
+      "step": 2309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784825667052403e-05,
+      "loss": 2.0653,
+      "step": 2310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846197553852778e-05,
+      "loss": 1.5045,
+      "step": 2311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844137571312507e-05,
+      "loss": 1.8816,
+      "step": 2312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784207672313054e-05,
+      "loss": 1.5468,
+      "step": 2313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840015009534308e-05,
+      "loss": 1.5686,
+      "step": 2314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783795243075133e-05,
+      "loss": 1.7399,
+      "step": 2315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835888987009228e-05,
+      "loss": 1.3732,
+      "step": 2316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833824678535703e-05,
+      "loss": 1.4234,
+      "step": 2317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831759505558573e-05,
+      "loss": 1.8352,
+      "step": 2318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829693468305733e-05,
+      "loss": 1.3353,
+      "step": 2319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827626567005183e-05,
+      "loss": 1.6721,
+      "step": 2320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825558801885016e-05,
+      "loss": 1.6433,
+      "step": 2321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823490173173425e-05,
+      "loss": 1.5564,
+      "step": 2322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821420681098688e-05,
+      "loss": 1.5978,
+      "step": 2323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.781935032588919e-05,
+      "loss": 1.7944,
+      "step": 2324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7817279107773405e-05,
+      "loss": 1.804,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7815207026979894e-05,
+      "loss": 1.521,
+      "step": 2326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7813134083737332e-05,
+      "loss": 1.7562,
+      "step": 2327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7811060278274474e-05,
+      "loss": 1.717,
+      "step": 2328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780898561082018e-05,
+      "loss": 1.7952,
+      "step": 2329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806910081603396e-05,
+      "loss": 1.279,
+      "step": 2330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780483369085317e-05,
+      "loss": 1.8354,
+      "step": 2331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780275643879864e-05,
+      "loss": 1.4884,
+      "step": 2332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800678325669044e-05,
+      "loss": 1.3794,
+      "step": 2333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779859935169371e-05,
+      "loss": 1.5412,
+      "step": 2334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796519517102068e-05,
+      "loss": 1.4118,
+      "step": 2335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794438822123634e-05,
+      "loss": 2.3015,
+      "step": 2336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792357266988026e-05,
+      "loss": 1.1629,
+      "step": 2337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790274851924952e-05,
+      "loss": 1.6498,
+      "step": 2338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788191577164224e-05,
+      "loss": 1.5896,
+      "step": 2339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786107442935735e-05,
+      "loss": 1.5225,
+      "step": 2340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784022449469487e-05,
+      "loss": 1.3117,
+      "step": 2341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781936596995565e-05,
+      "loss": 1.7072,
+      "step": 2342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777984988574415e-05,
+      "loss": 1.5264,
+      "step": 2343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777762315945533e-05,
+      "loss": 1.5244,
+      "step": 2344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775673887830075e-05,
+      "loss": 1.4601,
+      "step": 2345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773584601628256e-05,
+      "loss": 1.4403,
+      "step": 2346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771494457570635e-05,
+      "loss": 1.4735,
+      "step": 2347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769403455887872e-05,
+      "loss": 1.6096,
+      "step": 2348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767311596810716e-05,
+      "loss": 1.2588,
+      "step": 2349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765218880570017e-05,
+      "loss": 1.641,
+      "step": 2350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763125307396718e-05,
+      "loss": 1.9725,
+      "step": 2351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761030877521858e-05,
+      "loss": 1.7105,
+      "step": 2352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758935591176562e-05,
+      "loss": 1.6399,
+      "step": 2353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7756839448592056e-05,
+      "loss": 1.3588,
+      "step": 2354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775474244999967e-05,
+      "loss": 1.3428,
+      "step": 2355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775264459563081e-05,
+      "loss": 1.6529,
+      "step": 2356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.775054588571699e-05,
+      "loss": 1.4105,
+      "step": 2357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7748446320489807e-05,
+      "loss": 1.4459,
+      "step": 2358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7746345900180968e-05,
+      "loss": 1.7485,
+      "step": 2359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7744244625022256e-05,
+      "loss": 1.6964,
+      "step": 2360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7742142495245567e-05,
+      "loss": 1.4828,
+      "step": 2361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740039511082875e-05,
+      "loss": 1.7842,
+      "step": 2362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737935672766256e-05,
+      "loss": 1.4265,
+      "step": 2363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735830980527887e-05,
+      "loss": 1.5912,
+      "step": 2364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733725434600023e-05,
+      "loss": 1.5799,
+      "step": 2365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7731619035215026e-05,
+      "loss": 1.8037,
+      "step": 2366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729511782605345e-05,
+      "loss": 1.3188,
+      "step": 2367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772740367700353e-05,
+      "loss": 1.9559,
+      "step": 2368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725294718642222e-05,
+      "loss": 1.5518,
+      "step": 2369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723184907754154e-05,
+      "loss": 1.6641,
+      "step": 2370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721074244572152e-05,
+      "loss": 1.3074,
+      "step": 2371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771896272932914e-05,
+      "loss": 1.7992,
+      "step": 2372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716850362258142e-05,
+      "loss": 1.6254,
+      "step": 2373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771473714359226e-05,
+      "loss": 1.3077,
+      "step": 2374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77126230735647e-05,
+      "loss": 1.6537,
+      "step": 2375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710508152408764e-05,
+      "loss": 1.7954,
+      "step": 2376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708392380357844e-05,
+      "loss": 2.0889,
+      "step": 2377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706275757645427e-05,
+      "loss": 0.8259,
+      "step": 2378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704158284505086e-05,
+      "loss": 2.2486,
+      "step": 2379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702039961170506e-05,
+      "loss": 1.4705,
+      "step": 2380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769992078787545e-05,
+      "loss": 1.2882,
+      "step": 2381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769780076485378e-05,
+      "loss": 1.5853,
+      "step": 2382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769567989233945e-05,
+      "loss": 1.2831,
+      "step": 2383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769355817056651e-05,
+      "loss": 1.0985,
+      "step": 2384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769143559976911e-05,
+      "loss": 1.299,
+      "step": 2385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689312180181475e-05,
+      "loss": 1.3449,
+      "step": 2386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687187912037945e-05,
+      "loss": 1.4242,
+      "step": 2387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768506279557294e-05,
+      "loss": 1.5858,
+      "step": 2388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.768293683102098e-05,
+      "loss": 1.3669,
+      "step": 2389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7680810018616675e-05,
+      "loss": 1.6055,
+      "step": 2390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7678682358594727e-05,
+      "loss": 1.3821,
+      "step": 2391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767655385118994e-05,
+      "loss": 1.4589,
+      "step": 2392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7674424496637203e-05,
+      "loss": 1.5243,
+      "step": 2393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7672294295171497e-05,
+      "loss": 1.3774,
+      "step": 2394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767016324702791e-05,
+      "loss": 1.3103,
+      "step": 2395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766803135244161e-05,
+      "loss": 1.481,
+      "step": 2396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766589861164786e-05,
+      "loss": 1.5569,
+      "step": 2397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766376502488202e-05,
+      "loss": 1.4869,
+      "step": 2398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7661630592379548e-05,
+      "loss": 1.6258,
+      "step": 2399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659495314375986e-05,
+      "loss": 1.4608,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765735919110697e-05,
+      "loss": 1.6317,
+      "step": 2401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765522222280824e-05,
+      "loss": 1.439,
+      "step": 2402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765308440971561e-05,
+      "loss": 1.7311,
+      "step": 2403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650945752065013e-05,
+      "loss": 1.4905,
+      "step": 2404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764880625009245e-05,
+      "loss": 1.2827,
+      "step": 2405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646665904034032e-05,
+      "loss": 1.6138,
+      "step": 2406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644524714125955e-05,
+      "loss": 1.6193,
+      "step": 2407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642382680604512e-05,
+      "loss": 1.6026,
+      "step": 2408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640239803706088e-05,
+      "loss": 1.6241,
+      "step": 2409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763809608366716e-05,
+      "loss": 1.4521,
+      "step": 2410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635951520724293e-05,
+      "loss": 1.5578,
+      "step": 2411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763380611511416e-05,
+      "loss": 2.0113,
+      "step": 2412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631659867073513e-05,
+      "loss": 1.5277,
+      "step": 2413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629512776839206e-05,
+      "loss": 1.4438,
+      "step": 2414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627364844648175e-05,
+      "loss": 1.3936,
+      "step": 2415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762521607073746e-05,
+      "loss": 1.0666,
+      "step": 2416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623066455344185e-05,
+      "loss": 1.3861,
+      "step": 2417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620915998705578e-05,
+      "loss": 1.4214,
+      "step": 2418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618764701058948e-05,
+      "loss": 1.6131,
+      "step": 2419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616612562641702e-05,
+      "loss": 1.4597,
+      "step": 2420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614459583691346e-05,
+      "loss": 1.4385,
+      "step": 2421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7612305764445464e-05,
+      "loss": 1.2707,
+      "step": 2422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7610151105141748e-05,
+      "loss": 1.1847,
+      "step": 2423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7607995606017973e-05,
+      "loss": 1.5916,
+      "step": 2424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7605839267312007e-05,
+      "loss": 1.3716,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760368208926182e-05,
+      "loss": 1.643,
+      "step": 2426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760152407210546e-05,
+      "loss": 1.504,
+      "step": 2427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759936521608108e-05,
+      "loss": 1.7142,
+      "step": 2428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597205521426925e-05,
+      "loss": 1.3714,
+      "step": 2429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759504498838132e-05,
+      "loss": 1.5329,
+      "step": 2430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75928836171827e-05,
+      "loss": 1.5848,
+      "step": 2431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590721408069572e-05,
+      "loss": 1.6176,
+      "step": 2432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758855836128056e-05,
+      "loss": 1.3945,
+      "step": 2433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586394477054355e-05,
+      "loss": 1.9171,
+      "step": 2434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7584229755629767e-05,
+      "loss": 1.9026,
+      "step": 2435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582064197245673e-05,
+      "loss": 1.5019,
+      "step": 2436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579897802141058e-05,
+      "loss": 1.475,
+      "step": 2437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577730570554995e-05,
+      "loss": 1.3595,
+      "step": 2438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575562502726645e-05,
+      "loss": 1.3014,
+      "step": 2439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573393598895277e-05,
+      "loss": 1.4639,
+      "step": 2440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757122385930023e-05,
+      "loss": 1.999,
+      "step": 2441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756905328418095e-05,
+      "loss": 1.7967,
+      "step": 2442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756688187377697e-05,
+      "loss": 1.4047,
+      "step": 2443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756470962832792e-05,
+      "loss": 1.2305,
+      "step": 2444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562536548073517e-05,
+      "loss": 1.6775,
+      "step": 2445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560362633253566e-05,
+      "loss": 1.631,
+      "step": 2446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558187884107978e-05,
+      "loss": 1.3983,
+      "step": 2447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556012300876747e-05,
+      "loss": 1.9605,
+      "step": 2448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7553835883799954e-05,
+      "loss": 1.3665,
+      "step": 2449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551658633117784e-05,
+      "loss": 1.6947,
+      "step": 2450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549480549070508e-05,
+      "loss": 1.5889,
+      "step": 2451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547301631898484e-05,
+      "loss": 1.687,
+      "step": 2452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754512188184217e-05,
+      "loss": 1.4716,
+      "step": 2453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7542941299142113e-05,
+      "loss": 1.2482,
+      "step": 2454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7540759884038954e-05,
+      "loss": 1.4898,
+      "step": 2455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753857763677342e-05,
+      "loss": 1.5265,
+      "step": 2456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7536394557586333e-05,
+      "loss": 1.3815,
+      "step": 2457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7534210646718606e-05,
+      "loss": 1.4538,
+      "step": 2458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7532025904411253e-05,
+      "loss": 1.4313,
+      "step": 2459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529840330905366e-05,
+      "loss": 1.5255,
+      "step": 2460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527653926442133e-05,
+      "loss": 1.7673,
+      "step": 2461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525466691262837e-05,
+      "loss": 1.6669,
+      "step": 2462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523278625608854e-05,
+      "loss": 1.6184,
+      "step": 2463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521089729721643e-05,
+      "loss": 1.7635,
+      "step": 2464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518900003842764e-05,
+      "loss": 1.4908,
+      "step": 2465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516709448213863e-05,
+      "loss": 1.5854,
+      "step": 2466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751451806307668e-05,
+      "loss": 1.6539,
+      "step": 2467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7512325848673044e-05,
+      "loss": 1.4948,
+      "step": 2468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510132805244876e-05,
+      "loss": 1.4371,
+      "step": 2469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.75079389330342e-05,
+      "loss": 1.5392,
+      "step": 2470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750574423228311e-05,
+      "loss": 1.4462,
+      "step": 2471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503548703233808e-05,
+      "loss": 1.3271,
+      "step": 2472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750135234612858e-05,
+      "loss": 1.3105,
+      "step": 2473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499155161209808e-05,
+      "loss": 1.5186,
+      "step": 2474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749695714871996e-05,
+      "loss": 1.8438,
+      "step": 2475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74947583089016e-05,
+      "loss": 1.4276,
+      "step": 2476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492558641997378e-05,
+      "loss": 1.3904,
+      "step": 2477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490358148250047e-05,
+      "loss": 1.3139,
+      "step": 2478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488156827902433e-05,
+      "loss": 1.4547,
+      "step": 2479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748595468119747e-05,
+      "loss": 1.6808,
+      "step": 2480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483751708378176e-05,
+      "loss": 1.3606,
+      "step": 2481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481547909687656e-05,
+      "loss": 1.217,
+      "step": 2482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7479343285369115e-05,
+      "loss": 1.6996,
+      "step": 2483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477137835665843e-05,
+      "loss": 1.6285,
+      "step": 2484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474931560821225e-05,
+      "loss": 1.9917,
+      "step": 2485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7472724461078734e-05,
+      "loss": 1.4267,
+      "step": 2486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7470516536681934e-05,
+      "loss": 1.5292,
+      "step": 2487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.746830778787448e-05,
+      "loss": 1.6324,
+      "step": 2488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7466098214900126e-05,
+      "loss": 1.7711,
+      "step": 2489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7463887818002697e-05,
+      "loss": 1.7237,
+      "step": 2490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7461676597426134e-05,
+      "loss": 1.6554,
+      "step": 2491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745946455341445e-05,
+      "loss": 1.5112,
+      "step": 2492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745725168621176e-05,
+      "loss": 1.4349,
+      "step": 2493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745503799606226e-05,
+      "loss": 1.6638,
+      "step": 2494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745282348321025e-05,
+      "loss": 1.573,
+      "step": 2495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450608147900108e-05,
+      "loss": 1.4132,
+      "step": 2496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744839199037631e-05,
+      "loss": 1.6668,
+      "step": 2497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446175010883412e-05,
+      "loss": 1.6062,
+      "step": 2498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443957209666082e-05,
+      "loss": 1.5008,
+      "step": 2499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441738586969062e-05,
+      "loss": 1.7883,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439519143037183e-05,
+      "loss": 1.4874,
+      "step": 2501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437298878115378e-05,
+      "loss": 1.4744,
+      "step": 2502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435077792448666e-05,
+      "loss": 1.2847,
+      "step": 2503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743285588628215e-05,
+      "loss": 1.2901,
+      "step": 2504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743063315986103e-05,
+      "loss": 1.676,
+      "step": 2505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74284096134306e-05,
+      "loss": 1.3577,
+      "step": 2506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426185247236237e-05,
+      "loss": 1.5904,
+      "step": 2507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423960061523413e-05,
+      "loss": 1.6286,
+      "step": 2508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421734056537687e-05,
+      "loss": 1.6425,
+      "step": 2509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741950723252471e-05,
+      "loss": 1.8541,
+      "step": 2510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7417279589730226e-05,
+      "loss": 1.278,
+      "step": 2511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415051128400062e-05,
+      "loss": 1.7043,
+      "step": 2512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7412821848780147e-05,
+      "loss": 1.6275,
+      "step": 2513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410591751116492e-05,
+      "loss": 2.0522,
+      "step": 2514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408360835655196e-05,
+      "loss": 1.3978,
+      "step": 2515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406129102642455e-05,
+      "loss": 1.6331,
+      "step": 2516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403896552324554e-05,
+      "loss": 1.4974,
+      "step": 2517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401663184947867e-05,
+      "loss": 2.0108,
+      "step": 2518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.739942900075885e-05,
+      "loss": 1.4074,
+      "step": 2519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7397194000004066e-05,
+      "loss": 1.683,
+      "step": 2520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.739495818293016e-05,
+      "loss": 1.6046,
+      "step": 2521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7392721549783856e-05,
+      "loss": 1.6901,
+      "step": 2522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7390484100811986e-05,
+      "loss": 1.3202,
+      "step": 2523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7388245836261463e-05,
+      "loss": 1.5058,
+      "step": 2524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738600675637929e-05,
+      "loss": 1.871,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738376686141256e-05,
+      "loss": 1.4272,
+      "step": 2526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381526151608463e-05,
+      "loss": 1.5842,
+      "step": 2527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379284627214265e-05,
+      "loss": 1.4844,
+      "step": 2528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377042288477336e-05,
+      "loss": 1.4749,
+      "step": 2529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737479913564513e-05,
+      "loss": 1.5996,
+      "step": 2530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7372555168965184e-05,
+      "loss": 1.3506,
+      "step": 2531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737031038868514e-05,
+      "loss": 2.0484,
+      "step": 2532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368064795052713e-05,
+      "loss": 1.7354,
+      "step": 2533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365818388315724e-05,
+      "loss": 1.5144,
+      "step": 2534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736357116872207e-05,
+      "loss": 2.1472,
+      "step": 2535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361323136519745e-05,
+      "loss": 1.6346,
+      "step": 2536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735907429195683e-05,
+      "loss": 1.6363,
+      "step": 2537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356824635281506e-05,
+      "loss": 1.6,
+      "step": 2538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354574166742017e-05,
+      "loss": 1.4448,
+      "step": 2539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735232288658673e-05,
+      "loss": 1.6083,
+      "step": 2540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350070795064077e-05,
+      "loss": 1.4575,
+      "step": 2541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734781789242259e-05,
+      "loss": 1.5988,
+      "step": 2542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734556417891089e-05,
+      "loss": 1.4102,
+      "step": 2543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343309654777686e-05,
+      "loss": 1.5071,
+      "step": 2544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341054320271777e-05,
+      "loss": 1.2428,
+      "step": 2545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338798175642045e-05,
+      "loss": 1.5813,
+      "step": 2546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336541221137478e-05,
+      "loss": 1.3608,
+      "step": 2547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733428345700713e-05,
+      "loss": 1.4363,
+      "step": 2548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733202488350017e-05,
+      "loss": 1.7805,
+      "step": 2549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329765500865834e-05,
+      "loss": 1.1701,
+      "step": 2550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732750530935346e-05,
+      "loss": 1.5918,
+      "step": 2551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7325244309212476e-05,
+      "loss": 1.6107,
+      "step": 2552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7322982500692387e-05,
+      "loss": 1.5146,
+      "step": 2553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7320719884042802e-05,
+      "loss": 1.6848,
+      "step": 2554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731845645951341e-05,
+      "loss": 1.5652,
+      "step": 2555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731619222735399e-05,
+      "loss": 1.0482,
+      "step": 2556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313927187814415e-05,
+      "loss": 1.2872,
+      "step": 2557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311661341144643e-05,
+      "loss": 1.7554,
+      "step": 2558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730939468759472e-05,
+      "loss": 1.1633,
+      "step": 2559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730712722741479e-05,
+      "loss": 1.5836,
+      "step": 2560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304858960855072e-05,
+      "loss": 1.9601,
+      "step": 2561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302589888165883e-05,
+      "loss": 1.5429,
+      "step": 2562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300320009597628e-05,
+      "loss": 1.4549,
+      "step": 2563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298049325400795e-05,
+      "loss": 1.6812,
+      "step": 2564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295777835825975e-05,
+      "loss": 1.3166,
+      "step": 2565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293505541123834e-05,
+      "loss": 1.5302,
+      "step": 2566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729123244154513e-05,
+      "loss": 1.1688,
+      "step": 2567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288958537340712e-05,
+      "loss": 1.5949,
+      "step": 2568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728668382876152e-05,
+      "loss": 1.5963,
+      "step": 2569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728440831605858e-05,
+      "loss": 1.5161,
+      "step": 2570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282131999483004e-05,
+      "loss": 1.3071,
+      "step": 2571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279854879285995e-05,
+      "loss": 1.4454,
+      "step": 2572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277576955718848e-05,
+      "loss": 1.4222,
+      "step": 2573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275298229032943e-05,
+      "loss": 1.5741,
+      "step": 2574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273018699479748e-05,
+      "loss": 1.7851,
+      "step": 2575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727073836731082e-05,
+      "loss": 1.5932,
+      "step": 2576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268457232777815e-05,
+      "loss": 1.3786,
+      "step": 2577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266175296132455e-05,
+      "loss": 1.842,
+      "step": 2578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263892557626567e-05,
+      "loss": 1.6167,
+      "step": 2579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726160901751207e-05,
+      "loss": 1.6383,
+      "step": 2580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259324676040958e-05,
+      "loss": 1.3044,
+      "step": 2581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257039533465325e-05,
+      "loss": 1.6398,
+      "step": 2582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725475359003734e-05,
+      "loss": 1.6087,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7252466846009282e-05,
+      "loss": 1.5426,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7250179301633497e-05,
+      "loss": 1.4205,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7247890957162427e-05,
+      "loss": 1.6813,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72456018128486e-05,
+      "loss": 1.5715,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.724331186894464e-05,
+      "loss": 1.6312,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7241021125703258e-05,
+      "loss": 1.4437,
+      "step": 2589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723872958337724e-05,
+      "loss": 1.4668,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723643724221948e-05,
+      "loss": 1.6975,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234144102482942e-05,
+      "loss": 1.4192,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7231850164420686e-05,
+      "loss": 1.5636,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229555428285864e-05,
+      "loss": 1.4375,
+      "step": 2594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227259894331712e-05,
+      "loss": 1.139,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224963562811554e-05,
+      "loss": 1.1649,
+      "step": 2596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222666433978798e-05,
+      "loss": 1.4725,
+      "step": 2597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220368508086948e-05,
+      "loss": 1.3335,
+      "step": 2598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7218069785389595e-05,
+      "loss": 1.4863,
+      "step": 2599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721577026614041e-05,
+      "loss": 1.4216,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213469950593156e-05,
+      "loss": 1.4146,
+      "step": 2601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721116883900169e-05,
+      "loss": 1.6166,
+      "step": 2602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720886693161995e-05,
+      "loss": 1.6012,
+      "step": 2603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206564228701962e-05,
+      "loss": 1.7924,
+      "step": 2604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204260730501842e-05,
+      "loss": 1.37,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720195643727379e-05,
+      "loss": 1.4788,
+      "step": 2606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199651349272108e-05,
+      "loss": 1.3886,
+      "step": 2607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7197345466751158e-05,
+      "loss": 2.2109,
+      "step": 2608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195038789965418e-05,
+      "loss": 1.5347,
+      "step": 2609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719273131916944e-05,
+      "loss": 1.535,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190423054617867e-05,
+      "loss": 1.6584,
+      "step": 2611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188113996565418e-05,
+      "loss": 1.6026,
+      "step": 2612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718580414526692e-05,
+      "loss": 1.5443,
+      "step": 2613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183493500977277e-05,
+      "loss": 1.2741,
+      "step": 2614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181182063951477e-05,
+      "loss": 1.2667,
+      "step": 2615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71788698344446e-05,
+      "loss": 2.3035,
+      "step": 2616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.717655681271181e-05,
+      "loss": 1.5433,
+      "step": 2617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7174242999008365e-05,
+      "loss": 1.4618,
+      "step": 2618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7171928393589604e-05,
+      "loss": 1.503,
+      "step": 2619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716961299671096e-05,
+      "loss": 1.5021,
+      "step": 2620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7167296808627948e-05,
+      "loss": 1.4692,
+      "step": 2621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164979829596166e-05,
+      "loss": 1.7899,
+      "step": 2622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7162662059871308e-05,
+      "loss": 1.619,
+      "step": 2623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160343499709154e-05,
+      "loss": 1.0764,
+      "step": 2624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715802414936557e-05,
+      "loss": 1.2129,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71557040090965e-05,
+      "loss": 1.5178,
+      "step": 2626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153383079157995e-05,
+      "loss": 1.42,
+      "step": 2627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151061359806175e-05,
+      "loss": 1.51,
+      "step": 2628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148738851297258e-05,
+      "loss": 1.6107,
+      "step": 2629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714641555388754e-05,
+      "loss": 1.6749,
+      "step": 2630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144091467833413e-05,
+      "loss": 1.9103,
+      "step": 2631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714176659339135e-05,
+      "loss": 1.2883,
+      "step": 2632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139440930817916e-05,
+      "loss": 2.2095,
+      "step": 2633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137114480369753e-05,
+      "loss": 1.6706,
+      "step": 2634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134787242303605e-05,
+      "loss": 1.4047,
+      "step": 2635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713245921687629e-05,
+      "loss": 1.5478,
+      "step": 2636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7130130404344726e-05,
+      "loss": 1.4588,
+      "step": 2637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127800804965895e-05,
+      "loss": 1.5274,
+      "step": 2638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125470418996894e-05,
+      "loss": 1.6291,
+      "step": 2639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7123139246694885e-05,
+      "loss": 1.6095,
+      "step": 2640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120807288317133e-05,
+      "loss": 1.0193,
+      "step": 2641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711847454412097e-05,
+      "loss": 1.5453,
+      "step": 2642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116141014363838e-05,
+      "loss": 1.2776,
+      "step": 2643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113806699303247e-05,
+      "loss": 1.5143,
+      "step": 2644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111471599196807e-05,
+      "loss": 1.1833,
+      "step": 2645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109135714302204e-05,
+      "loss": 1.3277,
+      "step": 2646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7106799044877216e-05,
+      "loss": 1.2405,
+      "step": 2647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104461591179705e-05,
+      "loss": 1.5718,
+      "step": 2648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7102123353467624e-05,
+      "loss": 1.255,
+      "step": 2649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709978433199901e-05,
+      "loss": 1.4142,
+      "step": 2650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7097444527031988e-05,
+      "loss": 1.6964,
+      "step": 2651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7095103938824762e-05,
+      "loss": 1.583,
+      "step": 2652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092762567635636e-05,
+      "loss": 1.867,
+      "step": 2653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709042041372298e-05,
+      "loss": 1.3146,
+      "step": 2654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088077477345275e-05,
+      "loss": 1.4792,
+      "step": 2655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708573375876107e-05,
+      "loss": 1.1793,
+      "step": 2656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083389258229012e-05,
+      "loss": 1.4962,
+      "step": 2657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081043976007827e-05,
+      "loss": 1.6171,
+      "step": 2658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078697912356323e-05,
+      "loss": 1.2432,
+      "step": 2659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707635106753341e-05,
+      "loss": 1.2394,
+      "step": 2660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074003441798065e-05,
+      "loss": 1.4717,
+      "step": 2661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707165503540937e-05,
+      "loss": 1.3515,
+      "step": 2662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7069305848626477e-05,
+      "loss": 1.6137,
+      "step": 2663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066955881708634e-05,
+      "loss": 1.3117,
+      "step": 2664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064605134915174e-05,
+      "loss": 1.5284,
+      "step": 2665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706225360850551e-05,
+      "loss": 1.9393,
+      "step": 2666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705990130273915e-05,
+      "loss": 1.8731,
+      "step": 2667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057548217875677e-05,
+      "loss": 1.5723,
+      "step": 2668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705519435417477e-05,
+      "loss": 1.6854,
+      "step": 2669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052839711896193e-05,
+      "loss": 1.8749,
+      "step": 2670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705048429129979e-05,
+      "loss": 1.3127,
+      "step": 2671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048128092645493e-05,
+      "loss": 1.7564,
+      "step": 2672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045771116193323e-05,
+      "loss": 1.6874,
+      "step": 2673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043413362203383e-05,
+      "loss": 1.3207,
+      "step": 2674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041054830935865e-05,
+      "loss": 1.4219,
+      "step": 2675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038695522651044e-05,
+      "loss": 1.6235,
+      "step": 2676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036335437609283e-05,
+      "loss": 1.3084,
+      "step": 2677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703397457607103e-05,
+      "loss": 1.7362,
+      "step": 2678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031612938296818e-05,
+      "loss": 1.6267,
+      "step": 2679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029250524547267e-05,
+      "loss": 1.3341,
+      "step": 2680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702688733508308e-05,
+      "loss": 1.3559,
+      "step": 2681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7024523370165042e-05,
+      "loss": 1.651,
+      "step": 2682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7022158630054043e-05,
+      "loss": 1.2955,
+      "step": 2683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7019793115011035e-05,
+      "loss": 1.6148,
+      "step": 2684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701742682529706e-05,
+      "loss": 2.1154,
+      "step": 2685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7015059761173263e-05,
+      "loss": 1.4363,
+      "step": 2686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012691922900854e-05,
+      "loss": 1.5751,
+      "step": 2687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701032331074114e-05,
+      "loss": 1.6303,
+      "step": 2688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7007953924955508e-05,
+      "loss": 1.5978,
+      "step": 2689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700558376580543e-05,
+      "loss": 1.6747,
+      "step": 2690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700321283355247e-05,
+      "loss": 1.2924,
+      "step": 2691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000841128458266e-05,
+      "loss": 1.8293,
+      "step": 2692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6998468650784554e-05,
+      "loss": 1.0779,
+      "step": 2693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699609540079315e-05,
+      "loss": 1.5554,
+      "step": 2694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699372137874595e-05,
+      "loss": 1.2938,
+      "step": 2695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699134658490494e-05,
+      "loss": 1.4695,
+      "step": 2696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69889710195322e-05,
+      "loss": 1.5279,
+      "step": 2697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986594682889876e-05,
+      "loss": 1.5286,
+      "step": 2698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984217575240212e-05,
+      "loss": 1.7901,
+      "step": 2699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981839696845536e-05,
+      "loss": 1.3658,
+      "step": 2700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979461047968254e-05,
+      "loss": 1.4042,
+      "step": 2701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977081628870872e-05,
+      "loss": 1.3531,
+      "step": 2702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697470143981596e-05,
+      "loss": 1.5139,
+      "step": 2703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972320481066194e-05,
+      "loss": 1.361,
+      "step": 2704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696993875288432e-05,
+      "loss": 1.3834,
+      "step": 2705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967556255533174e-05,
+      "loss": 1.4932,
+      "step": 2706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965172989275683e-05,
+      "loss": 1.5791,
+      "step": 2707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962788954374844e-05,
+      "loss": 1.529,
+      "step": 2708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6960404151093756e-05,
+      "loss": 1.4749,
+      "step": 2709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695801857969559e-05,
+      "loss": 1.5009,
+      "step": 2710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955632240443606e-05,
+      "loss": 1.1918,
+      "step": 2711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953245133601147e-05,
+      "loss": 1.3971,
+      "step": 2712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695085725943165e-05,
+      "loss": 1.4516,
+      "step": 2713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6948468618198622e-05,
+      "loss": 1.4401,
+      "step": 2714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6946079210165672e-05,
+      "loss": 1.5602,
+      "step": 2715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.694368903559647e-05,
+      "loss": 1.6268,
+      "step": 2716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6941298094754795e-05,
+      "loss": 2.0233,
+      "step": 2717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6938906387904496e-05,
+      "loss": 1.3965,
+      "step": 2718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693651391530951e-05,
+      "loss": 1.3186,
+      "step": 2719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934120677233862e-05,
+      "loss": 1.3476,
+      "step": 2720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931726673941656e-05,
+      "loss": 1.7484,
+      "step": 2721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692933190569708e-05,
+      "loss": 1.4424,
+      "step": 2722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692693637276441e-05,
+      "loss": 1.3637,
+      "step": 2723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924540075408013e-05,
+      "loss": 1.8014,
+      "step": 2724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6922143013892322e-05,
+      "loss": 1.4208,
+      "step": 2725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919745188481876e-05,
+      "loss": 1.9145,
+      "step": 2726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6917346599441277e-05,
+      "loss": 1.4364,
+      "step": 2727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914947247035233e-05,
+      "loss": 1.4497,
+      "step": 2728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912547131528516e-05,
+      "loss": 1.5221,
+      "step": 2729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910146253185997e-05,
+      "loss": 1.35,
+      "step": 2730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690774461227262e-05,
+      "loss": 1.4372,
+      "step": 2731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905342209053427e-05,
+      "loss": 1.6709,
+      "step": 2732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6902939043793525e-05,
+      "loss": 1.7087,
+      "step": 2733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900535116758126e-05,
+      "loss": 1.7268,
+      "step": 2734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6898130428212508e-05,
+      "loss": 1.8736,
+      "step": 2735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6895724978422046e-05,
+      "loss": 1.5306,
+      "step": 2736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893318767652193e-05,
+      "loss": 1.6331,
+      "step": 2737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890911796168485e-05,
+      "loss": 1.3594,
+      "step": 2738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688850406423655e-05,
+      "loss": 1.3277,
+      "step": 2739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886095572122082e-05,
+      "loss": 1.4838,
+      "step": 2740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883686320090885e-05,
+      "loss": 1.5405,
+      "step": 2741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688127630840882e-05,
+      "loss": 1.7264,
+      "step": 2742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687886553734185e-05,
+      "loss": 1.2663,
+      "step": 2743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687645400715602e-05,
+      "loss": 1.4966,
+      "step": 2744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687404171811745e-05,
+      "loss": 1.644,
+      "step": 2745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687162867049235e-05,
+      "loss": 1.376,
+      "step": 2746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6869214864547013e-05,
+      "loss": 1.2963,
+      "step": 2747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6866800300547814e-05,
+      "loss": 1.9046,
+      "step": 2748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.686438497876121e-05,
+      "loss": 1.3792,
+      "step": 2749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.686196889945375e-05,
+      "loss": 1.5598,
+      "step": 2750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6859552062892063e-05,
+      "loss": 1.7399,
+      "step": 2751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685713446934285e-05,
+      "loss": 1.4359,
+      "step": 2752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685471611907291e-05,
+      "loss": 1.6123,
+      "step": 2753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852297012349127e-05,
+      "loss": 2.0056,
+      "step": 2754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849877149438448e-05,
+      "loss": 1.627,
+      "step": 2755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847456530607934e-05,
+      "loss": 1.5971,
+      "step": 2756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845035156124697e-05,
+      "loss": 1.2742,
+      "step": 2757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684261302625596e-05,
+      "loss": 1.2212,
+      "step": 2758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840190141269013e-05,
+      "loss": 1.3101,
+      "step": 2759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837766501431237e-05,
+      "loss": 1.0172,
+      "step": 2760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683534210701009e-05,
+      "loss": 1.4029,
+      "step": 2761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832916958273117e-05,
+      "loss": 1.1171,
+      "step": 2762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683049105548795e-05,
+      "loss": 1.3143,
+      "step": 2763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828064398922293e-05,
+      "loss": 1.3541,
+      "step": 2764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682563698884394e-05,
+      "loss": 1.1695,
+      "step": 2765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682320882552078e-05,
+      "loss": 1.5996,
+      "step": 2766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6820779909220767e-05,
+      "loss": 1.2796,
+      "step": 2767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818350240211935e-05,
+      "loss": 1.2876,
+      "step": 2768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6815919818762428e-05,
+      "loss": 1.8258,
+      "step": 2769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813488645140443e-05,
+      "loss": 1.512,
+      "step": 2770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811056719614274e-05,
+      "loss": 1.6515,
+      "step": 2771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808624042452305e-05,
+      "loss": 1.4887,
+      "step": 2772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680619061392299e-05,
+      "loss": 1.6002,
+      "step": 2773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803756434294866e-05,
+      "loss": 1.682,
+      "step": 2774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680132150383656e-05,
+      "loss": 1.819,
+      "step": 2775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798885822816785e-05,
+      "loss": 1.7961,
+      "step": 2776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796449391504325e-05,
+      "loss": 1.4615,
+      "step": 2777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679401221016806e-05,
+      "loss": 1.3235,
+      "step": 2778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6791574279076928e-05,
+      "loss": 1.3706,
+      "step": 2779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.678913559849999e-05,
+      "loss": 1.5812,
+      "step": 2780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6786696168706356e-05,
+      "loss": 1.4944,
+      "step": 2781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.678425598996523e-05,
+      "loss": 1.4179,
+      "step": 2782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6781815062545902e-05,
+      "loss": 1.4899,
+      "step": 2783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6779373386717734e-05,
+      "loss": 1.4313,
+      "step": 2784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776930962750185e-05,
+      "loss": 1.4914,
+      "step": 2785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774487790912787e-05,
+      "loss": 1.5834,
+      "step": 2786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6772043871475156e-05,
+      "loss": 1.0747,
+      "step": 2787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769599204706993e-05,
+      "loss": 1.3226,
+      "step": 2788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767153790878077e-05,
+      "loss": 1.8242,
+      "step": 2789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676470763025827e-05,
+      "loss": 1.4297,
+      "step": 2790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676226072311753e-05,
+      "loss": 1.5902,
+      "step": 2791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6759813069725876e-05,
+      "loss": 1.5821,
+      "step": 2792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757364670353423e-05,
+      "loss": 1.588,
+      "step": 2793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675491552527036e-05,
+      "loss": 1.3459,
+      "step": 2794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675246563474697e-05,
+      "loss": 1.7067,
+      "step": 2795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750014999053605e-05,
+      "loss": 1.1113,
+      "step": 2796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674756361846071e-05,
+      "loss": 1.4783,
+      "step": 2797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6745111493238808e-05,
+      "loss": 1.6437,
+      "step": 2798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67426586236585e-05,
+      "loss": 1.7149,
+      "step": 2799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740205009990476e-05,
+      "loss": 1.4864,
+      "step": 2800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67377506525055e-05,
+      "loss": 1.2707,
+      "step": 2801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735295551474433e-05,
+      "loss": 1.643,
+      "step": 2802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67328397071682e-05,
+      "loss": 1.3629,
+      "step": 2803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730383119857817e-05,
+      "loss": 1.8541,
+      "step": 2804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672792578981439e-05,
+      "loss": 1.0985,
+      "step": 2805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725467717309083e-05,
+      "loss": 1.3303,
+      "step": 2806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672300890261317e-05,
+      "loss": 1.515,
+      "step": 2807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6720549345997988e-05,
+      "loss": 1.4389,
+      "step": 2808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718089047734967e-05,
+      "loss": 1.7931,
+      "step": 2809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715628008095605e-05,
+      "loss": 1.4195,
+      "step": 2810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713166227351498e-05,
+      "loss": 1.689,
+      "step": 2811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6710703705774317e-05,
+      "loss": 1.3903,
+      "step": 2812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.670824044363581e-05,
+      "loss": 1.3598,
+      "step": 2813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6705776441207813e-05,
+      "loss": 1.4164,
+      "step": 2814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.670331169876224e-05,
+      "loss": 1.4129,
+      "step": 2815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6700846216571092e-05,
+      "loss": 1.3876,
+      "step": 2816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698379994906443e-05,
+      "loss": 1.4282,
+      "step": 2817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695913034040454e-05,
+      "loss": 1.5595,
+      "step": 2818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693445334245372e-05,
+      "loss": 1.369,
+      "step": 2819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690976895793515e-05,
+      "loss": 1.3641,
+      "step": 2820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688507718957294e-05,
+      "loss": 1.5494,
+      "step": 2821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6686037804009186e-05,
+      "loss": 1.6641,
+      "step": 2822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668356715122177e-05,
+      "loss": 1.381,
+      "step": 2823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668109576086769e-05,
+      "loss": 1.3843,
+      "step": 2824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678623633219676e-05,
+      "loss": 1.2713,
+      "step": 2825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667615076855054e-05,
+      "loss": 1.4897,
+      "step": 2826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673677167133183e-05,
+      "loss": 1.4331,
+      "step": 2827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671202829240572e-05,
+      "loss": 1.6962,
+      "step": 2828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668727755145765e-05,
+      "loss": 1.8388,
+      "step": 2829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66662519451219e-05,
+      "loss": 1.3498,
+      "step": 2830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6663775399442194e-05,
+      "loss": 1.6855,
+      "step": 2831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666129811837995e-05,
+      "loss": 1.5562,
+      "step": 2832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658820102208547e-05,
+      "loss": 1.4055,
+      "step": 2833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656341351201446e-05,
+      "loss": 1.4284,
+      "step": 2834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665386186563219e-05,
+      "loss": 1.3857,
+      "step": 2835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651381645774408e-05,
+      "loss": 1.5889,
+      "step": 2836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66489006919018e-05,
+      "loss": 1.5154,
+      "step": 2837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646419004288154e-05,
+      "loss": 1.2762,
+      "step": 2838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643936583207338e-05,
+      "loss": 1.8343,
+      "step": 2839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6641453428933298e-05,
+      "loss": 1.1211,
+      "step": 2840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663896954174007e-05,
+      "loss": 1.8168,
+      "step": 2841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663648492190175e-05,
+      "loss": 1.5715,
+      "step": 2842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663399956969254e-05,
+      "loss": 1.6234,
+      "step": 2843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6631513485386715e-05,
+      "loss": 1.5751,
+      "step": 2844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6629026669258618e-05,
+      "loss": 1.7916,
+      "step": 2845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6626539121582687e-05,
+      "loss": 1.6646,
+      "step": 2846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6624050842633433e-05,
+      "loss": 1.4822,
+      "step": 2847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.662156183268545e-05,
+      "loss": 1.5519,
+      "step": 2848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6619072092013418e-05,
+      "loss": 1.4174,
+      "step": 2849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661658162089209e-05,
+      "loss": 1.6802,
+      "step": 2850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614090419596304e-05,
+      "loss": 1.6792,
+      "step": 2851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6611598488400973e-05,
+      "loss": 1.2709,
+      "step": 2852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66091058275811e-05,
+      "loss": 1.3901,
+      "step": 2853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660661243741176e-05,
+      "loss": 1.0957,
+      "step": 2854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604118318168115e-05,
+      "loss": 1.7142,
+      "step": 2855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66016234701254e-05,
+      "loss": 1.661,
+      "step": 2856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599127893558934e-05,
+      "loss": 1.1411,
+      "step": 2857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6596631588744126e-05,
+      "loss": 1.3974,
+      "step": 2858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594134555956445e-05,
+      "loss": 1.3089,
+      "step": 2859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591636795471454e-05,
+      "loss": 1.3976,
+      "step": 2860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65891383075648e-05,
+      "loss": 1.406,
+      "step": 2861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65866390925122e-05,
+      "loss": 1.6816,
+      "step": 2862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6584139150589457e-05,
+      "loss": 1.2265,
+      "step": 2863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658163848207245e-05,
+      "loss": 1.6566,
+      "step": 2864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657913708723714e-05,
+      "loss": 1.2093,
+      "step": 2865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576634966359574e-05,
+      "loss": 1.4729,
+      "step": 2866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657413211971587e-05,
+      "loss": 2.1208,
+      "step": 2867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571628547582235e-05,
+      "loss": 1.3627,
+      "step": 2868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569124250234948e-05,
+      "loss": 1.2227,
+      "step": 2869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656661922795037e-05,
+      "loss": 1.5487,
+      "step": 2870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656411348100494e-05,
+      "loss": 1.7662,
+      "step": 2871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561607009675197e-05,
+      "loss": 2.72,
+      "step": 2872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655909981423772e-05,
+      "loss": 1.8003,
+      "step": 2873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655659189496921e-05,
+      "loss": 1.2893,
+      "step": 2874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655408325214642e-05,
+      "loss": 1.6084,
+      "step": 2875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655157388604619e-05,
+      "loss": 1.4859,
+      "step": 2876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.654906379694545e-05,
+      "loss": 1.9196,
+      "step": 2877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6546552985121195e-05,
+      "loss": 1.3492,
+      "step": 2878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6544041450850506e-05,
+      "loss": 1.4346,
+      "step": 2879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6541529194410548e-05,
+      "loss": 1.5927,
+      "step": 2880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6539016216078558e-05,
+      "loss": 1.3022,
+      "step": 2881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653650251613186e-05,
+      "loss": 1.4853,
+      "step": 2882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653398809484785e-05,
+      "loss": 1.4379,
+      "step": 2883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531472952504012e-05,
+      "loss": 1.7671,
+      "step": 2884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65289570893779e-05,
+      "loss": 1.3458,
+      "step": 2885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526440505747158e-05,
+      "loss": 1.7491,
+      "step": 2886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65239232018895e-05,
+      "loss": 1.3293,
+      "step": 2887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521405178082722e-05,
+      "loss": 1.2657,
+      "step": 2888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6518886434604708e-05,
+      "loss": 1.3489,
+      "step": 2889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651636697173341e-05,
+      "loss": 1.5196,
+      "step": 2890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513846789746866e-05,
+      "loss": 1.5233,
+      "step": 2891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511325888923186e-05,
+      "loss": 1.5177,
+      "step": 2892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508804269540573e-05,
+      "loss": 1.4502,
+      "step": 2893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506281931877296e-05,
+      "loss": 1.4762,
+      "step": 2894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650375887621171e-05,
+      "loss": 1.4592,
+      "step": 2895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650123510282224e-05,
+      "loss": 1.5499,
+      "step": 2896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649871061198741e-05,
+      "loss": 1.4489,
+      "step": 2897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64961854039858e-05,
+      "loss": 1.4679,
+      "step": 2898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493659479096088e-05,
+      "loss": 1.2684,
+      "step": 2899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6491132837597017e-05,
+      "loss": 1.4501,
+      "step": 2900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648860547976742e-05,
+      "loss": 1.6702,
+      "step": 2901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64860774058862e-05,
+      "loss": 2.2821,
+      "step": 2902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483548616232346e-05,
+      "loss": 1.4727,
+      "step": 2903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648101911108492e-05,
+      "loss": 1.363,
+      "step": 2904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647848889072307e-05,
+      "loss": 1.6591,
+      "step": 2905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647595795542602e-05,
+      "loss": 1.6207,
+      "step": 2906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647342630547307e-05,
+      "loss": 1.405,
+      "step": 2907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470893941143596e-05,
+      "loss": 1.6451,
+      "step": 2908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6468360862717067e-05,
+      "loss": 1.3386,
+      "step": 2909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.646582707047301e-05,
+      "loss": 1.6269,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6463292564691052e-05,
+      "loss": 1.6399,
+      "step": 2911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6460757345650885e-05,
+      "loss": 1.5858,
+      "step": 2912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6458221413632285e-05,
+      "loss": 1.5073,
+      "step": 2913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645568476891511e-05,
+      "loss": 1.3963,
+      "step": 2914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645314741177928e-05,
+      "loss": 1.4385,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6450609342504812e-05,
+      "loss": 1.4483,
+      "step": 2916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64480705613718e-05,
+      "loss": 1.4574,
+      "step": 2917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445531068660404e-05,
+      "loss": 1.6007,
+      "step": 2918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442990864650878e-05,
+      "loss": 1.5663,
+      "step": 2919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644044994962354e-05,
+      "loss": 1.5154,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437908323858794e-05,
+      "loss": 1.3323,
+      "step": 2921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643536598763713e-05,
+      "loss": 1.5199,
+      "step": 2922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432822941239094e-05,
+      "loss": 1.8364,
+      "step": 2923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430279184945338e-05,
+      "loss": 2.2247,
+      "step": 2924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6427734719036574e-05,
+      "loss": 1.5098,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642518954379359e-05,
+      "loss": 1.4725,
+      "step": 2926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642264365949727e-05,
+      "loss": 1.6172,
+      "step": 2927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420097066428563e-05,
+      "loss": 1.7755,
+      "step": 2928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417549764868496e-05,
+      "loss": 1.3917,
+      "step": 2929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415001755098177e-05,
+      "loss": 1.5254,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412453037398796e-05,
+      "loss": 1.6671,
+      "step": 2931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409903612051613e-05,
+      "loss": 1.4759,
+      "step": 2932
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407353479337974e-05,
+      "loss": 1.4903,
+      "step": 2933
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404802639539296e-05,
+      "loss": 1.8393,
+      "step": 2934
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640225109293708e-05,
+      "loss": 1.5291,
+      "step": 2935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399698839812902e-05,
+      "loss": 1.875,
+      "step": 2936
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6397145880448416e-05,
+      "loss": 2.1431,
+      "step": 2937
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394592215125357e-05,
+      "loss": 1.5959,
+      "step": 2938
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392037844125533e-05,
+      "loss": 1.7625,
+      "step": 2939
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389482767730833e-05,
+      "loss": 1.3268,
+      "step": 2940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638692698622322e-05,
+      "loss": 1.7383,
+      "step": 2941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6384370499884744e-05,
+      "loss": 1.4441,
+      "step": 2942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.638181330899752e-05,
+      "loss": 1.2817,
+      "step": 2943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637925541384375e-05,
+      "loss": 1.3823,
+      "step": 2944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6376696814705716e-05,
+      "loss": 1.4676,
+      "step": 2945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6374137511865765e-05,
+      "loss": 1.8602,
+      "step": 2946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371577505606334e-05,
+      "loss": 1.4812,
+      "step": 2947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369016796209933e-05,
+      "loss": 1.62,
+      "step": 2948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636645538395915e-05,
+      "loss": 1.5496,
+      "step": 2949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363893269136647e-05,
+      "loss": 1.1249,
+      "step": 2950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636133045202517e-05,
+      "loss": 1.7057,
+      "step": 2951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635876693290754e-05,
+      "loss": 1.4416,
+      "step": 2952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635620271206665e-05,
+      "loss": 1.8708,
+      "step": 2953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353637789785477e-05,
+      "loss": 1.5029,
+      "step": 2954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6351072166347076e-05,
+      "loss": 1.564,
+      "step": 2955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634850584203458e-05,
+      "loss": 1.5388,
+      "step": 2956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345938817131186e-05,
+      "loss": 1.5228,
+      "step": 2957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634337109192019e-05,
+      "loss": 1.537,
+      "step": 2958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340802666684945e-05,
+      "loss": 1.8487,
+      "step": 2959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6338233541708896e-05,
+      "loss": 1.4454,
+      "step": 2960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335663717275556e-05,
+      "loss": 1.5763,
+      "step": 2961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633309319366852e-05,
+      "loss": 1.7982,
+      "step": 2962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633052197117146e-05,
+      "loss": 1.5286,
+      "step": 2963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632795005006812e-05,
+      "loss": 1.439,
+      "step": 2964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632537743064233e-05,
+      "loss": 1.6088,
+      "step": 2965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322804113177988e-05,
+      "loss": 1.5864,
+      "step": 2966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320230097959074e-05,
+      "loss": 1.3793,
+      "step": 2967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317655385269648e-05,
+      "loss": 1.462,
+      "step": 2968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6315079975393836e-05,
+      "loss": 1.6197,
+      "step": 2969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312503868615857e-05,
+      "loss": 1.1337,
+      "step": 2970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309927065219987e-05,
+      "loss": 1.8076,
+      "step": 2971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63073495654906e-05,
+      "loss": 1.3758,
+      "step": 2972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630477136971213e-05,
+      "loss": 1.6526,
+      "step": 2973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6302192478169095e-05,
+      "loss": 1.5412,
+      "step": 2974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629961289114609e-05,
+      "loss": 1.5492,
+      "step": 2975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629703260892779e-05,
+      "loss": 2.0298,
+      "step": 2976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629445163179894e-05,
+      "loss": 1.3555,
+      "step": 2977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629186996004436e-05,
+      "loss": 1.6746,
+      "step": 2978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6289287593948953e-05,
+      "loss": 1.6202,
+      "step": 2979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62867045337977e-05,
+      "loss": 1.2116,
+      "step": 2980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284120779875656e-05,
+      "loss": 1.3967,
+      "step": 2981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281536332467945e-05,
+      "loss": 1.5216,
+      "step": 2982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278951191859783e-05,
+      "loss": 1.6179,
+      "step": 2983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276365358336444e-05,
+      "loss": 1.535,
+      "step": 2984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6273778832183297e-05,
+      "loss": 1.3103,
+      "step": 2985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627119161368578e-05,
+      "loss": 1.4566,
+      "step": 2986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6268603703129392e-05,
+      "loss": 1.3809,
+      "step": 2987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626601510079974e-05,
+      "loss": 1.5269,
+      "step": 2988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263425806982474e-05,
+      "loss": 1.3521,
+      "step": 2989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626083582196335e-05,
+      "loss": 1.7644,
+      "step": 2990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258245146028178e-05,
+      "loss": 1.7424,
+      "step": 2991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255653779462858e-05,
+      "loss": 1.4557,
+      "step": 2992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253061722553353e-05,
+      "loss": 1.398,
+      "step": 2993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250468975585717e-05,
+      "loss": 1.5957,
+      "step": 2994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6247875538846073e-05,
+      "loss": 1.5172,
+      "step": 2995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245281412620614e-05,
+      "loss": 1.4102,
+      "step": 2996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242686597195624e-05,
+      "loss": 1.6782,
+      "step": 2997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624009109285745e-05,
+      "loss": 1.613,
+      "step": 2998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623749489989252e-05,
+      "loss": 1.2662,
+      "step": 2999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.3338,
+      "step": 3000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623230044922848e-05,
+      "loss": 1.4012,
+      "step": 3001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229702192102607e-05,
+      "loss": 1.4414,
+      "step": 3002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227103247496448e-05,
+      "loss": 1.5481,
+      "step": 3003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6224503615696806e-05,
+      "loss": 1.2664,
+      "step": 3004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221903296990568e-05,
+      "loss": 1.3535,
+      "step": 3005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219302291664697e-05,
+      "loss": 1.3601,
+      "step": 3006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.621670060000622e-05,
+      "loss": 1.5586,
+      "step": 3007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.621409822230225e-05,
+      "loss": 1.8586,
+      "step": 3008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6211495158839974e-05,
+      "loss": 1.2973,
+      "step": 3009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6208891409906655e-05,
+      "loss": 1.5558,
+      "step": 3010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6206286975789625e-05,
+      "loss": 1.5115,
+      "step": 3011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203681856776303e-05,
+      "loss": 1.3612,
+      "step": 3012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201076053154174e-05,
+      "loss": 1.9438,
+      "step": 3013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6198469565210807e-05,
+      "loss": 1.267,
+      "step": 3014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619586239323383e-05,
+      "loss": 1.3242,
+      "step": 3015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619325453751097e-05,
+      "loss": 1.3807,
+      "step": 3016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190645998330013e-05,
+      "loss": 1.6454,
+      "step": 3017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6188036775978826e-05,
+      "loss": 1.1699,
+      "step": 3018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618542687074535e-05,
+      "loss": 1.4139,
+      "step": 3019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61828162829176e-05,
+      "loss": 1.2853,
+      "step": 3020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618020501278367e-05,
+      "loss": 1.853,
+      "step": 3021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177593060631726e-05,
+      "loss": 1.6466,
+      "step": 3022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617498042675001e-05,
+      "loss": 1.7478,
+      "step": 3023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172367111426846e-05,
+      "loss": 1.4195,
+      "step": 3024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169753114950618e-05,
+      "loss": 1.544,
+      "step": 3025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6167138437609798e-05,
+      "loss": 1.4509,
+      "step": 3026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616452307969293e-05,
+      "loss": 1.5346,
+      "step": 3027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161907041488637e-05,
+      "loss": 1.7875,
+      "step": 3028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159290323285604e-05,
+      "loss": 1.3505,
+      "step": 3029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156672925372603e-05,
+      "loss": 1.361,
+      "step": 3030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615405484803848e-05,
+      "loss": 1.2778,
+      "step": 3031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151436091572147e-05,
+      "loss": 1.2804,
+      "step": 3032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6148816656262605e-05,
+      "loss": 1.6883,
+      "step": 3033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614619654239892e-05,
+      "loss": 1.2637,
+      "step": 3034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143575750270232e-05,
+      "loss": 1.5428,
+      "step": 3035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140954280165767e-05,
+      "loss": 1.9922,
+      "step": 3036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613833213237481e-05,
+      "loss": 1.7038,
+      "step": 3037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613570930718673e-05,
+      "loss": 1.5089,
+      "step": 3038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.613308580489097e-05,
+      "loss": 1.7882,
+      "step": 3039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.613046162577705e-05,
+      "loss": 1.5388,
+      "step": 3040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6127836770134563e-05,
+      "loss": 1.5712,
+      "step": 3041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612521123825317e-05,
+      "loss": 1.2649,
+      "step": 3042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6122585030422614e-05,
+      "loss": 1.1286,
+      "step": 3043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611995814693271e-05,
+      "loss": 1.6546,
+      "step": 3044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117330588073353e-05,
+      "loss": 1.6018,
+      "step": 3045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611470235413451e-05,
+      "loss": 1.3951,
+      "step": 3046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112073445406208e-05,
+      "loss": 1.6533,
+      "step": 3047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610944386217857e-05,
+      "loss": 1.7879,
+      "step": 3048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106813604741783e-05,
+      "loss": 1.3803,
+      "step": 3049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610418267338611e-05,
+      "loss": 1.8864,
+      "step": 3050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101551068401886e-05,
+      "loss": 1.3975,
+      "step": 3051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098918790079524e-05,
+      "loss": 2.2429,
+      "step": 3052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609628583870951e-05,
+      "loss": 1.6041,
+      "step": 3053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093652214582404e-05,
+      "loss": 1.3492,
+      "step": 3054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091017917988842e-05,
+      "loss": 1.445,
+      "step": 3055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088382949219525e-05,
+      "loss": 1.8816,
+      "step": 3056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6085747308565242e-05,
+      "loss": 1.3542,
+      "step": 3057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608311099631685e-05,
+      "loss": 1.4003,
+      "step": 3058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080474012765276e-05,
+      "loss": 1.5289,
+      "step": 3059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607783635820153e-05,
+      "loss": 1.4755,
+      "step": 3060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075198032916686e-05,
+      "loss": 1.4054,
+      "step": 3061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072559037201898e-05,
+      "loss": 1.6104,
+      "step": 3062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069919371348398e-05,
+      "loss": 1.393,
+      "step": 3063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067279035647484e-05,
+      "loss": 1.4883,
+      "step": 3064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064638030390525e-05,
+      "loss": 1.6516,
+      "step": 3065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606199635586898e-05,
+      "loss": 1.5675,
+      "step": 3066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059354012374364e-05,
+      "loss": 1.5839,
+      "step": 3067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056711000198275e-05,
+      "loss": 1.1615,
+      "step": 3068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054067319632385e-05,
+      "loss": 1.3329,
+      "step": 3069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605142297096844e-05,
+      "loss": 1.5609,
+      "step": 3070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048777954498252e-05,
+      "loss": 1.62,
+      "step": 3071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6046132270513717e-05,
+      "loss": 1.6605,
+      "step": 3072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6043485919306795e-05,
+      "loss": 1.339,
+      "step": 3073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6040838901169535e-05,
+      "loss": 1.8774,
+      "step": 3074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603819121639404e-05,
+      "loss": 1.6599,
+      "step": 3075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603554286527249e-05,
+      "loss": 1.518,
+      "step": 3076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6032893848097164e-05,
+      "loss": 1.6304,
+      "step": 3077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6030244165160378e-05,
+      "loss": 1.2053,
+      "step": 3078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027593816754545e-05,
+      "loss": 1.6186,
+      "step": 3079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024942803172146e-05,
+      "loss": 1.6406,
+      "step": 3080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602229112470573e-05,
+      "loss": 1.3557,
+      "step": 3081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019638781647925e-05,
+      "loss": 1.3879,
+      "step": 3082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016985774291436e-05,
+      "loss": 1.3277,
+      "step": 3083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601433210292903e-05,
+      "loss": 1.3691,
+      "step": 3084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011677767853556e-05,
+      "loss": 1.2041,
+      "step": 3085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600902276935793e-05,
+      "loss": 1.5628,
+      "step": 3086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006367107735153e-05,
+      "loss": 1.611,
+      "step": 3087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003710783278286e-05,
+      "loss": 1.6174,
+      "step": 3088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6001053796280468e-05,
+      "loss": 1.4434,
+      "step": 3089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5998396147034913e-05,
+      "loss": 1.5297,
+      "step": 3090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995737835834905e-05,
+      "loss": 1.243,
+      "step": 3091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5993078862973803e-05,
+      "loss": 1.6612,
+      "step": 3092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599041922874504e-05,
+      "loss": 1.6805,
+      "step": 3093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598775893344212e-05,
+      "loss": 1.4154,
+      "step": 3094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985097977358618e-05,
+      "loss": 1.3952,
+      "step": 3095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982436360788188e-05,
+      "loss": 1.7138,
+      "step": 3096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597977408402455e-05,
+      "loss": 1.4765,
+      "step": 3097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5977111147361507e-05,
+      "loss": 1.5617,
+      "step": 3098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597444755109292e-05,
+      "loss": 1.5946,
+      "step": 3099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971783295512735e-05,
+      "loss": 1.3555,
+      "step": 3100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5969118380914964e-05,
+      "loss": 1.7299,
+      "step": 3101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5966452807593696e-05,
+      "loss": 1.5125,
+      "step": 3102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5963786575843094e-05,
+      "loss": 1.5169,
+      "step": 3103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5961119685957385e-05,
+      "loss": 1.4961,
+      "step": 3104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5958452138230876e-05,
+      "loss": 1.464,
+      "step": 3105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.595578393295795e-05,
+      "loss": 1.4319,
+      "step": 3106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.595311507043305e-05,
+      "loss": 1.2028,
+      "step": 3107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59504455509507e-05,
+      "loss": 1.5285,
+      "step": 3108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5947775374805498e-05,
+      "loss": 1.0861,
+      "step": 3109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5945104542292114e-05,
+      "loss": 1.3426,
+      "step": 3110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942433053705282e-05,
+      "loss": 1.7142,
+      "step": 3111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939760909339823e-05,
+      "loss": 1.6345,
+      "step": 3112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937088109490614e-05,
+      "loss": 1.4235,
+      "step": 3113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593441465445262e-05,
+      "loss": 2.475,
+      "step": 3114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931740544520864e-05,
+      "loss": 1.4022,
+      "step": 3115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929065779990453e-05,
+      "loss": 1.4021,
+      "step": 3116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5926390361156556e-05,
+      "loss": 1.5715,
+      "step": 3117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923714288314426e-05,
+      "loss": 1.2922,
+      "step": 3118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5921037561759376e-05,
+      "loss": 1.5212,
+      "step": 3119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59183601817868e-05,
+      "loss": 1.5651,
+      "step": 3120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591568214869216e-05,
+      "loss": 2.4885,
+      "step": 3121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5913003462770992e-05,
+      "loss": 1.1201,
+      "step": 3122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910324124318902e-05,
+      "loss": 1.501,
+      "step": 3123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907644133631567e-05,
+      "loss": 1.2496,
+      "step": 3124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904963491004745e-05,
+      "loss": 1.4059,
+      "step": 3125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590228219673425e-05,
+      "loss": 1.7443,
+      "step": 3126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899600251115985e-05,
+      "loss": 1.7484,
+      "step": 3127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589691765444591e-05,
+      "loss": 1.5321,
+      "step": 3128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5894234407020067e-05,
+      "loss": 1.3384,
+      "step": 3129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891550509134566e-05,
+      "loss": 1.1974,
+      "step": 3130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888865961085585e-05,
+      "loss": 1.6586,
+      "step": 3131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886180763169387e-05,
+      "loss": 1.6554,
+      "step": 3132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588349491568229e-05,
+      "loss": 1.5024,
+      "step": 3133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880808418920694e-05,
+      "loss": 1.4205,
+      "step": 3134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878121273181072e-05,
+      "loss": 1.4622,
+      "step": 3135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587543347875996e-05,
+      "loss": 1.5521,
+      "step": 3136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5872745035953965e-05,
+      "loss": 1.4637,
+      "step": 3137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5870055945059778e-05,
+      "loss": 1.3945,
+      "step": 3138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5867366206374157e-05,
+      "loss": 1.1749,
+      "step": 3139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586467582019392e-05,
+      "loss": 1.4299,
+      "step": 3140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5861984786815977e-05,
+      "loss": 1.5387,
+      "step": 3141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859293106537283e-05,
+      "loss": 1.8827,
+      "step": 3142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856600779654888e-05,
+      "loss": 1.729,
+      "step": 3143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853907806465903e-05,
+      "loss": 1.185,
+      "step": 3144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5851214187267516e-05,
+      "loss": 1.6733,
+      "step": 3145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848519922356973e-05,
+      "loss": 1.2115,
+      "step": 3146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5845825012031607e-05,
+      "loss": 1.6254,
+      "step": 3147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843129456588807e-05,
+      "loss": 1.3605,
+      "step": 3148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5840433256326056e-05,
+      "loss": 1.51,
+      "step": 3149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5837736411540882e-05,
+      "loss": 1.4452,
+      "step": 3150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5835038922530902e-05,
+      "loss": 1.4158,
+      "step": 3151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5832340789593793e-05,
+      "loss": 2.8708,
+      "step": 3152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829642013027312e-05,
+      "loss": 1.5699,
+      "step": 3153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826942593129282e-05,
+      "loss": 1.3608,
+      "step": 3154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58242425301976e-05,
+      "loss": 1.5783,
+      "step": 3155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821541824530226e-05,
+      "loss": 1.875,
+      "step": 3156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818840476425202e-05,
+      "loss": 1.5651,
+      "step": 3157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816138486180642e-05,
+      "loss": 1.3131,
+      "step": 3158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581343585409471e-05,
+      "loss": 1.4872,
+      "step": 3159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581073258046567e-05,
+      "loss": 1.4127,
+      "step": 3160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580802866559183e-05,
+      "loss": 1.2612,
+      "step": 3161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580532410977159e-05,
+      "loss": 1.5496,
+      "step": 3162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580261891330341e-05,
+      "loss": 1.8928,
+      "step": 3163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579991307648582e-05,
+      "loss": 1.769,
+      "step": 3164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797206599617424e-05,
+      "loss": 1.3944,
+      "step": 3165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794499482996902e-05,
+      "loss": 2.094,
+      "step": 3166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791791726922986e-05,
+      "loss": 1.2217,
+      "step": 3167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789083331694505e-05,
+      "loss": 1.2297,
+      "step": 3168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5786374297610336e-05,
+      "loss": 1.7577,
+      "step": 3169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5783664624969434e-05,
+      "loss": 1.5745,
+      "step": 3170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5780954314070833e-05,
+      "loss": 1.4956,
+      "step": 3171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5778243365213626e-05,
+      "loss": 1.5475,
+      "step": 3172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577553177869698e-05,
+      "loss": 1.4241,
+      "step": 3173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577281955482013e-05,
+      "loss": 1.3209,
+      "step": 3174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5770106693882387e-05,
+      "loss": 1.5739,
+      "step": 3175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767393196183132e-05,
+      "loss": 1.5108,
+      "step": 3176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5764679062021807e-05,
+      "loss": 1.9966,
+      "step": 3177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576196429169794e-05,
+      "loss": 1.2259,
+      "step": 3178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575924888551111e-05,
+      "loss": 1.4633,
+      "step": 3179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575653284376099e-05,
+      "loss": 1.5229,
+      "step": 3180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753816166747294e-05,
+      "loss": 1.8251,
+      "step": 3181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575109885476983e-05,
+      "loss": 1.5225,
+      "step": 3182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574838090812847e-05,
+      "loss": 1.3861,
+      "step": 3183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745662327123145e-05,
+      "loss": 1.3763,
+      "step": 3184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574294311205387e-05,
+      "loss": 1.2859,
+      "step": 3185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5740223263220728e-05,
+      "loss": 1.3433,
+      "step": 3186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573750278092386e-05,
+      "loss": 1.4764,
+      "step": 3187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5734781665463495e-05,
+      "loss": 1.7185,
+      "step": 3188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732059917139912e-05,
+      "loss": 1.543,
+      "step": 3189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572933753625348e-05,
+      "loss": 1.2244,
+      "step": 3190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726614523104622e-05,
+      "loss": 1.4451,
+      "step": 3191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723890877993837e-05,
+      "loss": 1.4518,
+      "step": 3192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721166601221697e-05,
+      "loss": 1.5602,
+      "step": 3193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571844169308884e-05,
+      "loss": 1.467,
+      "step": 3194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571571615389597e-05,
+      "loss": 1.5183,
+      "step": 3195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712989983943867e-05,
+      "loss": 1.245,
+      "step": 3196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5710263183533378e-05,
+      "loss": 1.673,
+      "step": 3197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570753575296542e-05,
+      "loss": 1.3897,
+      "step": 3198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704807692540977e-05,
+      "loss": 1.2617,
+      "step": 3199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702079002561105e-05,
+      "loss": 1.733,
+      "step": 3200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569934968332693e-05,
+      "loss": 1.3641,
+      "step": 3201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5696619735139654e-05,
+      "loss": 1.4743,
+      "step": 3202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5693889158300533e-05,
+      "loss": 1.6649,
+      "step": 3203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56911579531109e-05,
+      "loss": 1.6041,
+      "step": 3204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568842611987216e-05,
+      "loss": 1.3852,
+      "step": 3205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5685693658885783e-05,
+      "loss": 1.4872,
+      "step": 3206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5682960570453315e-05,
+      "loss": 1.7049,
+      "step": 3207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5680226854876364e-05,
+      "loss": 1.826,
+      "step": 3208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567749251245661e-05,
+      "loss": 1.4053,
+      "step": 3209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56747575434958e-05,
+      "loss": 1.5793,
+      "step": 3210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5672021948295756e-05,
+      "loss": 1.5107,
+      "step": 3211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669285727158363e-05,
+      "loss": 1.1895,
+      "step": 3212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666548880385575e-05,
+      "loss": 1.68,
+      "step": 3213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663811408279422e-05,
+      "loss": 1.2609,
+      "step": 3214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5661073311141996e-05,
+      "loss": 1.8583,
+      "step": 3215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565833458927546e-05,
+      "loss": 1.818,
+      "step": 3216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565559524298205e-05,
+      "loss": 1.4045,
+      "step": 3217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565285527256406e-05,
+      "loss": 1.8964,
+      "step": 3218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565011467832387e-05,
+      "loss": 2.1255,
+      "step": 3219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647373460563907e-05,
+      "loss": 1.5784,
+      "step": 3220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564463161958669e-05,
+      "loss": 1.2948,
+      "step": 3221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641889155694786e-05,
+      "loss": 1.5195,
+      "step": 3222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563914606919085e-05,
+      "loss": 1.7255,
+      "step": 3223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636402360377588e-05,
+      "loss": 1.6401,
+      "step": 3224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633658029557785e-05,
+      "loss": 1.5099,
+      "step": 3225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563091307703429e-05,
+      "loss": 1.5594,
+      "step": 3226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562816750311003e-05,
+      "loss": 1.2332,
+      "step": 3227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562542130808799e-05,
+      "loss": 1.6034,
+      "step": 3228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622674492271223e-05,
+      "loss": 1.5394,
+      "step": 3229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619927055962857e-05,
+      "loss": 1.8351,
+      "step": 3230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561717899946609e-05,
+      "loss": 1.3657,
+      "step": 3231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561443032308418e-05,
+      "loss": 1.5338,
+      "step": 3232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561168102712046e-05,
+      "loss": 1.8489,
+      "step": 3233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5608931111878325e-05,
+      "loss": 1.5514,
+      "step": 3234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.560618057766125e-05,
+      "loss": 1.5188,
+      "step": 3235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.560342942477276e-05,
+      "loss": 1.3885,
+      "step": 3236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5600677653516468e-05,
+      "loss": 1.2749,
+      "step": 3237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5597925264196048e-05,
+      "loss": 1.5986,
+      "step": 3238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5595172257115234e-05,
+      "loss": 1.3257,
+      "step": 3239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5592418632577838e-05,
+      "loss": 1.3222,
+      "step": 3240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5589664390887737e-05,
+      "loss": 1.7912,
+      "step": 3241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558690953234887e-05,
+      "loss": 1.5329,
+      "step": 3242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5584154057265265e-05,
+      "loss": 1.4071,
+      "step": 3243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581397965940986e-05,
+      "loss": 1.5763,
+      "step": 3244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578641258680192e-05,
+      "loss": 1.7962,
+      "step": 3245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575883935787097e-05,
+      "loss": 1.8057,
+      "step": 3246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557312599756599e-05,
+      "loss": 1.5978,
+      "step": 3247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570367444321218e-05,
+      "loss": 1.7621,
+      "step": 3248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567608276357203e-05,
+      "loss": 1.3491,
+      "step": 3249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556484849397844e-05,
+      "loss": 1.5221,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562088097489472e-05,
+      "loss": 0.998,
+      "step": 3251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559327087194943e-05,
+      "loss": 0.9476,
+      "step": 3252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556565463399525e-05,
+      "loss": 1.0043,
+      "step": 3253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553803226407992e-05,
+      "loss": 1.1111,
+      "step": 3254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551040376525163e-05,
+      "loss": 0.9168,
+      "step": 3255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5548276914055933e-05,
+      "loss": 1.2027,
+      "step": 3256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554551283930527e-05,
+      "loss": 0.8383,
+      "step": 3257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5542748152578204e-05,
+      "loss": 1.1465,
+      "step": 3258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553998285417983e-05,
+      "loss": 0.9568,
+      "step": 3259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553721694441531e-05,
+      "loss": 0.7924,
+      "step": 3260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534450423589883e-05,
+      "loss": 1.1345,
+      "step": 3261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5531683292008845e-05,
+      "loss": 0.7804,
+      "step": 3262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552891554997756e-05,
+      "loss": 0.8137,
+      "step": 3263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5526147197801474e-05,
+      "loss": 1.055,
+      "step": 3264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552337823578608e-05,
+      "loss": 0.9127,
+      "step": 3265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552060866423695e-05,
+      "loss": 1.0734,
+      "step": 3266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5517838483459718e-05,
+      "loss": 0.9999,
+      "step": 3267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5515067693760092e-05,
+      "loss": 1.0282,
+      "step": 3268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5512296295443845e-05,
+      "loss": 0.8155,
+      "step": 3269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5509524288816812e-05,
+      "loss": 1.1039,
+      "step": 3270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5506751674184895e-05,
+      "loss": 1.0874,
+      "step": 3271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503978451854072e-05,
+      "loss": 1.1715,
+      "step": 3272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550120462213038e-05,
+      "loss": 1.0271,
+      "step": 3273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498430185319925e-05,
+      "loss": 1.1039,
+      "step": 3274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549565514172889e-05,
+      "loss": 1.189,
+      "step": 3275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492879491663498e-05,
+      "loss": 1.1609,
+      "step": 3276
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549010323543007e-05,
+      "loss": 0.9113,
+      "step": 3277
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487326373334973e-05,
+      "loss": 1.1983,
+      "step": 3278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484548905684658e-05,
+      "loss": 1.0111,
+      "step": 3279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481770832785622e-05,
+      "loss": 1.0219,
+      "step": 3280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478992154944447e-05,
+      "loss": 1.0622,
+      "step": 3281
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5476212872467777e-05,
+      "loss": 0.9908,
+      "step": 3282
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547343298566231e-05,
+      "loss": 0.9053,
+      "step": 3283
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470652494834828e-05,
+      "loss": 0.9045,
+      "step": 3284
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467871400292174e-05,
+      "loss": 0.9491,
+      "step": 3285
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5465089702341253e-05,
+      "loss": 1.1655,
+      "step": 3286
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546230740128904e-05,
+      "loss": 1.0352,
+      "step": 3287
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545952449744258e-05,
+      "loss": 0.8286,
+      "step": 3288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545674099110898e-05,
+      "loss": 0.9579,
+      "step": 3289
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453956882595406e-05,
+      "loss": 1.172,
+      "step": 3290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451172172209115e-05,
+      "loss": 1.0927,
+      "step": 3291
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54483868602574e-05,
+      "loss": 0.9464,
+      "step": 3292
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544560094704764e-05,
+      "loss": 1.2146,
+      "step": 3293
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544281443288728e-05,
+      "loss": 1.1827,
+      "step": 3294
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440027318083815e-05,
+      "loss": 0.8445,
+      "step": 3295
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543723960294483e-05,
+      "loss": 0.8347,
+      "step": 3296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543445128777796e-05,
+      "loss": 0.8046,
+      "step": 3297
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54316623728909e-05,
+      "loss": 0.7245,
+      "step": 3298
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5428872858591438e-05,
+      "loss": 0.9162,
+      "step": 3299
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5426082745187402e-05,
+      "loss": 0.93,
+      "step": 3300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5423292032986695e-05,
+      "loss": 0.6418,
+      "step": 3301
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5420500722297292e-05,
+      "loss": 0.6595,
+      "step": 3302
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5417708813427222e-05,
+      "loss": 0.9148,
+      "step": 3303
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5414916306684593e-05,
+      "loss": 0.8985,
+      "step": 3304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541212320237757e-05,
+      "loss": 0.9927,
+      "step": 3305
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5409329500814387e-05,
+      "loss": 0.8463,
+      "step": 3306
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5406535202303345e-05,
+      "loss": 1.014,
+      "step": 3307
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5403740307152804e-05,
+      "loss": 0.6688,
+      "step": 3308
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400944815671204e-05,
+      "loss": 0.9197,
+      "step": 3309
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5398148728167037e-05,
+      "loss": 1.0024,
+      "step": 3310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395352044948864e-05,
+      "loss": 0.9539,
+      "step": 3311
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392554766325317e-05,
+      "loss": 0.9931,
+      "step": 3312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538975689260509e-05,
+      "loss": 1.009,
+      "step": 3313
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538695842409694e-05,
+      "loss": 1.3161,
+      "step": 3314
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5384159361109697e-05,
+      "loss": 0.9769,
+      "step": 3315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5381359703952253e-05,
+      "loss": 0.9155,
+      "step": 3316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537855945293356e-05,
+      "loss": 1.029,
+      "step": 3317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375758608362642e-05,
+      "loss": 1.1964,
+      "step": 3318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537295717054859e-05,
+      "loss": 0.6048,
+      "step": 3319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537015513980055e-05,
+      "loss": 0.9917,
+      "step": 3320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5367352516427748e-05,
+      "loss": 0.9525,
+      "step": 3321
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364549300739467e-05,
+      "loss": 1.2795,
+      "step": 3322
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361745493045053e-05,
+      "loss": 0.6208,
+      "step": 3323
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358941093653928e-05,
+      "loss": 1.0181,
+      "step": 3324
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356136102875562e-05,
+      "loss": 0.985,
+      "step": 3325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535333052101951e-05,
+      "loss": 1.2098,
+      "step": 3326
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535052434839538e-05,
+      "loss": 1.052,
+      "step": 3327
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347717585312837e-05,
+      "loss": 1.0336,
+      "step": 3328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534491023208164e-05,
+      "loss": 1.1479,
+      "step": 3329
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5342102289011584e-05,
+      "loss": 0.7062,
+      "step": 3330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533929375641254e-05,
+      "loss": 0.9469,
+      "step": 3331
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5336484634594452e-05,
+      "loss": 0.8737,
+      "step": 3332
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5333674923867313e-05,
+      "loss": 1.0543,
+      "step": 3333
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5330864624541194e-05,
+      "loss": 1.125,
+      "step": 3334
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5328053736926224e-05,
+      "loss": 0.8119,
+      "step": 3335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53252422613326e-05,
+      "loss": 0.9795,
+      "step": 3336
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5322430198070584e-05,
+      "loss": 1.0353,
+      "step": 3337
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53196175474505e-05,
+      "loss": 1.0304,
+      "step": 3338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5316804309782744e-05,
+      "loss": 0.7989,
+      "step": 3339
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313990485377762e-05,
+      "loss": 0.9866,
+      "step": 3340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531117607454608e-05,
+      "loss": 1.0581,
+      "step": 3341
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5308361077598285e-05,
+      "loss": 0.8907,
+      "step": 3342
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5305545494845024e-05,
+      "loss": 1.2773,
+      "step": 3343
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302729326597013e-05,
+      "loss": 0.9146,
+      "step": 3344
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299912573165022e-05,
+      "loss": 1.1619,
+      "step": 3345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529709523485991e-05,
+      "loss": 1.2831,
+      "step": 3346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529427731199258e-05,
+      "loss": 1.0326,
+      "step": 3347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291458804873992e-05,
+      "loss": 1.1022,
+      "step": 3348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288639713815198e-05,
+      "loss": 0.8965,
+      "step": 3349
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528582003912729e-05,
+      "loss": 0.8525,
+      "step": 3350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528299978112144e-05,
+      "loss": 1.0626,
+      "step": 3351
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5280178940108878e-05,
+      "loss": 1.1454,
+      "step": 3352
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277357516400897e-05,
+      "loss": 0.9667,
+      "step": 3353
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5274535510308853e-05,
+      "loss": 1.1191,
+      "step": 3354
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5271712922144174e-05,
+      "loss": 0.9503,
+      "step": 3355
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268889752218344e-05,
+      "loss": 0.8213,
+      "step": 3356
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526606600084292e-05,
+      "loss": 0.9526,
+      "step": 3357
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526324166832951e-05,
+      "loss": 1.0822,
+      "step": 3358
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5260416754989794e-05,
+      "loss": 0.7268,
+      "step": 3359
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257591261135528e-05,
+      "loss": 0.9278,
+      "step": 3360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254765187078506e-05,
+      "loss": 0.8483,
+      "step": 3361
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251938533130605e-05,
+      "loss": 0.9327,
+      "step": 3362
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5249111299603762e-05,
+      "loss": 0.8564,
+      "step": 3363
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5246283486809979e-05,
+      "loss": 1.347,
+      "step": 3364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.524345509506132e-05,
+      "loss": 0.9401,
+      "step": 3365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5240626124669905e-05,
+      "loss": 0.7253,
+      "step": 3366
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5237796575947934e-05,
+      "loss": 1.1823,
+      "step": 3367
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.523496644920766e-05,
+      "loss": 1.1267,
+      "step": 3368
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232135744761403e-05,
+      "loss": 0.9249,
+      "step": 3369
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5229304462921543e-05,
+      "loss": 1.2175,
+      "step": 3370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522647260400053e-05,
+      "loss": 0.8065,
+      "step": 3371
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223640168310873e-05,
+      "loss": 0.8325,
+      "step": 3372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220807156165144e-05,
+      "loss": 1.2722,
+      "step": 3373
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521797356787599e-05,
+      "loss": 0.8785,
+      "step": 3374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52151394037561e-05,
+      "loss": 1.1324,
+      "step": 3375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521230466411825e-05,
+      "loss": 0.7625,
+      "step": 3376
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209469349275261e-05,
+      "loss": 1.2244,
+      "step": 3377
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5206633459540024e-05,
+      "loss": 0.8167,
+      "step": 3378
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203796995225497e-05,
+      "loss": 0.9843,
+      "step": 3379
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200959956644701e-05,
+      "loss": 0.9915,
+      "step": 3380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198122344110716e-05,
+      "loss": 1.0626,
+      "step": 3381
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195284157936685e-05,
+      "loss": 0.9863,
+      "step": 3382
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519244539843582e-05,
+      "loss": 1.4046,
+      "step": 3383
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189606065921392e-05,
+      "loss": 1.0181,
+      "step": 3384
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186766160706738e-05,
+      "loss": 0.9071,
+      "step": 3385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183925683105254e-05,
+      "loss": 0.8917,
+      "step": 3386
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181084633430398e-05,
+      "loss": 0.9763,
+      "step": 3387
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178243011995703e-05,
+      "loss": 0.9,
+      "step": 3388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517540081911475e-05,
+      "loss": 0.9924,
+      "step": 3389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5172558055101197e-05,
+      "loss": 0.9256,
+      "step": 3390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169714720268754e-05,
+      "loss": 1.0794,
+      "step": 3391
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166870814931191e-05,
+      "loss": 0.8982,
+      "step": 3392
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164026339402358e-05,
+      "loss": 1.0915,
+      "step": 3393
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161181293996152e-05,
+      "loss": 0.8688,
+      "step": 3394
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158335679026543e-05,
+      "loss": 0.9525,
+      "step": 3395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5155489494807558e-05,
+      "loss": 0.7319,
+      "step": 3396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5152642741653285e-05,
+      "loss": 0.9049,
+      "step": 3397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5149795419877884e-05,
+      "loss": 1.1258,
+      "step": 3398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5146947529795565e-05,
+      "loss": 0.8903,
+      "step": 3399
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5144099071720613e-05,
+      "loss": 1.1894,
+      "step": 3400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5141250045967369e-05,
+      "loss": 1.239,
+      "step": 3401
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513840045285024e-05,
+      "loss": 1.175,
+      "step": 3402
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135550292683684e-05,
+      "loss": 0.8917,
+      "step": 3403
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513269956578224e-05,
+      "loss": 1.1397,
+      "step": 3404
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129848272460504e-05,
+      "loss": 0.9798,
+      "step": 3405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126996413033121e-05,
+      "loss": 1.9488,
+      "step": 3406
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124143987814815e-05,
+      "loss": 0.7778,
+      "step": 3407
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121290997120368e-05,
+      "loss": 1.1122,
+      "step": 3408
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118437441264613e-05,
+      "loss": 0.8265,
+      "step": 3409
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115583320562463e-05,
+      "loss": 1.0522,
+      "step": 3410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112728635328883e-05,
+      "loss": 1.0295,
+      "step": 3411
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109873385878904e-05,
+      "loss": 1.0859,
+      "step": 3412
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107017572527615e-05,
+      "loss": 1.034,
+      "step": 3413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5104161195590172e-05,
+      "loss": 0.7714,
+      "step": 3414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101304255381792e-05,
+      "loss": 0.8641,
+      "step": 3415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5098446752217748e-05,
+      "loss": 0.844,
+      "step": 3416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095588686413387e-05,
+      "loss": 1.0108,
+      "step": 3417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509273005828411e-05,
+      "loss": 0.9393,
+      "step": 3418
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089870868145378e-05,
+      "loss": 1.2103,
+      "step": 3419
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087011116312718e-05,
+      "loss": 1.305,
+      "step": 3420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084150803101723e-05,
+      "loss": 0.8721,
+      "step": 3421
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5081289928828043e-05,
+      "loss": 1.2135,
+      "step": 3422
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5078428493807384e-05,
+      "loss": 0.8368,
+      "step": 3423
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5075566498355528e-05,
+      "loss": 1.2047,
+      "step": 3424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072703942788306e-05,
+      "loss": 0.8886,
+      "step": 3425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069840827421623e-05,
+      "loss": 0.9544,
+      "step": 3426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066977152571429e-05,
+      "loss": 0.9536,
+      "step": 3427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064112918553753e-05,
+      "loss": 0.7182,
+      "step": 3428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5061248125684675e-05,
+      "loss": 1.1413,
+      "step": 3429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5058382774280341e-05,
+      "loss": 1.0271,
+      "step": 3430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5055516864656954e-05,
+      "loss": 1.1113,
+      "step": 3431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5052650397130788e-05,
+      "loss": 1.0263,
+      "step": 3432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5049783372018171e-05,
+      "loss": 0.8803,
+      "step": 3433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504691578963549e-05,
+      "loss": 1.1225,
+      "step": 3434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044047650299205e-05,
+      "loss": 1.296,
+      "step": 3435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041178954325825e-05,
+      "loss": 0.9007,
+      "step": 3436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038309702031924e-05,
+      "loss": 0.8745,
+      "step": 3437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035439893734144e-05,
+      "loss": 0.8424,
+      "step": 3438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032569529749182e-05,
+      "loss": 1.2336,
+      "step": 3439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029698610393798e-05,
+      "loss": 0.9984,
+      "step": 3440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502682713598481e-05,
+      "loss": 0.9473,
+      "step": 3441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023955106839104e-05,
+      "loss": 0.7814,
+      "step": 3442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021082523273617e-05,
+      "loss": 0.8953,
+      "step": 3443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5018209385605362e-05,
+      "loss": 0.9953,
+      "step": 3444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015335694151399e-05,
+      "loss": 1.0679,
+      "step": 3445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501246144922886e-05,
+      "loss": 1.0884,
+      "step": 3446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009586651154931e-05,
+      "loss": 0.921,
+      "step": 3447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006711300246854e-05,
+      "loss": 1.064,
+      "step": 3448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5003835396821953e-05,
+      "loss": 0.8435,
+      "step": 3449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000958941197582e-05,
+      "loss": 0.7727,
+      "step": 3450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998081933691188e-05,
+      "loss": 0.7266,
+      "step": 3451
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995204374620258e-05,
+      "loss": 1.0633,
+      "step": 3452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992326264302347e-05,
+      "loss": 1.0454,
+      "step": 3453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989447603055066e-05,
+      "loss": 0.6552,
+      "step": 3454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986568391196092e-05,
+      "loss": 0.6858,
+      "step": 3455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983688629043163e-05,
+      "loss": 1.0022,
+      "step": 3456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980808316914073e-05,
+      "loss": 1.1196,
+      "step": 3457
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977927455126683e-05,
+      "loss": 0.871,
+      "step": 3458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975046043998907e-05,
+      "loss": 0.8341,
+      "step": 3459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4972164083848729e-05,
+      "loss": 0.9038,
+      "step": 3460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969281574994181e-05,
+      "loss": 1.1499,
+      "step": 3461
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.496639851775337e-05,
+      "loss": 1.0832,
+      "step": 3462
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4963514912444457e-05,
+      "loss": 1.0345,
+      "step": 3463
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4960630759385655e-05,
+      "loss": 1.1416,
+      "step": 3464
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4957746058895253e-05,
+      "loss": 0.7964,
+      "step": 3465
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4954860811291591e-05,
+      "loss": 0.9127,
+      "step": 3466
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951975016893068e-05,
+      "loss": 1.0341,
+      "step": 3467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949088676018148e-05,
+      "loss": 1.3621,
+      "step": 3468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946201788985358e-05,
+      "loss": 1.1,
+      "step": 3469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4943314356113274e-05,
+      "loss": 0.8936,
+      "step": 3470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940426377720548e-05,
+      "loss": 0.9851,
+      "step": 3471
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937537854125874e-05,
+      "loss": 1.1511,
+      "step": 3472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934648785648023e-05,
+      "loss": 0.988,
+      "step": 3473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931759172605815e-05,
+      "loss": 1.0126,
+      "step": 3474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928869015318134e-05,
+      "loss": 1.1899,
+      "step": 3475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492597831410393e-05,
+      "loss": 0.9379,
+      "step": 3476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49230870692822e-05,
+      "loss": 1.0787,
+      "step": 3477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920195281172007e-05,
+      "loss": 0.9255,
+      "step": 3478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4917302950092482e-05,
+      "loss": 1.0459,
+      "step": 3479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914410076362807e-05,
+      "loss": 1.1252,
+      "step": 3480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4911516660302218e-05,
+      "loss": 0.9743,
+      "step": 3481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908622702230029e-05,
+      "loss": 0.9627,
+      "step": 3482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905728202465596e-05,
+      "loss": 1.1866,
+      "step": 3483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902833161328347e-05,
+      "loss": 1.1405,
+      "step": 3484
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899937579137763e-05,
+      "loss": 0.7024,
+      "step": 3485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897041456213387e-05,
+      "loss": 1.0145,
+      "step": 3486
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894144792874823e-05,
+      "loss": 1.0219,
+      "step": 3487
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891247589441728e-05,
+      "loss": 0.8508,
+      "step": 3488
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888349846233828e-05,
+      "loss": 0.8488,
+      "step": 3489
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885451563570903e-05,
+      "loss": 1.0092,
+      "step": 3490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882552741772794e-05,
+      "loss": 1.1529,
+      "step": 3491
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879653381159397e-05,
+      "loss": 0.8634,
+      "step": 3492
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876753482050678e-05,
+      "loss": 0.8149,
+      "step": 3493
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4873853044766657e-05,
+      "loss": 1.2586,
+      "step": 3494
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4870952069627405e-05,
+      "loss": 0.9724,
+      "step": 3495
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4868050556953064e-05,
+      "loss": 0.7354,
+      "step": 3496
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4865148507063833e-05,
+      "loss": 0.9917,
+      "step": 3497
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4862245920279963e-05,
+      "loss": 1.1142,
+      "step": 3498
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4859342796921773e-05,
+      "loss": 0.9095,
+      "step": 3499
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4856439137309642e-05,
+      "loss": 1.1114,
+      "step": 3500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853534941763996e-05,
+      "loss": 0.7406,
+      "step": 3501
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850630210605335e-05,
+      "loss": 1.0491,
+      "step": 3502
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847724944154205e-05,
+      "loss": 1.302,
+      "step": 3503
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844819142731224e-05,
+      "loss": 1.2236,
+      "step": 3504
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841912806657057e-05,
+      "loss": 0.7287,
+      "step": 3505
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839005936252435e-05,
+      "loss": 0.8693,
+      "step": 3506
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483609853183815e-05,
+      "loss": 0.718,
+      "step": 3507
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833190593735046e-05,
+      "loss": 1.0142,
+      "step": 3508
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830282122264024e-05,
+      "loss": 0.9536,
+      "step": 3509
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482737311774606e-05,
+      "loss": 0.8502,
+      "step": 3510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482446358050217e-05,
+      "loss": 1.2336,
+      "step": 3511
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821553510853441e-05,
+      "loss": 0.8399,
+      "step": 3512
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481864290912101e-05,
+      "loss": 1.1922,
+      "step": 3513
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4815731775626084e-05,
+      "loss": 0.7154,
+      "step": 3514
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4812820110689914e-05,
+      "loss": 1.0432,
+      "step": 3515
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809907914633823e-05,
+      "loss": 0.991,
+      "step": 3516
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806995187779185e-05,
+      "loss": 1.0473,
+      "step": 3517
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4804081930447433e-05,
+      "loss": 0.9818,
+      "step": 3518
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801168142960063e-05,
+      "loss": 0.8165,
+      "step": 3519
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798253825638625e-05,
+      "loss": 0.8779,
+      "step": 3520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795338978804732e-05,
+      "loss": 1.0692,
+      "step": 3521
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792423602780049e-05,
+      "loss": 2.6003,
+      "step": 3522
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789507697886306e-05,
+      "loss": 1.0515,
+      "step": 3523
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786591264445284e-05,
+      "loss": 0.9064,
+      "step": 3524
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783674302778831e-05,
+      "loss": 1.3585,
+      "step": 3525
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478075681320885e-05,
+      "loss": 0.8757,
+      "step": 3526
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4777838796057298e-05,
+      "loss": 0.9289,
+      "step": 3527
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4774920251646192e-05,
+      "loss": 0.9485,
+      "step": 3528
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4772001180297615e-05,
+      "loss": 1.9639,
+      "step": 3529
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4769081582333695e-05,
+      "loss": 0.9072,
+      "step": 3530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4766161458076628e-05,
+      "loss": 1.2003,
+      "step": 3531
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4763240807848667e-05,
+      "loss": 0.988,
+      "step": 3532
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760319631972117e-05,
+      "loss": 0.8915,
+      "step": 3533
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4757397930769348e-05,
+      "loss": 0.9364,
+      "step": 3534
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754475704562782e-05,
+      "loss": 0.8203,
+      "step": 3535
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4751552953674905e-05,
+      "loss": 1.2026,
+      "step": 3536
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748629678428258e-05,
+      "loss": 0.8296,
+      "step": 3537
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4745705879145442e-05,
+      "loss": 1.3071,
+      "step": 3538
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742781556149104e-05,
+      "loss": 0.9714,
+      "step": 3539
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739856709761968e-05,
+      "loss": 1.1715,
+      "step": 3540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736931340306802e-05,
+      "loss": 1.025,
+      "step": 3541
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4734005448106437e-05,
+      "loss": 1.0393,
+      "step": 3542
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731079033483759e-05,
+      "loss": 0.8813,
+      "step": 3543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728152096761716e-05,
+      "loss": 0.971,
+      "step": 3544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725224638263309e-05,
+      "loss": 1.0437,
+      "step": 3545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722296658311596e-05,
+      "loss": 1.1866,
+      "step": 3546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719368157229699e-05,
+      "loss": 0.9924,
+      "step": 3547
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716439135340795e-05,
+      "loss": 1.0506,
+      "step": 3548
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713509592968109e-05,
+      "loss": 1.0429,
+      "step": 3549
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710579530434936e-05,
+      "loss": 0.8914,
+      "step": 3550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707648948064627e-05,
+      "loss": 1.4063,
+      "step": 3551
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704717846180585e-05,
+      "loss": 1.3888,
+      "step": 3552
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701786225106269e-05,
+      "loss": 0.7413,
+      "step": 3553
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698854085165202e-05,
+      "loss": 1.0333,
+      "step": 3554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695921426680962e-05,
+      "loss": 0.9999,
+      "step": 3555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692988249977179e-05,
+      "loss": 1.0868,
+      "step": 3556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690054555377546e-05,
+      "loss": 1.0393,
+      "step": 3557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687120343205814e-05,
+      "loss": 1.2948,
+      "step": 3558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4684185613785786e-05,
+      "loss": 0.8449,
+      "step": 3559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4681250367441328e-05,
+      "loss": 0.6876,
+      "step": 3560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4678314604496355e-05,
+      "loss": 1.1864,
+      "step": 3561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4675378325274851e-05,
+      "loss": 1.0122,
+      "step": 3562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4672441530100843e-05,
+      "loss": 1.0006,
+      "step": 3563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4669504219298422e-05,
+      "loss": 1.0915,
+      "step": 3564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666566393191744e-05,
+      "loss": 0.9814,
+      "step": 3565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4663628052105003e-05,
+      "loss": 0.9652,
+      "step": 3566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660689196362463e-05,
+      "loss": 1.2702,
+      "step": 3567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657749826288446e-05,
+      "loss": 0.8348,
+      "step": 3568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654809942207325e-05,
+      "loss": 0.9666,
+      "step": 3569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651869544443528e-05,
+      "loss": 0.6251,
+      "step": 3570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464892863332155e-05,
+      "loss": 0.9603,
+      "step": 3571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645987209165933e-05,
+      "loss": 1.0408,
+      "step": 3572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4643045272301275e-05,
+      "loss": 1.0573,
+      "step": 3573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464010282305224e-05,
+      "loss": 1.2126,
+      "step": 3574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463715986174354e-05,
+      "loss": 1.0472,
+      "step": 3575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4634216388699945e-05,
+      "loss": 0.9166,
+      "step": 3576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631272404246284e-05,
+      "loss": 0.9447,
+      "step": 3577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4628327908707438e-05,
+      "loss": 1.1082,
+      "step": 3578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625382902408356e-05,
+      "loss": 1.0834,
+      "step": 3579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622437385674027e-05,
+      "loss": 0.9801,
+      "step": 3580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619491358829502e-05,
+      "loss": 0.8313,
+      "step": 3581
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616544822199901e-05,
+      "loss": 0.8767,
+      "step": 3582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613597776110382e-05,
+      "loss": 1.1766,
+      "step": 3583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461065022088617e-05,
+      "loss": 0.8416,
+      "step": 3584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460770215685254e-05,
+      "loss": 0.5835,
+      "step": 3585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4604753584334828e-05,
+      "loss": 0.908,
+      "step": 3586
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601804503658427e-05,
+      "loss": 0.7097,
+      "step": 3587
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459885491514878e-05,
+      "loss": 1.0217,
+      "step": 3588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459590481913139e-05,
+      "loss": 1.2156,
+      "step": 3589
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592954215931819e-05,
+      "loss": 1.1839,
+      "step": 3590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590003105875677e-05,
+      "loss": 0.9135,
+      "step": 3591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4587051489288633e-05,
+      "loss": 1.4125,
+      "step": 3592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4584099366496422e-05,
+      "loss": 1.1393,
+      "step": 3593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.458114673782482e-05,
+      "loss": 1.1894,
+      "step": 3594
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4578193603599662e-05,
+      "loss": 1.0267,
+      "step": 3595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.457523996414685e-05,
+      "loss": 0.9377,
+      "step": 3596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572285819792327e-05,
+      "loss": 1.1749,
+      "step": 3597
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4569331170862101e-05,
+      "loss": 1.1414,
+      "step": 3598
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566376017682234e-05,
+      "loss": 0.7771,
+      "step": 3599
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563420360578842e-05,
+      "loss": 0.9081,
+      "step": 3600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4560464199878092e-05,
+      "loss": 0.8147,
+      "step": 3601
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557507535906221e-05,
+      "loss": 0.6601,
+      "step": 3602
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554550368989506e-05,
+      "loss": 1.04,
+      "step": 3603
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455159269945429e-05,
+      "loss": 1.0224,
+      "step": 3604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548634527626961e-05,
+      "loss": 0.7492,
+      "step": 3605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4545675853833977e-05,
+      "loss": 0.8988,
+      "step": 3606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542716678401836e-05,
+      "loss": 1.0097,
+      "step": 3607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45397570016571e-05,
+      "loss": 1.0644,
+      "step": 3608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536796823926387e-05,
+      "loss": 1.0634,
+      "step": 3609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453383614553637e-05,
+      "loss": 1.1365,
+      "step": 3610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4530874966813771e-05,
+      "loss": 0.936,
+      "step": 3611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527913288085373e-05,
+      "loss": 1.0932,
+      "step": 3612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524951109678015e-05,
+      "loss": 0.9566,
+      "step": 3613
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521988431918586e-05,
+      "loss": 1.1192,
+      "step": 3614
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519025255134035e-05,
+      "loss": 0.7706,
+      "step": 3615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516061579651361e-05,
+      "loss": 1.0599,
+      "step": 3616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513097405797622e-05,
+      "loss": 0.9976,
+      "step": 3617
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510132733899935e-05,
+      "loss": 0.7654,
+      "step": 3618
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507167564285462e-05,
+      "loss": 1.0672,
+      "step": 3619
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504201897281427e-05,
+      "loss": 1.2496,
+      "step": 3620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501235733215106e-05,
+      "loss": 0.8838,
+      "step": 3621
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449826907241383e-05,
+      "loss": 1.2571,
+      "step": 3622
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4495301915204989e-05,
+      "loss": 1.1912,
+      "step": 3623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4492334261916021e-05,
+      "loss": 0.9383,
+      "step": 3624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.448936611287442e-05,
+      "loss": 1.145,
+      "step": 3625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4486397468407742e-05,
+      "loss": 1.1706,
+      "step": 3626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4483428328843594e-05,
+      "loss": 0.989,
+      "step": 3627
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.448045869450963e-05,
+      "loss": 1.1273,
+      "step": 3628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477488565733567e-05,
+      "loss": 1.0525,
+      "step": 3629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474517942843173e-05,
+      "loss": 0.9212,
+      "step": 3630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471546826166275e-05,
+      "loss": 0.9458,
+      "step": 3631
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446857521603075e-05,
+      "loss": 1.2971,
+      "step": 3632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465603112764528e-05,
+      "loss": 1.3468,
+      "step": 3633
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462630516695602e-05,
+      "loss": 0.9741,
+      "step": 3634
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4459657428152008e-05,
+      "loss": 1.1992,
+      "step": 3635
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456683847461841e-05,
+      "loss": 1.0594,
+      "step": 3636
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453709774953259e-05,
+      "loss": 0.9696,
+      "step": 3637
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450735210954461e-05,
+      "loss": 1.1503,
+      "step": 3638
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4447760155793704e-05,
+      "loss": 0.8514,
+      "step": 3639
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4444784609799306e-05,
+      "loss": 0.7815,
+      "step": 3640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441808573299632e-05,
+      "loss": 0.9822,
+      "step": 3641
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44388320466231e-05,
+      "loss": 0.689,
+      "step": 3642
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435855030098186e-05,
+      "loss": 0.8073,
+      "step": 3643
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432877524053429e-05,
+      "loss": 0.9477,
+      "step": 3644
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429899528817397e-05,
+      "loss": 0.8799,
+      "step": 3645
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426921044718738e-05,
+      "loss": 0.9362,
+      "step": 3646
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423942072086142e-05,
+      "loss": 0.8251,
+      "step": 3647
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442096261124835e-05,
+      "loss": 0.8619,
+      "step": 3648
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4417982662534164e-05,
+      "loss": 1.3131,
+      "step": 3649
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415002226272437e-05,
+      "loss": 0.9862,
+      "step": 3650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412021302792079e-05,
+      "loss": 0.8806,
+      "step": 3651
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4409039892422043e-05,
+      "loss": 0.9074,
+      "step": 3652
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4406057995491348e-05,
+      "loss": 0.9782,
+      "step": 3653
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403075612329065e-05,
+      "loss": 0.9752,
+      "step": 3654
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400092743264309e-05,
+      "loss": 1.479,
+      "step": 3655
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4397109388626256e-05,
+      "loss": 0.8795,
+      "step": 3656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.439412554874414e-05,
+      "loss": 1.1146,
+      "step": 3657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4391141223947238e-05,
+      "loss": 0.8502,
+      "step": 3658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4388156414564889e-05,
+      "loss": 0.7828,
+      "step": 3659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4385171120926482e-05,
+      "loss": 0.7112,
+      "step": 3660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4382185343361461e-05,
+      "loss": 0.9791,
+      "step": 3661
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437919908219932e-05,
+      "loss": 1.1309,
+      "step": 3662
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376212337769608e-05,
+      "loss": 1.0361,
+      "step": 3663
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373225110401929e-05,
+      "loss": 1.1432,
+      "step": 3664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437023740042594e-05,
+      "loss": 1.0386,
+      "step": 3665
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436724920817135e-05,
+      "loss": 1.3473,
+      "step": 3666
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364260533967918e-05,
+      "loss": 1.1735,
+      "step": 3667
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361271378145467e-05,
+      "loss": 0.8297,
+      "step": 3668
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435828174103386e-05,
+      "loss": 0.8462,
+      "step": 3669
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435529162296302e-05,
+      "loss": 1.0102,
+      "step": 3670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435230102426293e-05,
+      "loss": 1.4491,
+      "step": 3671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349309945263605e-05,
+      "loss": 1.0487,
+      "step": 3672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4346318386295137e-05,
+      "loss": 0.959,
+      "step": 3673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343326347687656e-05,
+      "loss": 0.9868,
+      "step": 3674
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340333829771348e-05,
+      "loss": 1.0506,
+      "step": 3675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337340832876457e-05,
+      "loss": 1.0393,
+      "step": 3676
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433434735733327e-05,
+      "loss": 0.9039,
+      "step": 3677
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331353403472136e-05,
+      "loss": 0.9993,
+      "step": 3678
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328358971623455e-05,
+      "loss": 0.9712,
+      "step": 3679
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325364062117678e-05,
+      "loss": 1.1945,
+      "step": 3680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322368675285306e-05,
+      "loss": 1.1994,
+      "step": 3681
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4319372811456898e-05,
+      "loss": 1.2013,
+      "step": 3682
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431637647096306e-05,
+      "loss": 1.0708,
+      "step": 3683
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313379654134458e-05,
+      "loss": 0.9643,
+      "step": 3684
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310382361301809e-05,
+      "loss": 1.3856,
+      "step": 3685
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307384592795872e-05,
+      "loss": 0.841,
+      "step": 3686
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430438634894747e-05,
+      "loss": 1.0299,
+      "step": 3687
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301387630087478e-05,
+      "loss": 0.9776,
+      "step": 3688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4298388436546816e-05,
+      "loss": 0.8935,
+      "step": 3689
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4295388768656463e-05,
+      "loss": 0.9874,
+      "step": 3690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4292388626747446e-05,
+      "loss": 0.9259,
+      "step": 3691
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4289388011150851e-05,
+      "loss": 0.8993,
+      "step": 3692
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4286386922197805e-05,
+      "loss": 1.0682,
+      "step": 3693
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4283385360219497e-05,
+      "loss": 1.1111,
+      "step": 3694
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.428038332554717e-05,
+      "loss": 1.0488,
+      "step": 3695
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4277380818512108e-05,
+      "loss": 1.2103,
+      "step": 3696
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274377839445652e-05,
+      "loss": 0.88,
+      "step": 3697
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4271374388679202e-05,
+      "loss": 0.7971,
+      "step": 3698
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268370466544204e-05,
+      "loss": 1.1068,
+      "step": 3699
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265366073372151e-05,
+      "loss": 0.7788,
+      "step": 3700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42623612094946e-05,
+      "loss": 1.0154,
+      "step": 3701
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425935587524315e-05,
+      "loss": 1.0623,
+      "step": 3702
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4256350070949456e-05,
+      "loss": 1.2346,
+      "step": 3703
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253343796945225e-05,
+      "loss": 1.0074,
+      "step": 3704
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250337053562212e-05,
+      "loss": 0.8524,
+      "step": 3705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247329841132232e-05,
+      "loss": 0.8141,
+      "step": 3706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244322159987145e-05,
+      "loss": 1.0277,
+      "step": 3707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424131401045886e-05,
+      "loss": 0.8315,
+      "step": 3708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238305392879349e-05,
+      "loss": 0.9132,
+      "step": 3709
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235296307580621e-05,
+      "loss": 1.137,
+      "step": 3710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423228675489475e-05,
+      "loss": 1.1309,
+      "step": 3711
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229276735153858e-05,
+      "loss": 1.3212,
+      "step": 3712
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422626624869011e-05,
+      "loss": 1.0444,
+      "step": 3713
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422325529583573e-05,
+      "loss": 1.2368,
+      "step": 3714
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220243876922996e-05,
+      "loss": 1.1393,
+      "step": 3715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4217231992284232e-05,
+      "loss": 1.1444,
+      "step": 3716
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4214219642251816e-05,
+      "loss": 1.1906,
+      "step": 3717
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211206827158175e-05,
+      "loss": 0.7461,
+      "step": 3718
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4208193547335792e-05,
+      "loss": 0.9418,
+      "step": 3719
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205179803117193e-05,
+      "loss": 1.0713,
+      "step": 3720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202165594834964e-05,
+      "loss": 1.2907,
+      "step": 3721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4199150922821741e-05,
+      "loss": 1.3457,
+      "step": 3722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4196135787410205e-05,
+      "loss": 0.993,
+      "step": 3723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4193120188933094e-05,
+      "loss": 1.0568,
+      "step": 3724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.419010412772319e-05,
+      "loss": 0.842,
+      "step": 3725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.418708760411334e-05,
+      "loss": 0.7755,
+      "step": 3726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4184070618436427e-05,
+      "loss": 0.706,
+      "step": 3727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181053171025392e-05,
+      "loss": 0.8238,
+      "step": 3728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4178035262213228e-05,
+      "loss": 0.8557,
+      "step": 3729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175016892332975e-05,
+      "loss": 1.0997,
+      "step": 3730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4171998061717726e-05,
+      "loss": 1.191,
+      "step": 3731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416897877070063e-05,
+      "loss": 1.2389,
+      "step": 3732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165959019614875e-05,
+      "loss": 0.7744,
+      "step": 3733
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162938808793708e-05,
+      "loss": 0.9367,
+      "step": 3734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4159918138570425e-05,
+      "loss": 0.9462,
+      "step": 3735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156897009278374e-05,
+      "loss": 1.0604,
+      "step": 3736
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415387542125095e-05,
+      "loss": 1.2284,
+      "step": 3737
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150853374821607e-05,
+      "loss": 1.1424,
+      "step": 3738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147830870323838e-05,
+      "loss": 0.997,
+      "step": 3739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144807908091193e-05,
+      "loss": 1.1211,
+      "step": 3740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141784488457275e-05,
+      "loss": 0.8732,
+      "step": 3741
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4138760611755729e-05,
+      "loss": 0.9866,
+      "step": 3742
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135736278320262e-05,
+      "loss": 1.3116,
+      "step": 3743
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132711488484617e-05,
+      "loss": 1.1425,
+      "step": 3744
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4129686242582604e-05,
+      "loss": 0.8799,
+      "step": 3745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126660540948068e-05,
+      "loss": 1.1751,
+      "step": 3746
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123634383914915e-05,
+      "loss": 0.9441,
+      "step": 3747
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120607771817097e-05,
+      "loss": 0.5857,
+      "step": 3748
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117580704988614e-05,
+      "loss": 1.0344,
+      "step": 3749
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411455318376352e-05,
+      "loss": 1.0497,
+      "step": 3750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4111525208475918e-05,
+      "loss": 0.936,
+      "step": 3751
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4108496779459962e-05,
+      "loss": 1.097,
+      "step": 3752
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105467897049852e-05,
+      "loss": 0.8542,
+      "step": 3753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4102438561579845e-05,
+      "loss": 1.1052,
+      "step": 3754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.409940877338424e-05,
+      "loss": 0.7931,
+      "step": 3755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4096378532797392e-05,
+      "loss": 0.7436,
+      "step": 3756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4093347840153708e-05,
+      "loss": 0.6784,
+      "step": 3757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4090316695787634e-05,
+      "loss": 1.0954,
+      "step": 3758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4087285100033673e-05,
+      "loss": 1.3401,
+      "step": 3759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4084253053226386e-05,
+      "loss": 0.9152,
+      "step": 3760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4081220555700364e-05,
+      "loss": 0.9961,
+      "step": 3761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4078187607790268e-05,
+      "loss": 0.8981,
+      "step": 3762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4075154209830793e-05,
+      "loss": 1.0017,
+      "step": 3763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4072120362156695e-05,
+      "loss": 0.7766,
+      "step": 3764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069086065102773e-05,
+      "loss": 1.0822,
+      "step": 3765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066051319003877e-05,
+      "loss": 0.9,
+      "step": 3766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063016124194908e-05,
+      "loss": 0.6479,
+      "step": 3767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059980481010818e-05,
+      "loss": 0.9719,
+      "step": 3768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40569443897866e-05,
+      "loss": 0.8612,
+      "step": 3769
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405390785085731e-05,
+      "loss": 1.0697,
+      "step": 3770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405087086455804e-05,
+      "loss": 0.9376,
+      "step": 3771
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047833431223938e-05,
+      "loss": 1.0962,
+      "step": 3772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044795551190205e-05,
+      "loss": 0.8976,
+      "step": 3773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041757224792087e-05,
+      "loss": 0.8229,
+      "step": 3774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4038718452364872e-05,
+      "loss": 1.1261,
+      "step": 3775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035679234243912e-05,
+      "loss": 0.7496,
+      "step": 3776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032639570764595e-05,
+      "loss": 0.9538,
+      "step": 3777
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4029599462262369e-05,
+      "loss": 0.9428,
+      "step": 3778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026558909072722e-05,
+      "loss": 0.9835,
+      "step": 3779
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023517911531196e-05,
+      "loss": 0.9741,
+      "step": 3780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020476469973383e-05,
+      "loss": 1.2054,
+      "step": 3781
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4017434584734921e-05,
+      "loss": 0.8022,
+      "step": 3782
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014392256151495e-05,
+      "loss": 1.0728,
+      "step": 3783
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4011349484558846e-05,
+      "loss": 1.1088,
+      "step": 3784
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008306270292758e-05,
+      "loss": 1.0786,
+      "step": 3785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005262613689064e-05,
+      "loss": 0.9777,
+      "step": 3786
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.400221851508365e-05,
+      "loss": 0.7632,
+      "step": 3787
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3999173974812452e-05,
+      "loss": 1.1966,
+      "step": 3788
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.399612899321144e-05,
+      "loss": 1.1658,
+      "step": 3789
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3993083570616652e-05,
+      "loss": 0.826,
+      "step": 3790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3990037707364166e-05,
+      "loss": 0.8113,
+      "step": 3791
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986991403790107e-05,
+      "loss": 0.911,
+      "step": 3792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3983944660230648e-05,
+      "loss": 1.1254,
+      "step": 3793
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980897477022019e-05,
+      "loss": 0.963,
+      "step": 3794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977849854500492e-05,
+      "loss": 1.1156,
+      "step": 3795
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974801793002382e-05,
+      "loss": 0.8081,
+      "step": 3796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3971753292864061e-05,
+      "loss": 1.046,
+      "step": 3797
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968704354421953e-05,
+      "loss": 0.981,
+      "step": 3798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3965654978012517e-05,
+      "loss": 1.1902,
+      "step": 3799
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962605163972271e-05,
+      "loss": 1.0125,
+      "step": 3800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395955491263778e-05,
+      "loss": 0.9806,
+      "step": 3801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395650422434565e-05,
+      "loss": 1.2002,
+      "step": 3802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953453099432541e-05,
+      "loss": 0.993,
+      "step": 3803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950401538235166e-05,
+      "loss": 1.0221,
+      "step": 3804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947349541090276e-05,
+      "loss": 0.796,
+      "step": 3805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944297108334675e-05,
+      "loss": 0.9417,
+      "step": 3806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941244240305219e-05,
+      "loss": 1.2436,
+      "step": 3807
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3938190937338803e-05,
+      "loss": 0.8559,
+      "step": 3808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393513719977238e-05,
+      "loss": 0.9704,
+      "step": 3809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393208302794294e-05,
+      "loss": 0.773,
+      "step": 3810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3929028422187531e-05,
+      "loss": 1.0689,
+      "step": 3811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925973382843245e-05,
+      "loss": 1.0354,
+      "step": 3812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392291791024722e-05,
+      "loss": 0.9958,
+      "step": 3813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391986200473664e-05,
+      "loss": 0.8157,
+      "step": 3814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3916805666648749e-05,
+      "loss": 1.2483,
+      "step": 3815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913748896320822e-05,
+      "loss": 1.0069,
+      "step": 3816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3910691694090192e-05,
+      "loss": 0.9633,
+      "step": 3817
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907634060294241e-05,
+      "loss": 1.0712,
+      "step": 3818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.390457599527039e-05,
+      "loss": 0.9341,
+      "step": 3819
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3901517499356116e-05,
+      "loss": 1.0035,
+      "step": 3820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3898458572888937e-05,
+      "loss": 0.9853,
+      "step": 3821
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3895399216206422e-05,
+      "loss": 0.9362,
+      "step": 3822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.389233942964619e-05,
+      "loss": 0.9396,
+      "step": 3823
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3889279213545902e-05,
+      "loss": 0.951,
+      "step": 3824
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886218568243268e-05,
+      "loss": 0.8333,
+      "step": 3825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883157494076048e-05,
+      "loss": 0.9452,
+      "step": 3826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880095991382048e-05,
+      "loss": 0.7672,
+      "step": 3827
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877034060499116e-05,
+      "loss": 1.0377,
+      "step": 3828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387397170176516e-05,
+      "loss": 0.8587,
+      "step": 3829
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870908915518122e-05,
+      "loss": 1.1537,
+      "step": 3830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867845702095997e-05,
+      "loss": 1.1582,
+      "step": 3831
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864782061836833e-05,
+      "loss": 0.9727,
+      "step": 3832
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3861717995078707e-05,
+      "loss": 1.1915,
+      "step": 3833
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858653502159762e-05,
+      "loss": 1.0227,
+      "step": 3834
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855588583418183e-05,
+      "loss": 0.9025,
+      "step": 3835
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852523239192195e-05,
+      "loss": 1.1003,
+      "step": 3836
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849457469820078e-05,
+      "loss": 1.0136,
+      "step": 3837
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846391275640154e-05,
+      "loss": 0.9408,
+      "step": 3838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3843324656990795e-05,
+      "loss": 1.0104,
+      "step": 3839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840257614210415e-05,
+      "loss": 0.8786,
+      "step": 3840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383719014763748e-05,
+      "loss": 0.9834,
+      "step": 3841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834122257610507e-05,
+      "loss": 1.1932,
+      "step": 3842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3831053944468045e-05,
+      "loss": 0.919,
+      "step": 3843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827985208548698e-05,
+      "loss": 0.9551,
+      "step": 3844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3824916050191126e-05,
+      "loss": 0.9332,
+      "step": 3845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821846469734021e-05,
+      "loss": 1.2434,
+      "step": 3846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818776467516127e-05,
+      "loss": 1.0492,
+      "step": 3847
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815706043876236e-05,
+      "loss": 1.0115,
+      "step": 3848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812635199153188e-05,
+      "loss": 0.9498,
+      "step": 3849
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809563933685859e-05,
+      "loss": 0.8882,
+      "step": 3850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3806492247813187e-05,
+      "loss": 1.225,
+      "step": 3851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3803420141874146e-05,
+      "loss": 1.3019,
+      "step": 3852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3800347616207759e-05,
+      "loss": 0.9185,
+      "step": 3853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3797274671153093e-05,
+      "loss": 1.247,
+      "step": 3854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3794201307049266e-05,
+      "loss": 0.8283,
+      "step": 3855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3791127524235442e-05,
+      "loss": 0.958,
+      "step": 3856
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3788053323050825e-05,
+      "loss": 0.9775,
+      "step": 3857
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3784978703834671e-05,
+      "loss": 1.0548,
+      "step": 3858
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781903666926281e-05,
+      "loss": 0.8374,
+      "step": 3859
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778828212665e-05,
+      "loss": 1.0376,
+      "step": 3860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377575234139022e-05,
+      "loss": 1.0365,
+      "step": 3861
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772676053441383e-05,
+      "loss": 0.9875,
+      "step": 3862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3769599349157972e-05,
+      "loss": 0.7482,
+      "step": 3863
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766522228879512e-05,
+      "loss": 0.9145,
+      "step": 3864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763444692945584e-05,
+      "loss": 1.2335,
+      "step": 3865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760366741695815e-05,
+      "loss": 0.8935,
+      "step": 3866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757288375469865e-05,
+      "loss": 1.1358,
+      "step": 3867
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375420959460745e-05,
+      "loss": 0.9271,
+      "step": 3868
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751130399448334e-05,
+      "loss": 0.9605,
+      "step": 3869
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748050790332317e-05,
+      "loss": 1.1645,
+      "step": 3870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374497076759925e-05,
+      "loss": 1.2268,
+      "step": 3871
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741890331589032e-05,
+      "loss": 1.062,
+      "step": 3872
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738809482641609e-05,
+      "loss": 0.9054,
+      "step": 3873
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735728221096963e-05,
+      "loss": 0.981,
+      "step": 3874
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732646547295128e-05,
+      "loss": 1.1105,
+      "step": 3875
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729564461576184e-05,
+      "loss": 1.0279,
+      "step": 3876
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3726481964280254e-05,
+      "loss": 0.7822,
+      "step": 3877
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723399055747511e-05,
+      "loss": 0.9765,
+      "step": 3878
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3720315736318166e-05,
+      "loss": 0.9839,
+      "step": 3879
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717232006332483e-05,
+      "loss": 0.9864,
+      "step": 3880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714147866130767e-05,
+      "loss": 0.8771,
+      "step": 3881
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3711063316053368e-05,
+      "loss": 0.9797,
+      "step": 3882
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707978356440683e-05,
+      "loss": 0.8369,
+      "step": 3883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3704892987633152e-05,
+      "loss": 1.0167,
+      "step": 3884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3701807209971262e-05,
+      "loss": 1.1729,
+      "step": 3885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3698721023795545e-05,
+      "loss": 1.3145,
+      "step": 3886
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3695634429446583e-05,
+      "loss": 0.9301,
+      "step": 3887
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3692547427264989e-05,
+      "loss": 0.8901,
+      "step": 3888
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3689460017591432e-05,
+      "loss": 1.0185,
+      "step": 3889
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368637220076663e-05,
+      "loss": 1.069,
+      "step": 3890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3683283977131331e-05,
+      "loss": 1.0534,
+      "step": 3891
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680195347026345e-05,
+      "loss": 1.3306,
+      "step": 3892
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677106310792514e-05,
+      "loss": 0.7929,
+      "step": 3893
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674016868770732e-05,
+      "loss": 1.0376,
+      "step": 3894
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367092702130193e-05,
+      "loss": 0.8006,
+      "step": 3895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667836768727094e-05,
+      "loss": 1.093,
+      "step": 3896
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664746111387249e-05,
+      "loss": 0.9221,
+      "step": 3897
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366165504962346e-05,
+      "loss": 0.8752,
+      "step": 3898
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3658563583776847e-05,
+      "loss": 0.9371,
+      "step": 3899
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655471714188568e-05,
+      "loss": 0.9889,
+      "step": 3900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365237944119983e-05,
+      "loss": 1.1885,
+      "step": 3901
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649286765151875e-05,
+      "loss": 0.9992,
+      "step": 3902
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646193686386e-05,
+      "loss": 0.9635,
+      "step": 3903
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643100205243547e-05,
+      "loss": 0.943,
+      "step": 3904
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640006322065889e-05,
+      "loss": 0.9627,
+      "step": 3905
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636912037194456e-05,
+      "loss": 0.9309,
+      "step": 3906
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633817350970719e-05,
+      "loss": 1.05,
+      "step": 3907
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630722263736193e-05,
+      "loss": 1.663,
+      "step": 3908
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362762677583244e-05,
+      "loss": 1.0396,
+      "step": 3909
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624530887601055e-05,
+      "loss": 1.6688,
+      "step": 3910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3621434599383694e-05,
+      "loss": 1.1123,
+      "step": 3911
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618337911522044e-05,
+      "loss": 1.0196,
+      "step": 3912
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615240824357842e-05,
+      "loss": 0.615,
+      "step": 3913
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612143338232869e-05,
+      "loss": 0.8423,
+      "step": 3914
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609045453488947e-05,
+      "loss": 0.9969,
+      "step": 3915
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3605947170467942e-05,
+      "loss": 1.2024,
+      "step": 3916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3602848489511771e-05,
+      "loss": 1.0687,
+      "step": 3917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3599749410962387e-05,
+      "loss": 0.9848,
+      "step": 3918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3596649935161786e-05,
+      "loss": 1.1007,
+      "step": 3919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3593550062452017e-05,
+      "loss": 1.0007,
+      "step": 3920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3590449793175167e-05,
+      "loss": 1.0009,
+      "step": 3921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587349127673361e-05,
+      "loss": 0.9645,
+      "step": 3922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584248066288779e-05,
+      "loss": 1.1244,
+      "step": 3923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358114660936364e-05,
+      "loss": 1.0799,
+      "step": 3924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578044757240202e-05,
+      "loss": 0.8491,
+      "step": 3925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574942510260773e-05,
+      "loss": 0.938,
+      "step": 3926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571839868767704e-05,
+      "loss": 1.1766,
+      "step": 3927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568736833103385e-05,
+      "loss": 0.8891,
+      "step": 3928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565633403610252e-05,
+      "loss": 0.973,
+      "step": 3929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562529580630784e-05,
+      "loss": 1.0542,
+      "step": 3930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559425364507508e-05,
+      "loss": 1.1516,
+      "step": 3931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3556320755582988e-05,
+      "loss": 1.2722,
+      "step": 3932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553215754199834e-05,
+      "loss": 0.9198,
+      "step": 3933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3550110360700697e-05,
+      "loss": 0.9558,
+      "step": 3934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354700457542828e-05,
+      "loss": 1.1618,
+      "step": 3935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543898398725316e-05,
+      "loss": 0.9358,
+      "step": 3936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354079183093459e-05,
+      "loss": 1.2641,
+      "step": 3937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537684872398928e-05,
+      "loss": 1.0213,
+      "step": 3938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35345775234612e-05,
+      "loss": 1.292,
+      "step": 3939
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353146978446432e-05,
+      "loss": 1.1212,
+      "step": 3940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528361655751238e-05,
+      "loss": 1.0995,
+      "step": 3941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525253137664957e-05,
+      "loss": 1.1476,
+      "step": 3942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522144230548517e-05,
+      "loss": 1.054,
+      "step": 3943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3519034934745e-05,
+      "loss": 1.0596,
+      "step": 3944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515925250597537e-05,
+      "loss": 1.1743,
+      "step": 3945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512815178449294e-05,
+      "loss": 0.9104,
+      "step": 3946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509704718643486e-05,
+      "loss": 1.0676,
+      "step": 3947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350659387152337e-05,
+      "loss": 0.9101,
+      "step": 3948
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3503482637432242e-05,
+      "loss": 1.0986,
+      "step": 3949
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.350037101671344e-05,
+      "loss": 1.0188,
+      "step": 3950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3497259009710353e-05,
+      "loss": 1.1268,
+      "step": 3951
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3494146616766406e-05,
+      "loss": 1.0083,
+      "step": 3952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3491033838225067e-05,
+      "loss": 1.0074,
+      "step": 3953
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3487920674429847e-05,
+      "loss": 0.9153,
+      "step": 3954
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484807125724302e-05,
+      "loss": 0.9101,
+      "step": 3955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481693192452026e-05,
+      "loss": 1.1263,
+      "step": 3956
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478578874956659e-05,
+      "loss": 0.9924,
+      "step": 3957
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475464173581882e-05,
+      "loss": 0.8717,
+      "step": 3958
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347234908867142e-05,
+      "loss": 0.7618,
+      "step": 3959
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469233620569036e-05,
+      "loss": 1.1164,
+      "step": 3960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466117769618538e-05,
+      "loss": 1.1257,
+      "step": 3961
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463001536163782e-05,
+      "loss": 0.803,
+      "step": 3962
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459884920548655e-05,
+      "loss": 0.8205,
+      "step": 3963
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456767923117093e-05,
+      "loss": 0.7144,
+      "step": 3964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453650544213078e-05,
+      "loss": 1.0737,
+      "step": 3965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450532784180621e-05,
+      "loss": 1.1617,
+      "step": 3966
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344741464336379e-05,
+      "loss": 0.9848,
+      "step": 3967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444296122106684e-05,
+      "loss": 0.8439,
+      "step": 3968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3441177220753452e-05,
+      "loss": 0.875,
+      "step": 3969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438057939648275e-05,
+      "loss": 0.8072,
+      "step": 3970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3434938279135388e-05,
+      "loss": 0.7183,
+      "step": 3971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431818239559057e-05,
+      "loss": 0.9072,
+      "step": 3972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34286978212636e-05,
+      "loss": 0.8346,
+      "step": 3973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425577024593367e-05,
+      "loss": 1.1929,
+      "step": 3974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422455849892757e-05,
+      "loss": 0.9018,
+      "step": 3975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341933429750621e-05,
+      "loss": 1.7579,
+      "step": 3976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3416212367778197e-05,
+      "loss": 1.0824,
+      "step": 3977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413090061053249e-05,
+      "loss": 1.1204,
+      "step": 3978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3409967377675925e-05,
+      "loss": 0.9056,
+      "step": 3979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406844317990826e-05,
+      "loss": 1.0826,
+      "step": 3980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3403720882342603e-05,
+      "loss": 0.8128,
+      "step": 3981
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3400597071075944e-05,
+      "loss": 0.9442,
+      "step": 3982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3397472884535573e-05,
+      "loss": 0.8307,
+      "step": 3983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3394348323066267e-05,
+      "loss": 1.1303,
+      "step": 3984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3391223387012831e-05,
+      "loss": 1.0964,
+      "step": 3985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3388098076720122e-05,
+      "loss": 1.0512,
+      "step": 3986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3384972392533033e-05,
+      "loss": 0.7458,
+      "step": 3987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381846334796498e-05,
+      "loss": 1.0103,
+      "step": 3988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33787199038555e-05,
+      "loss": 1.1086,
+      "step": 3989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337559310005505e-05,
+      "loss": 1.1224,
+      "step": 3990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337246592374021e-05,
+      "loss": 1.0914,
+      "step": 3991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336933837525608e-05,
+      "loss": 1.2034,
+      "step": 3992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366210454947806e-05,
+      "loss": 0.807,
+      "step": 3993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363082163160562e-05,
+      "loss": 1.0613,
+      "step": 3994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359953500239575e-05,
+      "loss": 0.9309,
+      "step": 3995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356824466530116e-05,
+      "loss": 1.0782,
+      "step": 3996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3353695062377478e-05,
+      "loss": 1.1783,
+      "step": 3997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350565288127015e-05,
+      "loss": 0.7871,
+      "step": 3998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347435144124112e-05,
+      "loss": 0.9774,
+      "step": 3999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344304630714198e-05,
+      "loss": 0.9512,
+      "step": 4000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341173748242739e-05,
+      "loss": 0.9215,
+      "step": 4001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338042497055246e-05,
+      "loss": 1.1983,
+      "step": 4002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334910877497274e-05,
+      "loss": 0.8292,
+      "step": 4003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3331778889914403e-05,
+      "loss": 0.7726,
+      "step": 4004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3328646534652272e-05,
+      "loss": 0.9123,
+      "step": 4005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325513812056552e-05,
+      "loss": 1.0147,
+      "step": 4006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322380722472953e-05,
+      "loss": 0.6528,
+      "step": 4007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3319247266247226e-05,
+      "loss": 0.816,
+      "step": 4008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3316113443725169e-05,
+      "loss": 0.9257,
+      "step": 4009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312979255252617e-05,
+      "loss": 0.9316,
+      "step": 4010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309844701175437e-05,
+      "loss": 0.9822,
+      "step": 4011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330670978183955e-05,
+      "loss": 0.9737,
+      "step": 4012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303574497590906e-05,
+      "loss": 0.7343,
+      "step": 4013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3300438848775504e-05,
+      "loss": 1.0502,
+      "step": 4014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3297302835739377e-05,
+      "loss": 1.0715,
+      "step": 4015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32941664588286e-05,
+      "loss": 0.9267,
+      "step": 4016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3291029718389293e-05,
+      "loss": 0.8384,
+      "step": 4017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3287892614767606e-05,
+      "loss": 0.9732,
+      "step": 4018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3284755148309735e-05,
+      "loss": 0.7436,
+      "step": 4019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281617319361924e-05,
+      "loss": 0.92,
+      "step": 4020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278479128270439e-05,
+      "loss": 1.0291,
+      "step": 4021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32753405753816e-05,
+      "loss": 0.8247,
+      "step": 4022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272201661041764e-05,
+      "loss": 0.9316,
+      "step": 4023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269062385597322e-05,
+      "loss": 0.9383,
+      "step": 4024
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265922749394717e-05,
+      "loss": 1.1679,
+      "step": 4025
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262782752780418e-05,
+      "loss": 0.9308,
+      "step": 4026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3259642396100942e-05,
+      "loss": 1.0696,
+      "step": 4027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256501679702844e-05,
+      "loss": 1.1084,
+      "step": 4028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253360603932721e-05,
+      "loss": 1.0153,
+      "step": 4029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250219169137202e-05,
+      "loss": 0.8795,
+      "step": 4030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247077375662967e-05,
+      "loss": 0.8538,
+      "step": 4031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243935223856721e-05,
+      "loss": 1.079,
+      "step": 4032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240792714065222e-05,
+      "loss": 1.1126,
+      "step": 4033
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237649846635266e-05,
+      "loss": 1.0539,
+      "step": 4034
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323450662191368e-05,
+      "loss": 1.2143,
+      "step": 4035
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231363040247333e-05,
+      "loss": 0.7444,
+      "step": 4036
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228219101983145e-05,
+      "loss": 0.9892,
+      "step": 4037
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3225074807468054e-05,
+      "loss": 1.172,
+      "step": 4038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221930157049058e-05,
+      "loss": 0.8577,
+      "step": 4039
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3218785151073185e-05,
+      "loss": 1.0218,
+      "step": 4040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32156397898875e-05,
+      "loss": 0.9051,
+      "step": 4041
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321249407383911e-05,
+      "loss": 1.0289,
+      "step": 4042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209348003275163e-05,
+      "loss": 1.0906,
+      "step": 4043
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206201578542846e-05,
+      "loss": 1.1269,
+      "step": 4044
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320305479998938e-05,
+      "loss": 1.1006,
+      "step": 4045
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199907667962028e-05,
+      "loss": 0.9531,
+      "step": 4046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3196760182808095e-05,
+      "loss": 0.8793,
+      "step": 4047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3193612344874924e-05,
+      "loss": 1.084,
+      "step": 4048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3190464154509887e-05,
+      "loss": 1.2086,
+      "step": 4049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3187315612060417e-05,
+      "loss": 0.7748,
+      "step": 4050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3184166717873964e-05,
+      "loss": 0.9601,
+      "step": 4051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.318101747229802e-05,
+      "loss": 1.1825,
+      "step": 4052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177867875680131e-05,
+      "loss": 0.7857,
+      "step": 4053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174717928367864e-05,
+      "loss": 1.5533,
+      "step": 4054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171567630708838e-05,
+      "loss": 0.8329,
+      "step": 4055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31684169830507e-05,
+      "loss": 0.9517,
+      "step": 4056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165265985741142e-05,
+      "loss": 0.8963,
+      "step": 4057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162114639127896e-05,
+      "loss": 0.91,
+      "step": 4058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158962943558726e-05,
+      "loss": 1.1878,
+      "step": 4059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155810899381438e-05,
+      "loss": 1.0027,
+      "step": 4060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3152658506943878e-05,
+      "loss": 0.7997,
+      "step": 4061
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314950576659393e-05,
+      "loss": 1.2559,
+      "step": 4062
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146352678679513e-05,
+      "loss": 0.9901,
+      "step": 4063
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143199243548588e-05,
+      "loss": 1.524,
+      "step": 4064
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140045461549155e-05,
+      "loss": 0.8338,
+      "step": 4065
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136891333029247e-05,
+      "loss": 1.0021,
+      "step": 4066
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313373685833694e-05,
+      "loss": 1.1933,
+      "step": 4067
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313058203782035e-05,
+      "loss": 0.9644,
+      "step": 4068
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312742687182762e-05,
+      "loss": 0.8934,
+      "step": 4069
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3124271360706949e-05,
+      "loss": 1.037,
+      "step": 4070
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121115504806554e-05,
+      "loss": 0.805,
+      "step": 4071
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117959304474707e-05,
+      "loss": 0.8632,
+      "step": 4072
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3114802760059713e-05,
+      "loss": 0.785,
+      "step": 4073
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111645871909908e-05,
+      "loss": 1.0183,
+      "step": 4074
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108488640373673e-05,
+      "loss": 0.949,
+      "step": 4075
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105331065799425e-05,
+      "loss": 0.9796,
+      "step": 4076
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102173148535617e-05,
+      "loss": 1.0845,
+      "step": 4077
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099014888930747e-05,
+      "loss": 1.3808,
+      "step": 4078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3095856287333341e-05,
+      "loss": 1.1097,
+      "step": 4079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3092697344091967e-05,
+      "loss": 0.9766,
+      "step": 4080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3089538059555233e-05,
+      "loss": 0.9293,
+      "step": 4081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3086378434071785e-05,
+      "loss": 0.8782,
+      "step": 4082
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3083218467990297e-05,
+      "loss": 0.9618,
+      "step": 4083
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3080058161659494e-05,
+      "loss": 0.8609,
+      "step": 4084
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076897515428132e-05,
+      "loss": 0.8852,
+      "step": 4085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073736529645003e-05,
+      "loss": 0.9881,
+      "step": 4086
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070575204658937e-05,
+      "loss": 1.0232,
+      "step": 4087
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067413540818806e-05,
+      "loss": 0.8327,
+      "step": 4088
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064251538473514e-05,
+      "loss": 0.9788,
+      "step": 4089
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3061089197972004e-05,
+      "loss": 1.3518,
+      "step": 4090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057926519663258e-05,
+      "loss": 0.7292,
+      "step": 4091
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3054763503896295e-05,
+      "loss": 0.944,
+      "step": 4092
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305160015102017e-05,
+      "loss": 0.8878,
+      "step": 4093
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048436461383973e-05,
+      "loss": 1.0898,
+      "step": 4094
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045272435336839e-05,
+      "loss": 1.1612,
+      "step": 4095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304210807322793e-05,
+      "loss": 1.1323,
+      "step": 4096
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303894337540645e-05,
+      "loss": 1.126,
+      "step": 4097
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035778342221642e-05,
+      "loss": 0.7596,
+      "step": 4098
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3032612974022785e-05,
+      "loss": 1.1488,
+      "step": 4099
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029447271159192e-05,
+      "loss": 0.9226,
+      "step": 4100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026281233980216e-05,
+      "loss": 0.8579,
+      "step": 4101
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023114862835247e-05,
+      "loss": 0.9043,
+      "step": 4102
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019948158073707e-05,
+      "loss": 0.7705,
+      "step": 4103
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016781120045059e-05,
+      "loss": 0.8508,
+      "step": 4104
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3013613749098806e-05,
+      "loss": 1.0667,
+      "step": 4105
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301044604558448e-05,
+      "loss": 0.9192,
+      "step": 4106
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007278009851655e-05,
+      "loss": 0.7181,
+      "step": 4107
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300410964224994e-05,
+      "loss": 1.0321,
+      "step": 4108
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000940943128986e-05,
+      "loss": 1.0919,
+      "step": 4109
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997771912838466e-05,
+      "loss": 1.0224,
+      "step": 4110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994602551728107e-05,
+      "loss": 1.1867,
+      "step": 4111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2991432860147662e-05,
+      "loss": 0.7643,
+      "step": 4112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2988262838446922e-05,
+      "loss": 1.1703,
+      "step": 4113
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2985092486975716e-05,
+      "loss": 0.8553,
+      "step": 4114
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.298192180608391e-05,
+      "loss": 1.1685,
+      "step": 4115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2978750796121403e-05,
+      "loss": 1.1294,
+      "step": 4116
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2975579457438138e-05,
+      "loss": 0.6991,
+      "step": 4117
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972407790384082e-05,
+      "loss": 1.1475,
+      "step": 4118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296923579530925e-05,
+      "loss": 0.9593,
+      "step": 4119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966063472563686e-05,
+      "loss": 0.9878,
+      "step": 4120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2962890822497471e-05,
+      "loss": 1.0708,
+      "step": 4121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959717845460724e-05,
+      "loss": 1.0662,
+      "step": 4122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2956544541803609e-05,
+      "loss": 0.9464,
+      "step": 4123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29533709118763e-05,
+      "loss": 1.054,
+      "step": 4124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950196956029036e-05,
+      "loss": 0.9003,
+      "step": 4125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947022674612078e-05,
+      "loss": 0.9314,
+      "step": 4126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943848067975718e-05,
+      "loss": 0.872,
+      "step": 4127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940673136470299e-05,
+      "loss": 0.9193,
+      "step": 4128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937497880446186e-05,
+      "loss": 1.0773,
+      "step": 4129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934322300253785e-05,
+      "loss": 1.025,
+      "step": 4130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293114639624354e-05,
+      "loss": 1.1016,
+      "step": 4131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927970168765933e-05,
+      "loss": 1.0676,
+      "step": 4132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2924793618171468e-05,
+      "loss": 0.8457,
+      "step": 4133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29216167448107e-05,
+      "loss": 0.7644,
+      "step": 4134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291843954903421e-05,
+      "loss": 1.1201,
+      "step": 4135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915262031192623e-05,
+      "loss": 1.1313,
+      "step": 4136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291208419163659e-05,
+      "loss": 1.1191,
+      "step": 4137
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908906030716805e-05,
+      "loss": 0.9407,
+      "step": 4138
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905727548783995e-05,
+      "loss": 0.9537,
+      "step": 4139
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902548746188923e-05,
+      "loss": 0.7337,
+      "step": 4140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899369623282381e-05,
+      "loss": 0.8463,
+      "step": 4141
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896190180415209e-05,
+      "loss": 1.0953,
+      "step": 4142
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893010417938273e-05,
+      "loss": 0.9469,
+      "step": 4143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.288983033620247e-05,
+      "loss": 0.9193,
+      "step": 4144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2886649935558746e-05,
+      "loss": 0.8141,
+      "step": 4145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2883469216358073e-05,
+      "loss": 0.8941,
+      "step": 4146
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2880288178951461e-05,
+      "loss": 0.9104,
+      "step": 4147
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2877106823689951e-05,
+      "loss": 1.276,
+      "step": 4148
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2873925150924624e-05,
+      "loss": 0.76,
+      "step": 4149
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2870743161006599e-05,
+      "loss": 0.7684,
+      "step": 4150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867560854287016e-05,
+      "loss": 0.7717,
+      "step": 4151
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864378231117063e-05,
+      "loss": 0.8277,
+      "step": 4152
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861195291847966e-05,
+      "loss": 0.998,
+      "step": 4153
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858012036830969e-05,
+      "loss": 1.0371,
+      "step": 4154
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854828466417363e-05,
+      "loss": 0.9482,
+      "step": 4155
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285164458095848e-05,
+      "loss": 1.0109,
+      "step": 4156
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284846038080567e-05,
+      "loss": 0.9732,
+      "step": 4157
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2845275866310325e-05,
+      "loss": 1.1134,
+      "step": 4158
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842091037823879e-05,
+      "loss": 0.9223,
+      "step": 4159
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838905895697792e-05,
+      "loss": 1.0998,
+      "step": 4160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2835720440283561e-05,
+      "loss": 0.9451,
+      "step": 4161
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832534671932715e-05,
+      "loss": 1.2866,
+      "step": 4162
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2829348590996826e-05,
+      "loss": 0.812,
+      "step": 4163
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282616219782749e-05,
+      "loss": 0.939,
+      "step": 4164
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822975492776344e-05,
+      "loss": 1.0422,
+      "step": 4165
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819788476195055e-05,
+      "loss": 0.986,
+      "step": 4166
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816601148435332e-05,
+      "loss": 0.8023,
+      "step": 4167
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813413509848911e-05,
+      "loss": 0.6123,
+      "step": 4168
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810225560787561e-05,
+      "loss": 0.8654,
+      "step": 4169
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807037301603093e-05,
+      "loss": 0.9083,
+      "step": 4170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280384873264735e-05,
+      "loss": 1.002,
+      "step": 4171
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800659854272205e-05,
+      "loss": 1.0642,
+      "step": 4172
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797470666829563e-05,
+      "loss": 1.033,
+      "step": 4173
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794281170671374e-05,
+      "loss": 0.8498,
+      "step": 4174
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791091366149613e-05,
+      "loss": 0.8961,
+      "step": 4175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278790125361629e-05,
+      "loss": 0.8653,
+      "step": 4176
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2784710833423454e-05,
+      "loss": 1.285,
+      "step": 4177
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2781520105923184e-05,
+      "loss": 0.8858,
+      "step": 4178
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2778329071467592e-05,
+      "loss": 0.9262,
+      "step": 4179
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2775137730408825e-05,
+      "loss": 1.2273,
+      "step": 4180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2771946083099065e-05,
+      "loss": 0.9541,
+      "step": 4181
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2768754129890528e-05,
+      "loss": 0.9549,
+      "step": 4182
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276556187113546e-05,
+      "loss": 1.0103,
+      "step": 4183
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762369307186148e-05,
+      "loss": 0.8673,
+      "step": 4184
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759176438394903e-05,
+      "loss": 0.9728,
+      "step": 4185
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755983265114078e-05,
+      "loss": 1.125,
+      "step": 4186
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2752789787696056e-05,
+      "loss": 0.9949,
+      "step": 4187
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749596006493254e-05,
+      "loss": 1.0242,
+      "step": 4188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746401921858125e-05,
+      "loss": 0.825,
+      "step": 4189
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2743207534143145e-05,
+      "loss": 1.0715,
+      "step": 4190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274001284370084e-05,
+      "loss": 1.1342,
+      "step": 4191
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736817850883755e-05,
+      "loss": 0.9601,
+      "step": 4192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273362255604448e-05,
+      "loss": 0.7606,
+      "step": 4193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730426959535627e-05,
+      "loss": 0.9794,
+      "step": 4194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272723106170985e-05,
+      "loss": 1.4652,
+      "step": 4195
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272403486291983e-05,
+      "loss": 0.9312,
+      "step": 4196
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2720838363518287e-05,
+      "loss": 1.0082,
+      "step": 4197
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717641563857975e-05,
+      "loss": 0.9562,
+      "step": 4198
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714444464291666e-05,
+      "loss": 1.0533,
+      "step": 4199
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271124706517219e-05,
+      "loss": 0.9958,
+      "step": 4200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708049366852389e-05,
+      "loss": 0.6512,
+      "step": 4201
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704851369685147e-05,
+      "loss": 0.9732,
+      "step": 4202
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701653074023382e-05,
+      "loss": 1.083,
+      "step": 4203
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698454480220039e-05,
+      "loss": 0.8201,
+      "step": 4204
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695255588628103e-05,
+      "loss": 0.943,
+      "step": 4205
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692056399600588e-05,
+      "loss": 0.7562,
+      "step": 4206
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688856913490537e-05,
+      "loss": 1.0702,
+      "step": 4207
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685657130651036e-05,
+      "loss": 1.2422,
+      "step": 4208
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2682457051435195e-05,
+      "loss": 0.9943,
+      "step": 4209
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2679256676196158e-05,
+      "loss": 0.9642,
+      "step": 4210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2676056005287107e-05,
+      "loss": 1.4751,
+      "step": 4211
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2672855039061252e-05,
+      "loss": 0.8517,
+      "step": 4212
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2669653777871833e-05,
+      "loss": 0.9718,
+      "step": 4213
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2666452222072127e-05,
+      "loss": 0.8227,
+      "step": 4214
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663250372015447e-05,
+      "loss": 0.9356,
+      "step": 4215
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660048228055129e-05,
+      "loss": 1.079,
+      "step": 4216
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656845790544546e-05,
+      "loss": 1.0266,
+      "step": 4217
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653643059837109e-05,
+      "loss": 1.5003,
+      "step": 4218
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650440036286249e-05,
+      "loss": 0.903,
+      "step": 4219
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647236720245442e-05,
+      "loss": 0.913,
+      "step": 4220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264403311206819e-05,
+      "loss": 1.0152,
+      "step": 4221
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2640829212108026e-05,
+      "loss": 0.9809,
+      "step": 4222
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637625020718518e-05,
+      "loss": 1.1325,
+      "step": 4223
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634420538253263e-05,
+      "loss": 0.7476,
+      "step": 4224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631215765065899e-05,
+      "loss": 0.8885,
+      "step": 4225
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628010701510082e-05,
+      "loss": 0.9137,
+      "step": 4226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262480534793951e-05,
+      "loss": 0.9453,
+      "step": 4227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621599704707914e-05,
+      "loss": 0.7624,
+      "step": 4228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618393772169053e-05,
+      "loss": 1.0021,
+      "step": 4229
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261518755067671e-05,
+      "loss": 0.9123,
+      "step": 4230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261198104058472e-05,
+      "loss": 0.8485,
+      "step": 4231
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2608774242246931e-05,
+      "loss": 0.9048,
+      "step": 4232
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605567156017233e-05,
+      "loss": 0.9249,
+      "step": 4233
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602359782249543e-05,
+      "loss": 1.0037,
+      "step": 4234
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599152121297813e-05,
+      "loss": 0.9824,
+      "step": 4235
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595944173516022e-05,
+      "loss": 1.2441,
+      "step": 4236
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592735939258187e-05,
+      "loss": 0.9142,
+      "step": 4237
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2589527418878354e-05,
+      "loss": 0.9205,
+      "step": 4238
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.25863186127306e-05,
+      "loss": 0.8827,
+      "step": 4239
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258310952116903e-05,
+      "loss": 0.9909,
+      "step": 4240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579900144547785e-05,
+      "loss": 1.1891,
+      "step": 4241
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2576690483221041e-05,
+      "loss": 0.7623,
+      "step": 4242
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2573480537542996e-05,
+      "loss": 1.0943,
+      "step": 4243
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2570270307867887e-05,
+      "loss": 1.1016,
+      "step": 4244
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2567059794549978e-05,
+      "loss": 1.3648,
+      "step": 4245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2563848997943569e-05,
+      "loss": 0.8856,
+      "step": 4246
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2560637918402985e-05,
+      "loss": 0.7559,
+      "step": 4247
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557426556282587e-05,
+      "loss": 0.902,
+      "step": 4248
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2554214911936765e-05,
+      "loss": 1.1298,
+      "step": 4249
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255100298571994e-05,
+      "loss": 1.0039,
+      "step": 4250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2547790777986567e-05,
+      "loss": 1.4312,
+      "step": 4251
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2544578289091127e-05,
+      "loss": 1.2029,
+      "step": 4252
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2541365519388137e-05,
+      "loss": 0.7494,
+      "step": 4253
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538152469232146e-05,
+      "loss": 0.9563,
+      "step": 4254
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2534939138977725e-05,
+      "loss": 0.8617,
+      "step": 4255
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531725528979486e-05,
+      "loss": 0.9159,
+      "step": 4256
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528511639592063e-05,
+      "loss": 1.0456,
+      "step": 4257
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252529747117013e-05,
+      "loss": 1.2095,
+      "step": 4258
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522083024068387e-05,
+      "loss": 0.8855,
+      "step": 4259
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2518868298641564e-05,
+      "loss": 1.2859,
+      "step": 4260
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515653295244424e-05,
+      "loss": 0.7836,
+      "step": 4261
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512438014231757e-05,
+      "loss": 1.0915,
+      "step": 4262
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250922245595839e-05,
+      "loss": 0.8618,
+      "step": 4263
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506006620779171e-05,
+      "loss": 0.9113,
+      "step": 4264
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250279050904899e-05,
+      "loss": 0.9763,
+      "step": 4265
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499574121122759e-05,
+      "loss": 1.222,
+      "step": 4266
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496357457355423e-05,
+      "loss": 0.7937,
+      "step": 4267
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493140518101958e-05,
+      "loss": 0.8808,
+      "step": 4268
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248992330371737e-05,
+      "loss": 1.0714,
+      "step": 4269
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2486705814556698e-05,
+      "loss": 0.9542,
+      "step": 4270
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483488050975006e-05,
+      "loss": 0.8804,
+      "step": 4271
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480270013327389e-05,
+      "loss": 1.0564,
+      "step": 4272
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477051701968982e-05,
+      "loss": 0.7862,
+      "step": 4273
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2473833117254933e-05,
+      "loss": 1.0367,
+      "step": 4274
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2470614259540436e-05,
+      "loss": 0.9971,
+      "step": 4275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2467395129180708e-05,
+      "loss": 0.8738,
+      "step": 4276
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2464175726530996e-05,
+      "loss": 0.8135,
+      "step": 4277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2460956051946575e-05,
+      "loss": 0.7901,
+      "step": 4278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2457736105782761e-05,
+      "loss": 0.7721,
+      "step": 4279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2454515888394882e-05,
+      "loss": 0.9159,
+      "step": 4280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451295400138315e-05,
+      "loss": 0.9305,
+      "step": 4281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448074641368448e-05,
+      "loss": 1.2479,
+      "step": 4282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244485361244072e-05,
+      "loss": 0.7943,
+      "step": 4283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441632313710579e-05,
+      "loss": 1.088,
+      "step": 4284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438410745533516e-05,
+      "loss": 1.0678,
+      "step": 4285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435188908265048e-05,
+      "loss": 1.1634,
+      "step": 4286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431966802260723e-05,
+      "loss": 0.9146,
+      "step": 4287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428744427876112e-05,
+      "loss": 1.1077,
+      "step": 4288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425521785466827e-05,
+      "loss": 1.0515,
+      "step": 4289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422298875388501e-05,
+      "loss": 0.7786,
+      "step": 4290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419075697996798e-05,
+      "loss": 0.9828,
+      "step": 4291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415852253647414e-05,
+      "loss": 0.6483,
+      "step": 4292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412628542696073e-05,
+      "loss": 1.0477,
+      "step": 4293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409404565498526e-05,
+      "loss": 0.8903,
+      "step": 4294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406180322410559e-05,
+      "loss": 0.7198,
+      "step": 4295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402955813787983e-05,
+      "loss": 1.2591,
+      "step": 4296
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239973103998664e-05,
+      "loss": 0.9421,
+      "step": 4297
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396506001362399e-05,
+      "loss": 0.9903,
+      "step": 4298
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393280698271157e-05,
+      "loss": 1.0308,
+      "step": 4299
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390055131068852e-05,
+      "loss": 0.9598,
+      "step": 4300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386829300111436e-05,
+      "loss": 0.9859,
+      "step": 4301
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383603205754895e-05,
+      "loss": 0.8895,
+      "step": 4302
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238037684835525e-05,
+      "loss": 1.0248,
+      "step": 4303
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377150228268548e-05,
+      "loss": 0.8728,
+      "step": 4304
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373923345850854e-05,
+      "loss": 1.0499,
+      "step": 4305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370696201458278e-05,
+      "loss": 1.1902,
+      "step": 4306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2367468795446952e-05,
+      "loss": 0.7636,
+      "step": 4307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.236424112817304e-05,
+      "loss": 1.0098,
+      "step": 4308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2361013199992727e-05,
+      "loss": 0.9145,
+      "step": 4309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.235778501126223e-05,
+      "loss": 0.9312,
+      "step": 4310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2354556562337803e-05,
+      "loss": 0.9217,
+      "step": 4311
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351327853575718e-05,
+      "loss": 0.7917,
+      "step": 4312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234809888533228e-05,
+      "loss": 1.1234,
+      "step": 4313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234486965796383e-05,
+      "loss": 0.7352,
+      "step": 4314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2341640171826718e-05,
+      "loss": 0.8172,
+      "step": 4315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338410427277342e-05,
+      "loss": 0.9618,
+      "step": 4316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233518042467212e-05,
+      "loss": 1.0786,
+      "step": 4317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2331950164367499e-05,
+      "loss": 1.0021,
+      "step": 4318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328719646719955e-05,
+      "loss": 1.0127,
+      "step": 4319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325488872085996e-05,
+      "loss": 1.0782,
+      "step": 4320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322257840822152e-05,
+      "loss": 1.0841,
+      "step": 4321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319026553284984e-05,
+      "loss": 1.0001,
+      "step": 4322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315795009831079e-05,
+      "loss": 1.0232,
+      "step": 4323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312563210817061e-05,
+      "loss": 1.0681,
+      "step": 4324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.230933115659957e-05,
+      "loss": 1.0406,
+      "step": 4325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306098847535285e-05,
+      "loss": 0.938,
+      "step": 4326
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302866283980905e-05,
+      "loss": 0.8231,
+      "step": 4327
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299633466293164e-05,
+      "loss": 0.971,
+      "step": 4328
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296400394828816e-05,
+      "loss": 1.0633,
+      "step": 4329
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293167069944644e-05,
+      "loss": 0.7089,
+      "step": 4330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289933491997477e-05,
+      "loss": 0.9833,
+      "step": 4331
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.228669966134414e-05,
+      "loss": 0.859,
+      "step": 4332
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283465578341514e-05,
+      "loss": 0.7916,
+      "step": 4333
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280231243346494e-05,
+      "loss": 1.3776,
+      "step": 4334
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2276996656716005e-05,
+      "loss": 1.0808,
+      "step": 4335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273761818807002e-05,
+      "loss": 0.8823,
+      "step": 4336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270526729976467e-05,
+      "loss": 0.9943,
+      "step": 4337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267291390581404e-05,
+      "loss": 0.8865,
+      "step": 4338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2264055800978856e-05,
+      "loss": 0.9328,
+      "step": 4339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2260819961525886e-05,
+      "loss": 0.7761,
+      "step": 4340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.225758387257958e-05,
+      "loss": 1.1586,
+      "step": 4341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2254347534497064e-05,
+      "loss": 0.7975,
+      "step": 4342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.225111094763548e-05,
+      "loss": 1.1984,
+      "step": 4343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224787411235201e-05,
+      "loss": 0.902,
+      "step": 4344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244637029003849e-05,
+      "loss": 0.9574,
+      "step": 4345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241399697948227e-05,
+      "loss": 0.683,
+      "step": 4346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22381621195424e-05,
+      "loss": 1.5603,
+      "step": 4347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234924294143656e-05,
+      "loss": 1.0586,
+      "step": 4348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22316862221093e-05,
+      "loss": 1.3151,
+      "step": 4349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228447903796677e-05,
+      "loss": 0.7169,
+      "step": 4350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 1.1025,
+      "step": 4351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221970529766103e-05,
+      "loss": 0.8952,
+      "step": 4352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218731474762967e-05,
+      "loss": 0.7826,
+      "step": 4353
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221549217491118e-05,
+      "loss": 1.0309,
+      "step": 4354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2212252630568226e-05,
+      "loss": 0.768,
+      "step": 4355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209012842091598e-05,
+      "loss": 1.0301,
+      "step": 4356
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205772809838825e-05,
+      "loss": 0.8276,
+      "step": 4357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202532534167464e-05,
+      "loss": 0.8426,
+      "step": 4358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199292015435095e-05,
+      "loss": 1.0383,
+      "step": 4359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196051253999322e-05,
+      "loss": 0.9013,
+      "step": 4360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2192810250217787e-05,
+      "loss": 0.8087,
+      "step": 4361
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189569004448147e-05,
+      "loss": 1.6362,
+      "step": 4362
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218632751704809e-05,
+      "loss": 0.9542,
+      "step": 4363
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183085788375337e-05,
+      "loss": 0.9578,
+      "step": 4364
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2179843818787625e-05,
+      "loss": 0.9841,
+      "step": 4365
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176601608642721e-05,
+      "loss": 0.746,
+      "step": 4366
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173359158298424e-05,
+      "loss": 1.2153,
+      "step": 4367
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217011646811255e-05,
+      "loss": 1.1085,
+      "step": 4368
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166873538442956e-05,
+      "loss": 1.1201,
+      "step": 4369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2163630369647508e-05,
+      "loss": 1.1123,
+      "step": 4370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2160386962084106e-05,
+      "loss": 1.182,
+      "step": 4371
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2157143316110685e-05,
+      "loss": 1.0993,
+      "step": 4372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2153899432085192e-05,
+      "loss": 0.7501,
+      "step": 4373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2150655310365605e-05,
+      "loss": 0.7265,
+      "step": 4374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2147410951309937e-05,
+      "loss": 1.1307,
+      "step": 4375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2144166355276216e-05,
+      "loss": 1.0168,
+      "step": 4376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2140921522622498e-05,
+      "loss": 0.727,
+      "step": 4377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137676453706871e-05,
+      "loss": 0.7904,
+      "step": 4378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134431148887442e-05,
+      "loss": 1.0921,
+      "step": 4379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131185608522351e-05,
+      "loss": 1.3551,
+      "step": 4380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127939832969759e-05,
+      "loss": 0.8703,
+      "step": 4381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124693822587854e-05,
+      "loss": 1.0792,
+      "step": 4382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212144757773485e-05,
+      "loss": 0.985,
+      "step": 4383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2118201098768986e-05,
+      "loss": 0.8387,
+      "step": 4384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211495438604853e-05,
+      "loss": 1.0947,
+      "step": 4385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111707439931776e-05,
+      "loss": 0.9082,
+      "step": 4386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108460260777038e-05,
+      "loss": 0.886,
+      "step": 4387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2105212848942656e-05,
+      "loss": 0.9726,
+      "step": 4388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2101965204787004e-05,
+      "loss": 1.3126,
+      "step": 4389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2098717328668478e-05,
+      "loss": 1.1804,
+      "step": 4390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095469220945493e-05,
+      "loss": 0.7073,
+      "step": 4391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092220881976498e-05,
+      "loss": 1.013,
+      "step": 4392
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088972312119965e-05,
+      "loss": 0.8027,
+      "step": 4393
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085723511734386e-05,
+      "loss": 0.875,
+      "step": 4394
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082474481178288e-05,
+      "loss": 0.9304,
+      "step": 4395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079225220810218e-05,
+      "loss": 1.0731,
+      "step": 4396
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2075975730988747e-05,
+      "loss": 0.9861,
+      "step": 4397
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072726012072473e-05,
+      "loss": 1.0126,
+      "step": 4398
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069476064420023e-05,
+      "loss": 0.9345,
+      "step": 4399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2066225888390042e-05,
+      "loss": 0.8929,
+      "step": 4400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062975484341204e-05,
+      "loss": 0.8475,
+      "step": 4401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205972485263221e-05,
+      "loss": 0.9494,
+      "step": 4402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056473993621784e-05,
+      "loss": 1.0374,
+      "step": 4403
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2053222907668673e-05,
+      "loss": 0.797,
+      "step": 4404
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2049971595131655e-05,
+      "loss": 1.1598,
+      "step": 4405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2046720056369527e-05,
+      "loss": 0.9172,
+      "step": 4406
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2043468291741117e-05,
+      "loss": 1.0481,
+      "step": 4407
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2040216301605267e-05,
+      "loss": 0.9128,
+      "step": 4408
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203696408632086e-05,
+      "loss": 0.7126,
+      "step": 4409
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2033711646246787e-05,
+      "loss": 0.9833,
+      "step": 4410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030458981741977e-05,
+      "loss": 0.8737,
+      "step": 4411
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027206093165376e-05,
+      "loss": 0.7579,
+      "step": 4412
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2023952980875956e-05,
+      "loss": 0.6545,
+      "step": 4413
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202069964523272e-05,
+      "loss": 1.1041,
+      "step": 4414
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017446086594686e-05,
+      "loss": 0.8043,
+      "step": 4415
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014192305320904e-05,
+      "loss": 1.1262,
+      "step": 4416
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010938301770447e-05,
+      "loss": 0.9578,
+      "step": 4417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2007684076302407e-05,
+      "loss": 1.1635,
+      "step": 4418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004429629275907e-05,
+      "loss": 1.157,
+      "step": 4419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001174961050096e-05,
+      "loss": 1.0914,
+      "step": 4420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997920071984133e-05,
+      "loss": 1.1657,
+      "step": 4421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994664962437221e-05,
+      "loss": 1.1732,
+      "step": 4422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199140963276858e-05,
+      "loss": 1.0264,
+      "step": 4423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988154083337446e-05,
+      "loss": 0.8718,
+      "step": 4424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198489831450309e-05,
+      "loss": 1.0049,
+      "step": 4425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981642326624802e-05,
+      "loss": 1.1465,
+      "step": 4426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19783861200619e-05,
+      "loss": 1.0089,
+      "step": 4427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197512969517372e-05,
+      "loss": 0.9488,
+      "step": 4428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971873052319622e-05,
+      "loss": 1.0643,
+      "step": 4429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968616191859007e-05,
+      "loss": 0.8833,
+      "step": 4430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965359114151271e-05,
+      "loss": 0.8079,
+      "step": 4431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1962101819555855e-05,
+      "loss": 1.0944,
+      "step": 4432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958844308432225e-05,
+      "loss": 0.9009,
+      "step": 4433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195558658113986e-05,
+      "loss": 1.0689,
+      "step": 4434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952328638038261e-05,
+      "loss": 0.812,
+      "step": 4435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949070479486968e-05,
+      "loss": 0.9799,
+      "step": 4436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1945812105845533e-05,
+      "loss": 1.0895,
+      "step": 4437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1942553517473533e-05,
+      "loss": 0.8374,
+      "step": 4438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1939294714730573e-05,
+      "loss": 0.8527,
+      "step": 4439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1936035697976277e-05,
+      "loss": 0.8025,
+      "step": 4440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1932776467570294e-05,
+      "loss": 1.0328,
+      "step": 4441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1929517023872299e-05,
+      "loss": 1.0249,
+      "step": 4442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926257367241987e-05,
+      "loss": 1.0292,
+      "step": 4443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1922997498039082e-05,
+      "loss": 1.3069,
+      "step": 4444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919737416623323e-05,
+      "loss": 0.9958,
+      "step": 4445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916477123354478e-05,
+      "loss": 1.0833,
+      "step": 4446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1913216618592337e-05,
+      "loss": 0.8164,
+      "step": 4447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190995590269672e-05,
+      "loss": 0.8384,
+      "step": 4448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906694976027455e-05,
+      "loss": 0.7843,
+      "step": 4449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903433838944407e-05,
+      "loss": 0.7944,
+      "step": 4450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900172491807462e-05,
+      "loss": 0.9515,
+      "step": 4451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189691093497652e-05,
+      "loss": 0.8158,
+      "step": 4452
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893649168811516e-05,
+      "loss": 1.1887,
+      "step": 4453
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890387193672401e-05,
+      "loss": 1.0552,
+      "step": 4454
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887125009919154e-05,
+      "loss": 1.1515,
+      "step": 4455
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188386261791177e-05,
+      "loss": 0.6262,
+      "step": 4456
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880600018010275e-05,
+      "loss": 0.6346,
+      "step": 4457
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877337210574712e-05,
+      "loss": 1.2626,
+      "step": 4458
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874074195965146e-05,
+      "loss": 0.9424,
+      "step": 4459
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187081097454167e-05,
+      "loss": 1.1565,
+      "step": 4460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867547546664401e-05,
+      "loss": 0.7148,
+      "step": 4461
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186428391269347e-05,
+      "loss": 0.9294,
+      "step": 4462
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186102007298904e-05,
+      "loss": 1.0865,
+      "step": 4463
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185775602791129e-05,
+      "loss": 0.8199,
+      "step": 4464
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185449177782043e-05,
+      "loss": 1.0132,
+      "step": 4465
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851227323076679e-05,
+      "loss": 0.8927,
+      "step": 4466
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847962664040293e-05,
+      "loss": 1.1333,
+      "step": 4467
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184469780107154e-05,
+      "loss": 1.0616,
+      "step": 4468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1841432734530718e-05,
+      "loss": 0.9054,
+      "step": 4469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1838167464778143e-05,
+      "loss": 0.9722,
+      "step": 4470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1834901992174155e-05,
+      "loss": 1.1059,
+      "step": 4471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1831636317079114e-05,
+      "loss": 1.1107,
+      "step": 4472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1828370439853408e-05,
+      "loss": 0.8656,
+      "step": 4473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1825104360857437e-05,
+      "loss": 0.7392,
+      "step": 4474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182183808045164e-05,
+      "loss": 0.8485,
+      "step": 4475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181857159899646e-05,
+      "loss": 1.058,
+      "step": 4476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1815304916852373e-05,
+      "loss": 1.3738,
+      "step": 4477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812038034379878e-05,
+      "loss": 0.7647,
+      "step": 4478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1808770951939483e-05,
+      "loss": 0.9528,
+      "step": 4479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1805503669891736e-05,
+      "loss": 1.1075,
+      "step": 4480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1802236188597199e-05,
+      "loss": 1.0509,
+      "step": 4481
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179896850841645e-05,
+      "loss": 0.9572,
+      "step": 4482
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17957006297101e-05,
+      "loss": 2.3675,
+      "step": 4483
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792432552838772e-05,
+      "loss": 0.6967,
+      "step": 4484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789164278163122e-05,
+      "loss": 0.6535,
+      "step": 4485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785895806043815e-05,
+      "loss": 0.9294,
+      "step": 4486
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782627136841544e-05,
+      "loss": 0.8834,
+      "step": 4487
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779358270917027e-05,
+      "loss": 1.0946,
+      "step": 4488
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1776089208631e-05,
+      "loss": 2.4875,
+      "step": 4489
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772819950344218e-05,
+      "loss": 1.0694,
+      "step": 4490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769550496417465e-05,
+      "loss": 1.06,
+      "step": 4491
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766280847211542e-05,
+      "loss": 1.1566,
+      "step": 4492
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1763011003087268e-05,
+      "loss": 1.0595,
+      "step": 4493
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759740964405491e-05,
+      "loss": 0.8051,
+      "step": 4494
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1756470731527077e-05,
+      "loss": 0.901,
+      "step": 4495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175320030481291e-05,
+      "loss": 0.9783,
+      "step": 4496
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749929684623904e-05,
+      "loss": 1.1286,
+      "step": 4497
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746658871320984e-05,
+      "loss": 1.0112,
+      "step": 4498
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743387865265104e-05,
+      "loss": 0.9016,
+      "step": 4499
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740116666817237e-05,
+      "loss": 1.1365,
+      "step": 4500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173684527633837e-05,
+      "loss": 0.8643,
+      "step": 4501
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1733573694189533e-05,
+      "loss": 0.9027,
+      "step": 4502
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1730301920731749e-05,
+      "loss": 1.2195,
+      "step": 4503
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1727029956326077e-05,
+      "loss": 0.8107,
+      "step": 4504
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1723757801333602e-05,
+      "loss": 1.0063,
+      "step": 4505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.172048545611542e-05,
+      "loss": 0.9912,
+      "step": 4506
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1717212921032645e-05,
+      "loss": 1.1791,
+      "step": 4507
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171394019644643e-05,
+      "loss": 1.0802,
+      "step": 4508
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710667282717933e-05,
+      "loss": 0.9665,
+      "step": 4509
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170739418020833e-05,
+      "loss": 0.9581,
+      "step": 4510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1704120889278834e-05,
+      "loss": 1.1319,
+      "step": 4511
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700847410290667e-05,
+      "loss": 0.7471,
+      "step": 4512
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697573743605074e-05,
+      "loss": 0.9462,
+      "step": 4513
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169429988958332e-05,
+      "loss": 1.0849,
+      "step": 4514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691025848586694e-05,
+      "loss": 0.9249,
+      "step": 4515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1687751620976506e-05,
+      "loss": 0.89,
+      "step": 4516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168447720711408e-05,
+      "loss": 1.4621,
+      "step": 4517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681202607360761e-05,
+      "loss": 0.9888,
+      "step": 4518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167792782207793e-05,
+      "loss": 1.1287,
+      "step": 4519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674652851626967e-05,
+      "loss": 0.9432,
+      "step": 4520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671377696369285e-05,
+      "loss": 0.8741,
+      "step": 4521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668102356666314e-05,
+      "loss": 0.7222,
+      "step": 4522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664826832879508e-05,
+      "loss": 0.7119,
+      "step": 4523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661551125370332e-05,
+      "loss": 0.9605,
+      "step": 4524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658275234500282e-05,
+      "loss": 1.0199,
+      "step": 4525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165499916063087e-05,
+      "loss": 0.9532,
+      "step": 4526
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1651722904123628e-05,
+      "loss": 1.0261,
+      "step": 4527
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1648446465340104e-05,
+      "loss": 1.0131,
+      "step": 4528
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645169844641877e-05,
+      "loss": 0.7903,
+      "step": 4529
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641893042390533e-05,
+      "loss": 0.9762,
+      "step": 4530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638616058947687e-05,
+      "loss": 0.8794,
+      "step": 4531
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1635338894674968e-05,
+      "loss": 0.983,
+      "step": 4532
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632061549934037e-05,
+      "loss": 0.9489,
+      "step": 4533
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1628784025086559e-05,
+      "loss": 0.8937,
+      "step": 4534
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1625506320494222e-05,
+      "loss": 0.9035,
+      "step": 4535
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.162222843651875e-05,
+      "loss": 0.8918,
+      "step": 4536
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1618950373521867e-05,
+      "loss": 1.038,
+      "step": 4537
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1615672131865321e-05,
+      "loss": 0.9896,
+      "step": 4538
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1612393711910892e-05,
+      "loss": 1.5898,
+      "step": 4539
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609115114020363e-05,
+      "loss": 0.8274,
+      "step": 4540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605836338555552e-05,
+      "loss": 1.2591,
+      "step": 4541
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602557385878282e-05,
+      "loss": 1.0197,
+      "step": 4542
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1599278256350406e-05,
+      "loss": 0.838,
+      "step": 4543
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1595998950333794e-05,
+      "loss": 0.936,
+      "step": 4544
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592719468190332e-05,
+      "loss": 0.9581,
+      "step": 4545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158943981028193e-05,
+      "loss": 1.1923,
+      "step": 4546
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1586159976970514e-05,
+      "loss": 0.8755,
+      "step": 4547
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582879968618033e-05,
+      "loss": 1.1357,
+      "step": 4548
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1579599785586449e-05,
+      "loss": 1.0231,
+      "step": 4549
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576319428237754e-05,
+      "loss": 0.81,
+      "step": 4550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573038896933943e-05,
+      "loss": 0.8143,
+      "step": 4551
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156975819203705e-05,
+      "loss": 1.0972,
+      "step": 4552
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1566477313909113e-05,
+      "loss": 1.139,
+      "step": 4553
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563196262912193e-05,
+      "loss": 0.9387,
+      "step": 4554
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1559915039408374e-05,
+      "loss": 1.0057,
+      "step": 4555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556633643759755e-05,
+      "loss": 1.0793,
+      "step": 4556
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553352076328457e-05,
+      "loss": 1.181,
+      "step": 4557
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550070337476613e-05,
+      "loss": 0.8679,
+      "step": 4558
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546788427566384e-05,
+      "loss": 1.085,
+      "step": 4559
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543506346959947e-05,
+      "loss": 0.9994,
+      "step": 4560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1540224096019495e-05,
+      "loss": 1.1944,
+      "step": 4561
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536941675107238e-05,
+      "loss": 0.9894,
+      "step": 4562
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533659084585415e-05,
+      "loss": 0.8744,
+      "step": 4563
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530376324816276e-05,
+      "loss": 0.8541,
+      "step": 4564
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527093396162088e-05,
+      "loss": 0.8019,
+      "step": 4565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523810298985138e-05,
+      "loss": 0.9641,
+      "step": 4566
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.152052703364774e-05,
+      "loss": 1.1936,
+      "step": 4567
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1517243600512209e-05,
+      "loss": 0.9694,
+      "step": 4568
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1513959999940898e-05,
+      "loss": 0.9883,
+      "step": 4569
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1510676232296168e-05,
+      "loss": 0.6517,
+      "step": 4570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1507392297940394e-05,
+      "loss": 0.7721,
+      "step": 4571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1504108197235983e-05,
+      "loss": 1.0384,
+      "step": 4572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1500823930545349e-05,
+      "loss": 0.9974,
+      "step": 4573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1497539498230927e-05,
+      "loss": 1.1004,
+      "step": 4574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1494254900655175e-05,
+      "loss": 0.7815,
+      "step": 4575
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149097013818056e-05,
+      "loss": 1.2215,
+      "step": 4576
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1487685211169582e-05,
+      "loss": 0.9931,
+      "step": 4577
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484400119984739e-05,
+      "loss": 0.9896,
+      "step": 4578
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481114864988564e-05,
+      "loss": 0.8823,
+      "step": 4579
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1477829446543603e-05,
+      "loss": 0.9627,
+      "step": 4580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147454386501242e-05,
+      "loss": 1.0182,
+      "step": 4581
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471258120757587e-05,
+      "loss": 1.0245,
+      "step": 4582
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467972214141712e-05,
+      "loss": 1.1018,
+      "step": 4583
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146468614552741e-05,
+      "loss": 1.0597,
+      "step": 4584
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1461399915277315e-05,
+      "loss": 0.9956,
+      "step": 4585
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458113523754082e-05,
+      "loss": 0.825,
+      "step": 4586
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1454826971320379e-05,
+      "loss": 0.8901,
+      "step": 4587
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451540258338894e-05,
+      "loss": 1.2059,
+      "step": 4588
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448253385172334e-05,
+      "loss": 1.0284,
+      "step": 4589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444966352183421e-05,
+      "loss": 1.0685,
+      "step": 4590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441679159734905e-05,
+      "loss": 1.271,
+      "step": 4591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1438391808189533e-05,
+      "loss": 1.1618,
+      "step": 4592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435104297910084e-05,
+      "loss": 0.9329,
+      "step": 4593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431816629259359e-05,
+      "loss": 0.9042,
+      "step": 4594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1428528802600164e-05,
+      "loss": 0.9693,
+      "step": 4595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1425240818295326e-05,
+      "loss": 0.9252,
+      "step": 4596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1421952676707697e-05,
+      "loss": 0.8889,
+      "step": 4597
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1418664378200138e-05,
+      "loss": 0.7863,
+      "step": 4598
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1415375923135526e-05,
+      "loss": 0.9568,
+      "step": 4599
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1412087311876765e-05,
+      "loss": 0.9345,
+      "step": 4600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.140879854478677e-05,
+      "loss": 1.0696,
+      "step": 4601
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1405509622228473e-05,
+      "loss": 1.0178,
+      "step": 4602
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.140222054456482e-05,
+      "loss": 1.0425,
+      "step": 4603
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.139893131215878e-05,
+      "loss": 0.9794,
+      "step": 4604
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1395641925373343e-05,
+      "loss": 0.7845,
+      "step": 4605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392352384571501e-05,
+      "loss": 0.966,
+      "step": 4606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389062690116276e-05,
+      "loss": 0.6458,
+      "step": 4607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385772842370706e-05,
+      "loss": 1.0562,
+      "step": 4608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382482841697838e-05,
+      "loss": 0.9627,
+      "step": 4609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137919268846074e-05,
+      "loss": 0.9013,
+      "step": 4610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1375902383022506e-05,
+      "loss": 0.8855,
+      "step": 4611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137261192574623e-05,
+      "loss": 0.8666,
+      "step": 4612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1369321316995034e-05,
+      "loss": 0.8785,
+      "step": 4613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366030557132057e-05,
+      "loss": 1.182,
+      "step": 4614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362739646520447e-05,
+      "loss": 0.9433,
+      "step": 4615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359448585523374e-05,
+      "loss": 0.8104,
+      "step": 4616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356157374504027e-05,
+      "loss": 0.9296,
+      "step": 4617
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1352866013825605e-05,
+      "loss": 1.1165,
+      "step": 4618
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349574503851327e-05,
+      "loss": 0.8575,
+      "step": 4619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346282844944433e-05,
+      "loss": 0.8621,
+      "step": 4620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342991037468165e-05,
+      "loss": 0.9029,
+      "step": 4621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339699081785801e-05,
+      "loss": 1.0271,
+      "step": 4622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336406978260621e-05,
+      "loss": 0.9954,
+      "step": 4623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333114727255929e-05,
+      "loss": 0.8577,
+      "step": 4624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329822329135043e-05,
+      "loss": 0.9006,
+      "step": 4625
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1326529784261287e-05,
+      "loss": 0.6738,
+      "step": 4626
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323237092998019e-05,
+      "loss": 1.2334,
+      "step": 4627
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319944255708603e-05,
+      "loss": 0.9721,
+      "step": 4628
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131665127275642e-05,
+      "loss": 0.7378,
+      "step": 4629
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313358144504872e-05,
+      "loss": 0.9075,
+      "step": 4630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1310064871317365e-05,
+      "loss": 0.9018,
+      "step": 4631
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1306771453557336e-05,
+      "loss": 1.0435,
+      "step": 4632
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1303477891588229e-05,
+      "loss": 1.0304,
+      "step": 4633
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1300184185773504e-05,
+      "loss": 0.8984,
+      "step": 4634
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.129689033647664e-05,
+      "loss": 1.0484,
+      "step": 4635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1293596344061131e-05,
+      "loss": 0.955,
+      "step": 4636
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1290302208890483e-05,
+      "loss": 0.8184,
+      "step": 4637
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1287007931328226e-05,
+      "loss": 0.8969,
+      "step": 4638
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283713511737902e-05,
+      "loss": 1.0541,
+      "step": 4639
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280418950483059e-05,
+      "loss": 0.9988,
+      "step": 4640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277124247927275e-05,
+      "loss": 0.9968,
+      "step": 4641
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1273829404434137e-05,
+      "loss": 0.8451,
+      "step": 4642
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1270534420367252e-05,
+      "loss": 0.7854,
+      "step": 4643
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126723929609023e-05,
+      "loss": 1.0984,
+      "step": 4644
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263944031966714e-05,
+      "loss": 1.0132,
+      "step": 4645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260648628360346e-05,
+      "loss": 0.9613,
+      "step": 4646
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1257353085634799e-05,
+      "loss": 1.2104,
+      "step": 4647
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1254057404153745e-05,
+      "loss": 1.1887,
+      "step": 4648
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250761584280887e-05,
+      "loss": 1.4398,
+      "step": 4649
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1247465626379934e-05,
+      "loss": 1.2696,
+      "step": 4650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244169530814606e-05,
+      "loss": 1.1084,
+      "step": 4651
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240873297948654e-05,
+      "loss": 1.1145,
+      "step": 4652
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123757692814583e-05,
+      "loss": 0.8835,
+      "step": 4653
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234280421769906e-05,
+      "loss": 1.0705,
+      "step": 4654
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230983779184666e-05,
+      "loss": 0.7295,
+      "step": 4655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227687000753919e-05,
+      "loss": 0.9698,
+      "step": 4656
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224390086841472e-05,
+      "loss": 0.9497,
+      "step": 4657
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221093037811164e-05,
+      "loss": 1.0842,
+      "step": 4658
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121779585402684e-05,
+      "loss": 0.9629,
+      "step": 4659
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1214498535852364e-05,
+      "loss": 1.0439,
+      "step": 4660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211201083651603e-05,
+      "loss": 0.8466,
+      "step": 4661
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207903497788457e-05,
+      "loss": 0.7938,
+      "step": 4662
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204605778626834e-05,
+      "loss": 1.1601,
+      "step": 4663
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1201307926530647e-05,
+      "loss": 1.2977,
+      "step": 4664
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1198009941863832e-05,
+      "loss": 0.9331,
+      "step": 4665
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1194711824990344e-05,
+      "loss": 0.9302,
+      "step": 4666
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1191413576274146e-05,
+      "loss": 0.8541,
+      "step": 4667
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118811519607921e-05,
+      "loss": 1.1595,
+      "step": 4668
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118481668476954e-05,
+      "loss": 1.1588,
+      "step": 4669
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1181518042709142e-05,
+      "loss": 0.8141,
+      "step": 4670
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178219270262028e-05,
+      "loss": 0.8669,
+      "step": 4671
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174920367792246e-05,
+      "loss": 1.003,
+      "step": 4672
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1171621335663845e-05,
+      "loss": 0.9867,
+      "step": 4673
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168322174240889e-05,
+      "loss": 0.8432,
+      "step": 4674
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165022883887457e-05,
+      "loss": 0.8466,
+      "step": 4675
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161723464967643e-05,
+      "loss": 0.9375,
+      "step": 4676
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1158423917845558e-05,
+      "loss": 0.703,
+      "step": 4677
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155124242885322e-05,
+      "loss": 0.9371,
+      "step": 4678
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1151824440451069e-05,
+      "loss": 1.069,
+      "step": 4679
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148524510906957e-05,
+      "loss": 1.0764,
+      "step": 4680
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1145224454617145e-05,
+      "loss": 1.0872,
+      "step": 4681
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114192427194581e-05,
+      "loss": 1.2699,
+      "step": 4682
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138623963257148e-05,
+      "loss": 1.0027,
+      "step": 4683
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1135323528915363e-05,
+      "loss": 0.7768,
+      "step": 4684
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1132022969284679e-05,
+      "loss": 1.3011,
+      "step": 4685
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128722284729325e-05,
+      "loss": 0.6925,
+      "step": 4686
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125421475613553e-05,
+      "loss": 1.1287,
+      "step": 4687
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1122120542301622e-05,
+      "loss": 1.1452,
+      "step": 4688
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111881948515781e-05,
+      "loss": 1.1364,
+      "step": 4689
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115518304546402e-05,
+      "loss": 1.0064,
+      "step": 4690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112217000831705e-05,
+      "loss": 0.9209,
+      "step": 4691
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110891557437803e-05,
+      "loss": 1.0919,
+      "step": 4692
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1105614025549711e-05,
+      "loss": 0.9084,
+      "step": 4693
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102312354711091e-05,
+      "loss": 1.1164,
+      "step": 4694
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099010562226524e-05,
+      "loss": 0.8615,
+      "step": 4695
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1095708648460379e-05,
+      "loss": 0.877,
+      "step": 4696
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1092406613777047e-05,
+      "loss": 1.0184,
+      "step": 4697
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1089104458540916e-05,
+      "loss": 1.057,
+      "step": 4698
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1085802183116402e-05,
+      "loss": 1.1232,
+      "step": 4699
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1082499787867926e-05,
+      "loss": 0.9908,
+      "step": 4700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1079197273159925e-05,
+      "loss": 1.5388,
+      "step": 4701
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107589463935685e-05,
+      "loss": 1.0668,
+      "step": 4702
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072591886823164e-05,
+      "loss": 0.7361,
+      "step": 4703
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106928901592334e-05,
+      "loss": 1.1364,
+      "step": 4704
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106598602702187e-05,
+      "loss": 1.0996,
+      "step": 4705
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062682920483252e-05,
+      "loss": 1.091,
+      "step": 4706
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059379696672008e-05,
+      "loss": 0.7735,
+      "step": 4707
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105607635595266e-05,
+      "loss": 1.0883,
+      "step": 4708
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105277289868975e-05,
+      "loss": 1.1093,
+      "step": 4709
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049469325247834e-05,
+      "loss": 1.4689,
+      "step": 4710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1046165635991482e-05,
+      "loss": 0.9519,
+      "step": 4711
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042861831285265e-05,
+      "loss": 0.826,
+      "step": 4712
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1039557911493778e-05,
+      "loss": 1.0001,
+      "step": 4713
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036253876981633e-05,
+      "loss": 1.1681,
+      "step": 4714
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1032949728113438e-05,
+      "loss": 0.9964,
+      "step": 4715
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029645465253827e-05,
+      "loss": 0.8896,
+      "step": 4716
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1026341088767442e-05,
+      "loss": 0.9813,
+      "step": 4717
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023036599018938e-05,
+      "loss": 0.9438,
+      "step": 4718
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019731996372986e-05,
+      "loss": 0.9867,
+      "step": 4719
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016427281194265e-05,
+      "loss": 1.1249,
+      "step": 4720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1013122453847465e-05,
+      "loss": 1.0183,
+      "step": 4721
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009817514697293e-05,
+      "loss": 1.2329,
+      "step": 4722
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.100651246410846e-05,
+      "loss": 0.9593,
+      "step": 4723
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1003207302445712e-05,
+      "loss": 1.0665,
+      "step": 4724
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0999902030073773e-05,
+      "loss": 0.9591,
+      "step": 4725
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996596647357405e-05,
+      "loss": 1.0152,
+      "step": 4726
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0993291154661376e-05,
+      "loss": 0.7541,
+      "step": 4727
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989985552350463e-05,
+      "loss": 1.0544,
+      "step": 4728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.098667984078945e-05,
+      "loss": 0.7064,
+      "step": 4729
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.098337402034315e-05,
+      "loss": 1.0512,
+      "step": 4730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0980068091376375e-05,
+      "loss": 1.206,
+      "step": 4731
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0976762054253948e-05,
+      "loss": 0.9317,
+      "step": 4732
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0973455909340709e-05,
+      "loss": 1.0535,
+      "step": 4733
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0970149657001508e-05,
+      "loss": 0.8092,
+      "step": 4734
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966843297601208e-05,
+      "loss": 1.0583,
+      "step": 4735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0963536831504685e-05,
+      "loss": 0.9388,
+      "step": 4736
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960230259076819e-05,
+      "loss": 0.9977,
+      "step": 4737
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956923580682515e-05,
+      "loss": 1.1196,
+      "step": 4738
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0953616796686676e-05,
+      "loss": 0.8002,
+      "step": 4739
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950309907454224e-05,
+      "loss": 0.9721,
+      "step": 4740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947002913350095e-05,
+      "loss": 1.1079,
+      "step": 4741
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943695814739236e-05,
+      "loss": 0.8059,
+      "step": 4742
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940388611986592e-05,
+      "loss": 1.1437,
+      "step": 4743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937081305457137e-05,
+      "loss": 1.0838,
+      "step": 4744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0933773895515854e-05,
+      "loss": 1.1419,
+      "step": 4745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930466382527721e-05,
+      "loss": 0.6718,
+      "step": 4746
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927158766857748e-05,
+      "loss": 0.7341,
+      "step": 4747
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923851048870948e-05,
+      "loss": 0.9867,
+      "step": 4748
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092054322893234e-05,
+      "loss": 1.2122,
+      "step": 4749
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917235307406966e-05,
+      "loss": 1.038,
+      "step": 4750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913927284659865e-05,
+      "loss": 0.9952,
+      "step": 4751
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0910619161056102e-05,
+      "loss": 0.6576,
+      "step": 4752
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090731093696074e-05,
+      "loss": 0.9446,
+      "step": 4753
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0904002612738858e-05,
+      "loss": 1.1278,
+      "step": 4754
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0900694188755557e-05,
+      "loss": 0.9684,
+      "step": 4755
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897385665375926e-05,
+      "loss": 0.8864,
+      "step": 4756
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894077042965084e-05,
+      "loss": 0.9068,
+      "step": 4757
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0890768321888154e-05,
+      "loss": 1.3466,
+      "step": 4758
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887459502510273e-05,
+      "loss": 0.9552,
+      "step": 4759
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884150585196584e-05,
+      "loss": 0.8481,
+      "step": 4760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0880841570312247e-05,
+      "loss": 1.0736,
+      "step": 4761
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0877532458222423e-05,
+      "loss": 0.8293,
+      "step": 4762
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0874223249292296e-05,
+      "loss": 1.2591,
+      "step": 4763
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.087091394388705e-05,
+      "loss": 1.18,
+      "step": 4764
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0867604542371883e-05,
+      "loss": 1.1448,
+      "step": 4765
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0864295045112013e-05,
+      "loss": 0.9371,
+      "step": 4766
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860985452472653e-05,
+      "loss": 1.0332,
+      "step": 4767
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857675764819031e-05,
+      "loss": 0.9852,
+      "step": 4768
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854365982516402e-05,
+      "loss": 1.0805,
+      "step": 4769
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851056105930003e-05,
+      "loss": 1.2591,
+      "step": 4770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0847746135425103e-05,
+      "loss": 1.1397,
+      "step": 4771
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844436071366977e-05,
+      "loss": 1.3144,
+      "step": 4772
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841125914120903e-05,
+      "loss": 0.6863,
+      "step": 4773
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837815664052173e-05,
+      "loss": 0.7746,
+      "step": 4774
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834505321526098e-05,
+      "loss": 1.0478,
+      "step": 4775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831194886907985e-05,
+      "loss": 0.9628,
+      "step": 4776
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827884360563164e-05,
+      "loss": 0.9358,
+      "step": 4777
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082457374285696e-05,
+      "loss": 1.1801,
+      "step": 4778
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0821263034154723e-05,
+      "loss": 1.1534,
+      "step": 4779
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817952234821806e-05,
+      "loss": 1.271,
+      "step": 4780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0814641345223575e-05,
+      "loss": 0.7939,
+      "step": 4781
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.08113303657254e-05,
+      "loss": 1.1777,
+      "step": 4782
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0808019296692669e-05,
+      "loss": 0.9806,
+      "step": 4783
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804708138490771e-05,
+      "loss": 1.1218,
+      "step": 4784
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801396891485115e-05,
+      "loss": 1.0111,
+      "step": 4785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798085556041114e-05,
+      "loss": 1.0589,
+      "step": 4786
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0794774132524182e-05,
+      "loss": 0.984,
+      "step": 4787
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0791462621299764e-05,
+      "loss": 0.9386,
+      "step": 4788
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0788151022733297e-05,
+      "loss": 0.9884,
+      "step": 4789
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784839337190234e-05,
+      "loss": 0.9645,
+      "step": 4790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0781527565036037e-05,
+      "loss": 0.825,
+      "step": 4791
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778215706636176e-05,
+      "loss": 0.9144,
+      "step": 4792
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774903762356137e-05,
+      "loss": 0.9909,
+      "step": 4793
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0771591732561404e-05,
+      "loss": 1.0179,
+      "step": 4794
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0768279617617476e-05,
+      "loss": 0.9474,
+      "step": 4795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0764967417889872e-05,
+      "loss": 0.8595,
+      "step": 4796
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0761655133744103e-05,
+      "loss": 0.8272,
+      "step": 4797
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0758342765545693e-05,
+      "loss": 0.7601,
+      "step": 4798
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0755030313660188e-05,
+      "loss": 0.8301,
+      "step": 4799
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0751717778453132e-05,
+      "loss": 0.8519,
+      "step": 4800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0748405160290076e-05,
+      "loss": 0.6404,
+      "step": 4801
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074509245953659e-05,
+      "loss": 0.9431,
+      "step": 4802
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741779676558251e-05,
+      "loss": 1.0773,
+      "step": 4803
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0738466811720629e-05,
+      "loss": 0.9105,
+      "step": 4804
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735153865389327e-05,
+      "loss": 1.0175,
+      "step": 4805
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731840837929946e-05,
+      "loss": 1.0053,
+      "step": 4806
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0728527729708089e-05,
+      "loss": 0.8367,
+      "step": 4807
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0725214541089382e-05,
+      "loss": 1.0591,
+      "step": 4808
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721901272439446e-05,
+      "loss": 0.7722,
+      "step": 4809
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0718587924123925e-05,
+      "loss": 0.8377,
+      "step": 4810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0715274496508458e-05,
+      "loss": 1.1026,
+      "step": 4811
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.07119609899587e-05,
+      "loss": 0.9574,
+      "step": 4812
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070864740484032e-05,
+      "loss": 1.4197,
+      "step": 4813
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705333741518984e-05,
+      "loss": 0.8982,
+      "step": 4814
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702020000360368e-05,
+      "loss": 1.0778,
+      "step": 4815
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069870618173017e-05,
+      "loss": 0.9717,
+      "step": 4816
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695392285994085e-05,
+      "loss": 0.8432,
+      "step": 4817
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692078313517813e-05,
+      "loss": 0.9164,
+      "step": 4818
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688764264667072e-05,
+      "loss": 0.7605,
+      "step": 4819
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685450139807585e-05,
+      "loss": 0.7751,
+      "step": 4820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0682135939305086e-05,
+      "loss": 0.8005,
+      "step": 4821
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678821663525307e-05,
+      "loss": 0.7154,
+      "step": 4822
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675507312834001e-05,
+      "loss": 0.8917,
+      "step": 4823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672192887596924e-05,
+      "loss": 0.9929,
+      "step": 4824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668878388179838e-05,
+      "loss": 1.1359,
+      "step": 4825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665563814948512e-05,
+      "loss": 1.0123,
+      "step": 4826
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0662249168268736e-05,
+      "loss": 0.897,
+      "step": 4827
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0658934448506295e-05,
+      "loss": 0.9549,
+      "step": 4828
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0655619656026979e-05,
+      "loss": 0.9434,
+      "step": 4829
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0652304791196599e-05,
+      "loss": 0.8752,
+      "step": 4830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0648989854380969e-05,
+      "loss": 0.9067,
+      "step": 4831
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0645674845945905e-05,
+      "loss": 1.0842,
+      "step": 4832
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064235976625724e-05,
+      "loss": 1.0616,
+      "step": 4833
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0639044615680809e-05,
+      "loss": 0.763,
+      "step": 4834
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635729394582453e-05,
+      "loss": 0.8651,
+      "step": 4835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0632414103328028e-05,
+      "loss": 0.8245,
+      "step": 4836
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062909874228339e-05,
+      "loss": 0.9909,
+      "step": 4837
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0625783311814411e-05,
+      "loss": 0.8569,
+      "step": 4838
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622467812286963e-05,
+      "loss": 0.9224,
+      "step": 4839
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061915224406693e-05,
+      "loss": 0.872,
+      "step": 4840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615836607520203e-05,
+      "loss": 0.9464,
+      "step": 4841
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612520903012677e-05,
+      "loss": 1.0303,
+      "step": 4842
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0609205130910257e-05,
+      "loss": 1.1219,
+      "step": 4843
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605889291578864e-05,
+      "loss": 1.0787,
+      "step": 4844
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602573385384404e-05,
+      "loss": 1.124,
+      "step": 4845
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599257412692818e-05,
+      "loss": 1.0363,
+      "step": 4846
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595941373870038e-05,
+      "loss": 1.0307,
+      "step": 4847
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592625269282e-05,
+      "loss": 0.9088,
+      "step": 4848
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058930909929466e-05,
+      "loss": 1.0646,
+      "step": 4849
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058599286427397e-05,
+      "loss": 1.0699,
+      "step": 4850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582676564585898e-05,
+      "loss": 1.1847,
+      "step": 4851
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579360200596414e-05,
+      "loss": 1.1256,
+      "step": 4852
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0576043772671496e-05,
+      "loss": 0.9955,
+      "step": 4853
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572727281177127e-05,
+      "loss": 0.7508,
+      "step": 4854
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569410726479301e-05,
+      "loss": 0.7779,
+      "step": 4855
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566094108944016e-05,
+      "loss": 1.4554,
+      "step": 4856
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562777428937278e-05,
+      "loss": 0.7101,
+      "step": 4857
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0559460686825107e-05,
+      "loss": 0.7556,
+      "step": 4858
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0556143882973514e-05,
+      "loss": 1.0041,
+      "step": 4859
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0552827017748526e-05,
+      "loss": 1.1672,
+      "step": 4860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0549510091516184e-05,
+      "loss": 0.8395,
+      "step": 4861
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.054619310464252e-05,
+      "loss": 1.3269,
+      "step": 4862
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0542876057493586e-05,
+      "loss": 0.7696,
+      "step": 4863
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0539558950435433e-05,
+      "loss": 0.9798,
+      "step": 4864
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0536241783834123e-05,
+      "loss": 1.1914,
+      "step": 4865
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532924558055722e-05,
+      "loss": 1.2323,
+      "step": 4866
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529607273466302e-05,
+      "loss": 1.333,
+      "step": 4867
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526289930431943e-05,
+      "loss": 1.025,
+      "step": 4868
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0522972529318733e-05,
+      "loss": 1.1325,
+      "step": 4869
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0519655070492762e-05,
+      "loss": 0.9433,
+      "step": 4870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516337554320135e-05,
+      "loss": 0.9906,
+      "step": 4871
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051301998116695e-05,
+      "loss": 1.2548,
+      "step": 4872
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0509702351399319e-05,
+      "loss": 0.8991,
+      "step": 4873
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506384665383365e-05,
+      "loss": 0.9628,
+      "step": 4874
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050306692348521e-05,
+      "loss": 0.8698,
+      "step": 4875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0499749126070979e-05,
+      "loss": 0.9098,
+      "step": 4876
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496431273506815e-05,
+      "loss": 1.2373,
+      "step": 4877
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049311336615886e-05,
+      "loss": 1.1808,
+      "step": 4878
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489795404393259e-05,
+      "loss": 0.7791,
+      "step": 4879
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048647738857617e-05,
+      "loss": 1.0938,
+      "step": 4880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048315931907375e-05,
+      "loss": 0.8424,
+      "step": 4881
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479841196252167e-05,
+      "loss": 1.0848,
+      "step": 4882
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476523020477594e-05,
+      "loss": 1.306,
+      "step": 4883
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0473204792116206e-05,
+      "loss": 0.9806,
+      "step": 4884
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469886511534193e-05,
+      "loss": 0.8259,
+      "step": 4885
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466568179097739e-05,
+      "loss": 0.9445,
+      "step": 4886
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463249795173039e-05,
+      "loss": 1.1161,
+      "step": 4887
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459931360126296e-05,
+      "loss": 0.9071,
+      "step": 4888
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0456612874323722e-05,
+      "loss": 1.0275,
+      "step": 4889
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045329433813152e-05,
+      "loss": 0.9575,
+      "step": 4890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449975751915913e-05,
+      "loss": 0.86,
+      "step": 4891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0446657116043124e-05,
+      "loss": 0.9913,
+      "step": 4892
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0443338430879381e-05,
+      "loss": 1.0188,
+      "step": 4893
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0440019696790919e-05,
+      "loss": 1.1546,
+      "step": 4894
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0436700914143975e-05,
+      "loss": 0.9882,
+      "step": 4895
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0433382083304799e-05,
+      "loss": 1.4152,
+      "step": 4896
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.043006320463964e-05,
+      "loss": 1.2159,
+      "step": 4897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0426744278514748e-05,
+      "loss": 0.9961,
+      "step": 4898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423425305296393e-05,
+      "loss": 0.8684,
+      "step": 4899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0420106285350834e-05,
+      "loss": 1.3923,
+      "step": 4900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0416787219044344e-05,
+      "loss": 1.0552,
+      "step": 4901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413468106743202e-05,
+      "loss": 1.062,
+      "step": 4902
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0410148948813687e-05,
+      "loss": 0.9102,
+      "step": 4903
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406829745622085e-05,
+      "loss": 1.1074,
+      "step": 4904
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0403510497534689e-05,
+      "loss": 1.334,
+      "step": 4905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400191204917793e-05,
+      "loss": 0.9954,
+      "step": 4906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396871868137704e-05,
+      "loss": 1.1624,
+      "step": 4907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0393552487560721e-05,
+      "loss": 0.761,
+      "step": 4908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0390233063553161e-05,
+      "loss": 1.1464,
+      "step": 4909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0386913596481336e-05,
+      "loss": 1.3016,
+      "step": 4910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383594086711568e-05,
+      "loss": 1.1168,
+      "step": 4911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380274534610181e-05,
+      "loss": 1.0319,
+      "step": 4912
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0376954940543506e-05,
+      "loss": 0.8828,
+      "step": 4913
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037363530487788e-05,
+      "loss": 1.6235,
+      "step": 4914
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370315627979636e-05,
+      "loss": 0.8786,
+      "step": 4915
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366995910215128e-05,
+      "loss": 1.0702,
+      "step": 4916
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036367615195069e-05,
+      "loss": 0.7742,
+      "step": 4917
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0360356353552687e-05,
+      "loss": 0.9138,
+      "step": 4918
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0357036515387476e-05,
+      "loss": 1.09,
+      "step": 4919
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0353716637821407e-05,
+      "loss": 1.0874,
+      "step": 4920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350396721220854e-05,
+      "loss": 0.8146,
+      "step": 4921
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347076765952191e-05,
+      "loss": 0.7727,
+      "step": 4922
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0343756772381786e-05,
+      "loss": 0.9072,
+      "step": 4923
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0340436740876016e-05,
+      "loss": 1.2744,
+      "step": 4924
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0337116671801273e-05,
+      "loss": 0.8416,
+      "step": 4925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0333796565523934e-05,
+      "loss": 1.2922,
+      "step": 4926
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0330476422410397e-05,
+      "loss": 0.7864,
+      "step": 4927
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0327156242827054e-05,
+      "loss": 0.9668,
+      "step": 4928
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0323836027140306e-05,
+      "loss": 0.9501,
+      "step": 4929
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320515775716556e-05,
+      "loss": 0.9201,
+      "step": 4930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317195488922204e-05,
+      "loss": 1.1026,
+      "step": 4931
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0313875167123673e-05,
+      "loss": 0.9965,
+      "step": 4932
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310554810687375e-05,
+      "loss": 0.9107,
+      "step": 4933
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030723441997972e-05,
+      "loss": 1.0649,
+      "step": 4934
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303913995367138e-05,
+      "loss": 1.1146,
+      "step": 4935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0300593537216054e-05,
+      "loss": 1.3065,
+      "step": 4936
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297273045892899e-05,
+      "loss": 1.1867,
+      "step": 4937
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293952521764105e-05,
+      "loss": 1.1634,
+      "step": 4938
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029063196519611e-05,
+      "loss": 0.9414,
+      "step": 4939
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287311376555353e-05,
+      "loss": 1.0708,
+      "step": 4940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0283990756208282e-05,
+      "loss": 0.9773,
+      "step": 4941
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280670104521339e-05,
+      "loss": 1.0128,
+      "step": 4942
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277349421860985e-05,
+      "loss": 0.9747,
+      "step": 4943
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0274028708593664e-05,
+      "loss": 2.2388,
+      "step": 4944
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027070796508584e-05,
+      "loss": 1.0783,
+      "step": 4945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0267387191703973e-05,
+      "loss": 1.4603,
+      "step": 4946
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026406638881453e-05,
+      "loss": 0.8705,
+      "step": 4947
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260745556783973e-05,
+      "loss": 0.7753,
+      "step": 4948
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257424695978778e-05,
+      "loss": 0.9615,
+      "step": 4949
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254103806765425e-05,
+      "loss": 0.6761,
+      "step": 4950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250782889510378e-05,
+      "loss": 0.9571,
+      "step": 4951
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247461944580128e-05,
+      "loss": 0.9096,
+      "step": 4952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244140972341155e-05,
+      "loss": 1.0326,
+      "step": 4953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240819973159945e-05,
+      "loss": 0.6187,
+      "step": 4954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237498947402992e-05,
+      "loss": 1.0815,
+      "step": 4955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234177895436779e-05,
+      "loss": 0.9185,
+      "step": 4956
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0230856817627814e-05,
+      "loss": 0.813,
+      "step": 4957
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0227535714342589e-05,
+      "loss": 0.9073,
+      "step": 4958
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.02242145859476e-05,
+      "loss": 1.065,
+      "step": 4959
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0220893432809365e-05,
+      "loss": 1.073,
+      "step": 4960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0217572255294377e-05,
+      "loss": 1.0147,
+      "step": 4961
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0214251053769149e-05,
+      "loss": 1.1032,
+      "step": 4962
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0210929828600196e-05,
+      "loss": 0.7311,
+      "step": 4963
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207608580154032e-05,
+      "loss": 1.3116,
+      "step": 4964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020428730879717e-05,
+      "loss": 0.9237,
+      "step": 4965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0200966014896135e-05,
+      "loss": 0.9476,
+      "step": 4966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197644698817446e-05,
+      "loss": 1.1144,
+      "step": 4967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019432336092763e-05,
+      "loss": 0.9821,
+      "step": 4968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191002001593211e-05,
+      "loss": 1.0399,
+      "step": 4969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187680621180722e-05,
+      "loss": 1.2324,
+      "step": 4970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0184359220056693e-05,
+      "loss": 1.0708,
+      "step": 4971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181037798587656e-05,
+      "loss": 0.9152,
+      "step": 4972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0177716357140147e-05,
+      "loss": 0.9831,
+      "step": 4973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0174394896080713e-05,
+      "loss": 1.0606,
+      "step": 4974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171073415775884e-05,
+      "loss": 0.9734,
+      "step": 4975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167751916592206e-05,
+      "loss": 1.0343,
+      "step": 4976
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0164430398896229e-05,
+      "loss": 1.1551,
+      "step": 4977
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161108863054493e-05,
+      "loss": 1.0011,
+      "step": 4978
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157787309433549e-05,
+      "loss": 0.8958,
+      "step": 4979
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015446573839995e-05,
+      "loss": 1.3323,
+      "step": 4980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0151144150320247e-05,
+      "loss": 0.9824,
+      "step": 4981
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147822545560992e-05,
+      "loss": 0.9303,
+      "step": 4982
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014450092448875e-05,
+      "loss": 0.7139,
+      "step": 4983
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0141179287470068e-05,
+      "loss": 0.8015,
+      "step": 4984
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137857634871515e-05,
+      "loss": 0.8237,
+      "step": 4985
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134535967059649e-05,
+      "loss": 1.1532,
+      "step": 4986
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0131214284401031e-05,
+      "loss": 1.1621,
+      "step": 4987
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127892587262233e-05,
+      "loss": 1.151,
+      "step": 4988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0124570876009818e-05,
+      "loss": 0.9922,
+      "step": 4989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.012124915101035e-05,
+      "loss": 0.6279,
+      "step": 4990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0117927412630409e-05,
+      "loss": 0.8845,
+      "step": 4991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0114605661236552e-05,
+      "loss": 0.8794,
+      "step": 4992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0111283897195363e-05,
+      "loss": 1.0294,
+      "step": 4993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0107962120873414e-05,
+      "loss": 1.0574,
+      "step": 4994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010464033263728e-05,
+      "loss": 1.1216,
+      "step": 4995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101318532853535e-05,
+      "loss": 0.9504,
+      "step": 4996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097996721888759e-05,
+      "loss": 0.9278,
+      "step": 4997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094674900109532e-05,
+      "loss": 0.9247,
+      "step": 4998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091353067882433e-05,
+      "loss": 0.9408,
+      "step": 4999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088031225574044e-05,
+      "loss": 0.8386,
+      "step": 5000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084709373550947e-05,
+      "loss": 1.1627,
+      "step": 5001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008138751217973e-05,
+      "loss": 1.3667,
+      "step": 5002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078065641826968e-05,
+      "loss": 1.0113,
+      "step": 5003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074743762859258e-05,
+      "loss": 0.823,
+      "step": 5004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071421875643183e-05,
+      "loss": 0.8925,
+      "step": 5005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068099980545328e-05,
+      "loss": 0.8034,
+      "step": 5006
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064778077932283e-05,
+      "loss": 0.9807,
+      "step": 5007
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006145616817064e-05,
+      "loss": 0.9977,
+      "step": 5008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058134251626982e-05,
+      "loss": 0.9436,
+      "step": 5009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005481232866791e-05,
+      "loss": 0.8229,
+      "step": 5010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051490399660007e-05,
+      "loss": 1.1463,
+      "step": 5011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004816846496987e-05,
+      "loss": 0.923,
+      "step": 5012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044846524964088e-05,
+      "loss": 0.9225,
+      "step": 5013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004152458000926e-05,
+      "loss": 1.6149,
+      "step": 5014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038202630471976e-05,
+      "loss": 1.0369,
+      "step": 5015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034880676718833e-05,
+      "loss": 1.0979,
+      "step": 5016
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003155871911642e-05,
+      "loss": 1.0746,
+      "step": 5017
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002823675803134e-05,
+      "loss": 1.0782,
+      "step": 5018
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0024914793830187e-05,
+      "loss": 1.0469,
+      "step": 5019
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021592826879552e-05,
+      "loss": 0.796,
+      "step": 5020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0018270857546037e-05,
+      "loss": 0.9452,
+      "step": 5021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0014948886196242e-05,
+      "loss": 1.0314,
+      "step": 5022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0011626913196751e-05,
+      "loss": 0.8606,
+      "step": 5023
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0008304938914173e-05,
+      "loss": 1.095,
+      "step": 5024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0004982963715103e-05,
+      "loss": 0.7337,
+      "step": 5025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0001660987966135e-05,
+      "loss": 1.034,
+      "step": 5026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998339012033867e-06,
+      "loss": 0.909,
+      "step": 5027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9950170362849e-06,
+      "loss": 1.2543,
+      "step": 5028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99169506108583e-06,
+      "loss": 0.9363,
+      "step": 5029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98837308680325e-06,
+      "loss": 0.9493,
+      "step": 5030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985051113803763e-06,
+      "loss": 1.108,
+      "step": 5031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.981729142453964e-06,
+      "loss": 1.0486,
+      "step": 5032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97840717312045e-06,
+      "loss": 0.7558,
+      "step": 5033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975085206169816e-06,
+      "loss": 1.0216,
+      "step": 5034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971763241968662e-06,
+      "loss": 1.0021,
+      "step": 5035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968441280883584e-06,
+      "loss": 0.9665,
+      "step": 5036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965119323281174e-06,
+      "loss": 0.9079,
+      "step": 5037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961797369528027e-06,
+      "loss": 1.0889,
+      "step": 5038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958475419990743e-06,
+      "loss": 0.9143,
+      "step": 5039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955153475035913e-06,
+      "loss": 1.2848,
+      "step": 5040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951831535030133e-06,
+      "loss": 1.147,
+      "step": 5041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.948509600339996e-06,
+      "loss": 0.943,
+      "step": 5042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945187671332095e-06,
+      "loss": 0.8922,
+      "step": 5043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941865748373018e-06,
+      "loss": 0.78,
+      "step": 5044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.938543831829362e-06,
+      "loss": 1.4574,
+      "step": 5045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935221922067718e-06,
+      "loss": 1.3351,
+      "step": 5046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.931900019454675e-06,
+      "loss": 0.935,
+      "step": 5047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92857812435682e-06,
+      "loss": 1.0208,
+      "step": 5048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925256237140742e-06,
+      "loss": 1.0029,
+      "step": 5049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92193435817303e-06,
+      "loss": 0.904,
+      "step": 5050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918612487820274e-06,
+      "loss": 1.0086,
+      "step": 5051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915290626449055e-06,
+      "loss": 0.9658,
+      "step": 5052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91196877442596e-06,
+      "loss": 1.0362,
+      "step": 5053
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.908646932117572e-06,
+      "loss": 0.8793,
+      "step": 5054
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.90532509989047e-06,
+      "loss": 1.0579,
+      "step": 5055
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.902003278111245e-06,
+      "loss": 0.95,
+      "step": 5056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.898681467146469e-06,
+      "loss": 0.9195,
+      "step": 5057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.895359667362724e-06,
+      "loss": 1.1672,
+      "step": 5058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.892037879126588e-06,
+      "loss": 1.0196,
+      "step": 5059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.888716102804636e-06,
+      "loss": 1.0427,
+      "step": 5060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885394338763448e-06,
+      "loss": 1.0883,
+      "step": 5061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.882072587369595e-06,
+      "loss": 0.8845,
+      "step": 5062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878750848989654e-06,
+      "loss": 0.9209,
+      "step": 5063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.875429123990187e-06,
+      "loss": 0.7778,
+      "step": 5064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87210741273777e-06,
+      "loss": 1.0263,
+      "step": 5065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86878571559897e-06,
+      "loss": 0.9274,
+      "step": 5066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865464032940355e-06,
+      "loss": 0.9532,
+      "step": 5067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862142365128488e-06,
+      "loss": 1.0927,
+      "step": 5068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858820712529934e-06,
+      "loss": 1.3107,
+      "step": 5069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855499075511257e-06,
+      "loss": 0.9761,
+      "step": 5070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852177454439006e-06,
+      "loss": 1.1364,
+      "step": 5071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848855849679754e-06,
+      "loss": 1.0093,
+      "step": 5072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.845534261600053e-06,
+      "loss": 0.9606,
+      "step": 5073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842212690566455e-06,
+      "loss": 0.9161,
+      "step": 5074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83889113694551e-06,
+      "loss": 0.9935,
+      "step": 5075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.835569601103776e-06,
+      "loss": 1.1428,
+      "step": 5076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832248083407794e-06,
+      "loss": 1.0328,
+      "step": 5077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828926584224117e-06,
+      "loss": 1.2078,
+      "step": 5078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82560510391929e-06,
+      "loss": 1.0986,
+      "step": 5079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822283642859854e-06,
+      "loss": 0.7643,
+      "step": 5080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818962201412349e-06,
+      "loss": 1.0458,
+      "step": 5081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.815640779943309e-06,
+      "loss": 0.7539,
+      "step": 5082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81231937881928e-06,
+      "loss": 0.9571,
+      "step": 5083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80899799840679e-06,
+      "loss": 0.9556,
+      "step": 5084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805676639072374e-06,
+      "loss": 0.9681,
+      "step": 5085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802355301182558e-06,
+      "loss": 1.0989,
+      "step": 5086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.79903398510387e-06,
+      "loss": 1.0322,
+      "step": 5087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.795712691202832e-06,
+      "loss": 1.0631,
+      "step": 5088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.792391419845972e-06,
+      "loss": 1.667,
+      "step": 5089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.789070171399808e-06,
+      "loss": 1.1097,
+      "step": 5090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.785748946230856e-06,
+      "loss": 1.2615,
+      "step": 5091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78242774470563e-06,
+      "loss": 0.7532,
+      "step": 5092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.779106567190642e-06,
+      "loss": 1.0201,
+      "step": 5093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7757854140524e-06,
+      "loss": 1.1719,
+      "step": 5094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772464285657414e-06,
+      "loss": 1.0022,
+      "step": 5095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769143182372187e-06,
+      "loss": 0.6872,
+      "step": 5096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765822104563223e-06,
+      "loss": 0.9382,
+      "step": 5097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762501052597014e-06,
+      "loss": 1.0153,
+      "step": 5098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759180026840055e-06,
+      "loss": 0.9315,
+      "step": 5099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755859027658848e-06,
+      "loss": 0.9683,
+      "step": 5100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752538055419876e-06,
+      "loss": 1.2724,
+      "step": 5101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.749217110489625e-06,
+      "loss": 1.0018,
+      "step": 5102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74589619323458e-06,
+      "loss": 0.9697,
+      "step": 5103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742575304021223e-06,
+      "loss": 1.1811,
+      "step": 5104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739254443216028e-06,
+      "loss": 0.8364,
+      "step": 5105
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.735933611185473e-06,
+      "loss": 0.8105,
+      "step": 5106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73261280829603e-06,
+      "loss": 0.7232,
+      "step": 5107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729292034914164e-06,
+      "loss": 0.7213,
+      "step": 5108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72597129140634e-06,
+      "loss": 1.0718,
+      "step": 5109
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.722650578139019e-06,
+      "loss": 0.7797,
+      "step": 5110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71932989547866e-06,
+      "loss": 1.1011,
+      "step": 5111
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.716009243791721e-06,
+      "loss": 0.959,
+      "step": 5112
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712688623444648e-06,
+      "loss": 1.0522,
+      "step": 5113
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709368034803892e-06,
+      "loss": 1.0027,
+      "step": 5114
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7060474782359e-06,
+      "loss": 0.9817,
+      "step": 5115
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.702726954107101e-06,
+      "loss": 1.2846,
+      "step": 5116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.699406462783947e-06,
+      "loss": 0.6549,
+      "step": 5117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.696086004632866e-06,
+      "loss": 0.8584,
+      "step": 5118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.692765580020284e-06,
+      "loss": 1.0075,
+      "step": 5119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.689445189312632e-06,
+      "loss": 0.841,
+      "step": 5120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.686124832876327e-06,
+      "loss": 1.0449,
+      "step": 5121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.682804511077796e-06,
+      "loss": 0.6637,
+      "step": 5122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67948422428345e-06,
+      "loss": 0.943,
+      "step": 5123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.676163972859696e-06,
+      "loss": 1.1099,
+      "step": 5124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67284375717295e-06,
+      "loss": 0.9831,
+      "step": 5125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.669523577589608e-06,
+      "loss": 0.954,
+      "step": 5126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666203434476068e-06,
+      "loss": 0.8992,
+      "step": 5127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.662883328198732e-06,
+      "loss": 0.8533,
+      "step": 5128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.659563259123986e-06,
+      "loss": 0.8482,
+      "step": 5129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.656243227618219e-06,
+      "loss": 1.1077,
+      "step": 5130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652923234047812e-06,
+      "loss": 1.115,
+      "step": 5131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649603278779144e-06,
+      "loss": 0.9455,
+      "step": 5132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646283362178594e-06,
+      "loss": 0.7726,
+      "step": 5133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64296348461253e-06,
+      "loss": 1.142,
+      "step": 5134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.639643646447315e-06,
+      "loss": 0.9251,
+      "step": 5135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636323848049311e-06,
+      "loss": 0.8673,
+      "step": 5136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63300408978488e-06,
+      "loss": 0.8012,
+      "step": 5137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629684372020365e-06,
+      "loss": 1.0428,
+      "step": 5138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626364695122123e-06,
+      "loss": 0.793,
+      "step": 5139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.623045059456496e-06,
+      "loss": 1.0844,
+      "step": 5140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619725465389822e-06,
+      "loss": 0.8476,
+      "step": 5141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616405913288437e-06,
+      "loss": 1.1051,
+      "step": 5142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.613086403518666e-06,
+      "loss": 0.9108,
+      "step": 5143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.609766936446842e-06,
+      "loss": 1.0051,
+      "step": 5144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60644751243928e-06,
+      "loss": 1.1731,
+      "step": 5145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.603128131862299e-06,
+      "loss": 0.8887,
+      "step": 5146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599808795082209e-06,
+      "loss": 1.0216,
+      "step": 5147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596489502465315e-06,
+      "loss": 1.0202,
+      "step": 5148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593170254377915e-06,
+      "loss": 1.0084,
+      "step": 5149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.589851051186315e-06,
+      "loss": 1.1433,
+      "step": 5150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.5865318932568e-06,
+      "loss": 1.003,
+      "step": 5151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.58321278095566e-06,
+      "loss": 1.3184,
+      "step": 5152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.57989371464917e-06,
+      "loss": 1.0354,
+      "step": 5153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.57657469470361e-06,
+      "loss": 1.0285,
+      "step": 5154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.573255721485253e-06,
+      "loss": 0.8872,
+      "step": 5155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.569936795360363e-06,
+      "loss": 0.7879,
+      "step": 5156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566617916695203e-06,
+      "loss": 0.9563,
+      "step": 5157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.563299085856026e-06,
+      "loss": 0.9845,
+      "step": 5158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.559980303209086e-06,
+      "loss": 0.9144,
+      "step": 5159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55666156912062e-06,
+      "loss": 0.8106,
+      "step": 5160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.553342883956877e-06,
+      "loss": 0.9683,
+      "step": 5161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.550024248084089e-06,
+      "loss": 0.9641,
+      "step": 5162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.546705661868484e-06,
+      "loss": 1.1819,
+      "step": 5163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.543387125676283e-06,
+      "loss": 0.833,
+      "step": 5164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540068639873706e-06,
+      "loss": 0.9583,
+      "step": 5165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536750204826963e-06,
+      "loss": 0.8326,
+      "step": 5166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533431820902264e-06,
+      "loss": 0.9926,
+      "step": 5167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53011348846581e-06,
+      "loss": 1.0771,
+      "step": 5168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526795207883798e-06,
+      "loss": 1.1474,
+      "step": 5169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523476979522411e-06,
+      "loss": 1.094,
+      "step": 5170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.520158803747833e-06,
+      "loss": 0.909,
+      "step": 5171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.516840680926252e-06,
+      "loss": 0.9488,
+      "step": 5172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513522611423833e-06,
+      "loss": 0.6553,
+      "step": 5173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510204595606743e-06,
+      "loss": 1.081,
+      "step": 5174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506886633841143e-06,
+      "loss": 1.1541,
+      "step": 5175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503568726493189e-06,
+      "loss": 1.1397,
+      "step": 5176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500250873929021e-06,
+      "loss": 1.5846,
+      "step": 5177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496933076514794e-06,
+      "loss": 1.2837,
+      "step": 5178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49361533461664e-06,
+      "loss": 0.9319,
+      "step": 5179
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.490297648600685e-06,
+      "loss": 0.6993,
+      "step": 5180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486980018833057e-06,
+      "loss": 1.3279,
+      "step": 5181
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.483662445679868e-06,
+      "loss": 0.8686,
+      "step": 5182
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480344929507238e-06,
+      "loss": 1.0598,
+      "step": 5183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.477027470681269e-06,
+      "loss": 1.0842,
+      "step": 5184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.473710069568059e-06,
+      "loss": 0.9347,
+      "step": 5185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.470392726533703e-06,
+      "loss": 1.5629,
+      "step": 5186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.467075441944283e-06,
+      "loss": 0.9626,
+      "step": 5187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.463758216165877e-06,
+      "loss": 1.3354,
+      "step": 5188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.460441049564569e-06,
+      "loss": 1.1368,
+      "step": 5189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457123942506416e-06,
+      "loss": 1.2632,
+      "step": 5190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453806895357482e-06,
+      "loss": 0.8177,
+      "step": 5191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.450489908483819e-06,
+      "loss": 1.0775,
+      "step": 5192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.447172982251474e-06,
+      "loss": 0.9769,
+      "step": 5193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.443856117026488e-06,
+      "loss": 1.0356,
+      "step": 5194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440539313174896e-06,
+      "loss": 1.0946,
+      "step": 5195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437222571062723e-06,
+      "loss": 1.0345,
+      "step": 5196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433905891055987e-06,
+      "loss": 0.9646,
+      "step": 5197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430589273520704e-06,
+      "loss": 1.2613,
+      "step": 5198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427272718822876e-06,
+      "loss": 0.9641,
+      "step": 5199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423956227328508e-06,
+      "loss": 0.8566,
+      "step": 5200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420639799403588e-06,
+      "loss": 0.9088,
+      "step": 5201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417323435414103e-06,
+      "loss": 0.8324,
+      "step": 5202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414007135726031e-06,
+      "loss": 0.7648,
+      "step": 5203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410690900705341e-06,
+      "loss": 1.0024,
+      "step": 5204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407374730718e-06,
+      "loss": 1.0869,
+      "step": 5205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404058626129965e-06,
+      "loss": 1.3319,
+      "step": 5206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400742587307183e-06,
+      "loss": 0.9388,
+      "step": 5207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.397426614615597e-06,
+      "loss": 0.914,
+      "step": 5208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394110708421143e-06,
+      "loss": 0.8571,
+      "step": 5209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390794869089743e-06,
+      "loss": 1.0564,
+      "step": 5210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387479096987326e-06,
+      "loss": 1.1181,
+      "step": 5211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3841633924798e-06,
+      "loss": 0.5117,
+      "step": 5212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.380847755933072e-06,
+      "loss": 1.1522,
+      "step": 5213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37753218771304e-06,
+      "loss": 0.7996,
+      "step": 5214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374216688185594e-06,
+      "loss": 0.8414,
+      "step": 5215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370901257716611e-06,
+      "loss": 0.8926,
+      "step": 5216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.367585896671975e-06,
+      "loss": 0.8781,
+      "step": 5217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.364270605417549e-06,
+      "loss": 0.8407,
+      "step": 5218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.360955384319195e-06,
+      "loss": 0.9815,
+      "step": 5219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.357640233742761e-06,
+      "loss": 1.1944,
+      "step": 5220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.354325154054093e-06,
+      "loss": 1.0519,
+      "step": 5221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.351010145619031e-06,
+      "loss": 1.0703,
+      "step": 5222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347695208803403e-06,
+      "loss": 0.7136,
+      "step": 5223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.344380343973024e-06,
+      "loss": 0.8589,
+      "step": 5224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341065551493711e-06,
+      "loss": 1.0442,
+      "step": 5225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337750831731267e-06,
+      "loss": 0.9839,
+      "step": 5226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334436185051488e-06,
+      "loss": 0.6398,
+      "step": 5227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.331121611820167e-06,
+      "loss": 0.7906,
+      "step": 5228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32780711240308e-06,
+      "loss": 0.7951,
+      "step": 5229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324492687166002e-06,
+      "loss": 0.9644,
+      "step": 5230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321178336474698e-06,
+      "loss": 0.6915,
+      "step": 5231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.317864060694916e-06,
+      "loss": 0.9171,
+      "step": 5232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314549860192417e-06,
+      "loss": 0.8747,
+      "step": 5233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311235735332931e-06,
+      "loss": 1.1654,
+      "step": 5234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30792168648219e-06,
+      "loss": 1.0531,
+      "step": 5235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30460771400592e-06,
+      "loss": 1.2604,
+      "step": 5236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301293818269832e-06,
+      "loss": 1.0786,
+      "step": 5237
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297979999639632e-06,
+      "loss": 0.9948,
+      "step": 5238
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29466625848102e-06,
+      "loss": 0.9003,
+      "step": 5239
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291352595159682e-06,
+      "loss": 0.8044,
+      "step": 5240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288039010041303e-06,
+      "loss": 1.166,
+      "step": 5241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284725503491547e-06,
+      "loss": 1.1302,
+      "step": 5242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.281412075876078e-06,
+      "loss": 0.7546,
+      "step": 5243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278098727560556e-06,
+      "loss": 0.8657,
+      "step": 5244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274785458910622e-06,
+      "loss": 0.9287,
+      "step": 5245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271472270291914e-06,
+      "loss": 1.1525,
+      "step": 5246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268159162070058e-06,
+      "loss": 0.893,
+      "step": 5247
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264846134610677e-06,
+      "loss": 1.066,
+      "step": 5248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.26153318827937e-06,
+      "loss": 1.1093,
+      "step": 5249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.258220323441754e-06,
+      "loss": 1.0294,
+      "step": 5250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.254907540463412e-06,
+      "loss": 1.2094,
+      "step": 5251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.251594839709927e-06,
+      "loss": 0.9125,
+      "step": 5252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.248282221546873e-06,
+      "loss": 0.9544,
+      "step": 5253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244969686339813e-06,
+      "loss": 0.9419,
+      "step": 5254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.241657234454309e-06,
+      "loss": 0.8979,
+      "step": 5255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238344866255902e-06,
+      "loss": 0.8724,
+      "step": 5256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235032582110132e-06,
+      "loss": 0.9457,
+      "step": 5257
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231720382382526e-06,
+      "loss": 0.9901,
+      "step": 5258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228408267438603e-06,
+      "loss": 0.9051,
+      "step": 5259
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225096237643865e-06,
+      "loss": 1.0373,
+      "step": 5260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221784293363825e-06,
+      "loss": 0.9945,
+      "step": 5261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.218472434963965e-06,
+      "loss": 1.2625,
+      "step": 5262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215160662809768e-06,
+      "loss": 0.8058,
+      "step": 5263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211848977266706e-06,
+      "loss": 0.8987,
+      "step": 5264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208537378700241e-06,
+      "loss": 0.8764,
+      "step": 5265
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205225867475818e-06,
+      "loss": 1.2645,
+      "step": 5266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20191444395889e-06,
+      "loss": 0.8153,
+      "step": 5267
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198603108514889e-06,
+      "loss": 0.8948,
+      "step": 5268
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195291861509232e-06,
+      "loss": 0.9344,
+      "step": 5269
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191980703307336e-06,
+      "loss": 1.0562,
+      "step": 5270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188669634274601e-06,
+      "loss": 1.0386,
+      "step": 5271
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185358654776428e-06,
+      "loss": 0.9236,
+      "step": 5272
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182047765178195e-06,
+      "loss": 0.9404,
+      "step": 5273
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178736965845279e-06,
+      "loss": 0.7719,
+      "step": 5274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175426257143045e-06,
+      "loss": 1.2532,
+      "step": 5275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172115639436843e-06,
+      "loss": 1.4251,
+      "step": 5276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168805113092014e-06,
+      "loss": 1.0337,
+      "step": 5277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165494678473905e-06,
+      "loss": 0.9668,
+      "step": 5278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162184335947829e-06,
+      "loss": 0.9426,
+      "step": 5279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158874085879102e-06,
+      "loss": 1.1379,
+      "step": 5280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.155563928633028e-06,
+      "loss": 0.8598,
+      "step": 5281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.152253864574898e-06,
+      "loss": 1.124,
+      "step": 5282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.148943894069999e-06,
+      "loss": 0.9988,
+      "step": 5283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.145634017483603e-06,
+      "loss": 0.9082,
+      "step": 5284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.14232423518097e-06,
+      "loss": 0.9688,
+      "step": 5285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.139014547527354e-06,
+      "loss": 1.0918,
+      "step": 5286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135704954887993e-06,
+      "loss": 0.8672,
+      "step": 5287
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.132395457628119e-06,
+      "loss": 0.9665,
+      "step": 5288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129086056112955e-06,
+      "loss": 1.0014,
+      "step": 5289
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12577675070771e-06,
+      "loss": 1.0403,
+      "step": 5290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12246754177758e-06,
+      "loss": 0.8954,
+      "step": 5291
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.119158429687757e-06,
+      "loss": 0.9335,
+      "step": 5292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115849414803414e-06,
+      "loss": 1.1046,
+      "step": 5293
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112540497489727e-06,
+      "loss": 0.9475,
+      "step": 5294
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109231678111849e-06,
+      "loss": 0.8758,
+      "step": 5295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105922957034921e-06,
+      "loss": 0.893,
+      "step": 5296
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102614334624079e-06,
+      "loss": 1.1858,
+      "step": 5297
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09930581124445e-06,
+      "loss": 0.6201,
+      "step": 5298
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095997387261143e-06,
+      "loss": 1.13,
+      "step": 5299
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092689063039263e-06,
+      "loss": 0.8603,
+      "step": 5300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.089380838943901e-06,
+      "loss": 0.9938,
+      "step": 5301
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.086072715340138e-06,
+      "loss": 0.7756,
+      "step": 5302
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082764692593039e-06,
+      "loss": 1.0354,
+      "step": 5303
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079456771067658e-06,
+      "loss": 1.1115,
+      "step": 5304
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076148951129055e-06,
+      "loss": 0.7651,
+      "step": 5305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072841233142254e-06,
+      "loss": 0.6962,
+      "step": 5306
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06953361747228e-06,
+      "loss": 1.12,
+      "step": 5307
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.066226104484151e-06,
+      "loss": 0.9767,
+      "step": 5308
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062918694542865e-06,
+      "loss": 1.4015,
+      "step": 5309
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059611388013408e-06,
+      "loss": 1.0323,
+      "step": 5310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056304185260767e-06,
+      "loss": 1.0545,
+      "step": 5311
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052997086649907e-06,
+      "loss": 0.9007,
+      "step": 5312
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049690092545778e-06,
+      "loss": 0.6751,
+      "step": 5313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.046383203313329e-06,
+      "loss": 0.9449,
+      "step": 5314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.04307641931749e-06,
+      "loss": 0.9119,
+      "step": 5315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.039769740923183e-06,
+      "loss": 0.9289,
+      "step": 5316
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.03646316849532e-06,
+      "loss": 0.8548,
+      "step": 5317
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.033156702398795e-06,
+      "loss": 0.9513,
+      "step": 5318
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.029850342998494e-06,
+      "loss": 0.9847,
+      "step": 5319
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.026544090659296e-06,
+      "loss": 1.0814,
+      "step": 5320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023237945746053e-06,
+      "loss": 0.8147,
+      "step": 5321
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019931908623626e-06,
+      "loss": 1.0863,
+      "step": 5322
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.016625979656851e-06,
+      "loss": 1.116,
+      "step": 5323
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013320159210551e-06,
+      "loss": 0.8535,
+      "step": 5324
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010014447649542e-06,
+      "loss": 1.0518,
+      "step": 5325
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.006708845338627e-06,
+      "loss": 1.2429,
+      "step": 5326
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003403352642595e-06,
+      "loss": 0.8623,
+      "step": 5327
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.000097969926229e-06,
+      "loss": 0.5346,
+      "step": 5328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.996792697554293e-06,
+      "loss": 0.9233,
+      "step": 5329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993487535891541e-06,
+      "loss": 0.749,
+      "step": 5330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990182485302712e-06,
+      "loss": 0.8119,
+      "step": 5331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.986877546152539e-06,
+      "loss": 0.8683,
+      "step": 5332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983572718805737e-06,
+      "loss": 1.2716,
+      "step": 5333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980268003627016e-06,
+      "loss": 0.732,
+      "step": 5334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976963400981063e-06,
+      "loss": 1.0115,
+      "step": 5335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97365891123256e-06,
+      "loss": 0.7249,
+      "step": 5336
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970354534746178e-06,
+      "loss": 1.045,
+      "step": 5337
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.967050271886564e-06,
+      "loss": 0.7865,
+      "step": 5338
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.963746123018369e-06,
+      "loss": 0.7942,
+      "step": 5339
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.960442088506223e-06,
+      "loss": 1.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957138168714738e-06,
+      "loss": 0.8384,
+      "step": 5341
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.953834364008523e-06,
+      "loss": 0.9296,
+      "step": 5342
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950530674752165e-06,
+      "loss": 0.7772,
+      "step": 5343
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94722710131025e-06,
+      "loss": 1.4013,
+      "step": 5344
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943923644047343e-06,
+      "loss": 0.635,
+      "step": 5345
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.940620303327996e-06,
+      "loss": 1.1369,
+      "step": 5346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.937317079516753e-06,
+      "loss": 0.9876,
+      "step": 5347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.934013972978137e-06,
+      "loss": 1.2125,
+      "step": 5348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.93071098407666e-06,
+      "loss": 0.8528,
+      "step": 5349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92740811317684e-06,
+      "loss": 1.2293,
+      "step": 5350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.924105360643153e-06,
+      "loss": 0.8023,
+      "step": 5351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.920802726840076e-06,
+      "loss": 0.8499,
+      "step": 5352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.917500212132077e-06,
+      "loss": 1.6044,
+      "step": 5353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.9141978168836e-06,
+      "loss": 1.2211,
+      "step": 5354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910895541459085e-06,
+      "loss": 1.1716,
+      "step": 5355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907593386222956e-06,
+      "loss": 0.8467,
+      "step": 5356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904291351539623e-06,
+      "loss": 1.0388,
+      "step": 5357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900989437773481e-06,
+      "loss": 0.9532,
+      "step": 5358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897687645288916e-06,
+      "loss": 0.8229,
+      "step": 5359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894385974450292e-06,
+      "loss": 1.052,
+      "step": 5360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891084425621974e-06,
+      "loss": 0.753,
+      "step": 5361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8877829991683e-06,
+      "loss": 0.8954,
+      "step": 5362
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8844816954536e-06,
+      "loss": 0.7902,
+      "step": 5363
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.881180514842196e-06,
+      "loss": 0.8888,
+      "step": 5364
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.877879457698377e-06,
+      "loss": 1.1646,
+      "step": 5365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.874578524386447e-06,
+      "loss": 0.7879,
+      "step": 5366
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871277715270677e-06,
+      "loss": 0.9312,
+      "step": 5367
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.867977030715325e-06,
+      "loss": 1.0708,
+      "step": 5368
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.864676471084639e-06,
+      "loss": 1.0664,
+      "step": 5369
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861376036742857e-06,
+      "loss": 0.8458,
+      "step": 5370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858075728054192e-06,
+      "loss": 0.7136,
+      "step": 5371
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.854775545382858e-06,
+      "loss": 0.9962,
+      "step": 5372
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851475489093044e-06,
+      "loss": 0.7838,
+      "step": 5373
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848175559548933e-06,
+      "loss": 1.0613,
+      "step": 5374
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844875757114683e-06,
+      "loss": 1.0977,
+      "step": 5375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841576082154443e-06,
+      "loss": 0.9447,
+      "step": 5376
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838276535032358e-06,
+      "loss": 0.93,
+      "step": 5377
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834977116112547e-06,
+      "loss": 0.946,
+      "step": 5378
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.831677825759114e-06,
+      "loss": 0.8686,
+      "step": 5379
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.828378664336158e-06,
+      "loss": 0.661,
+      "step": 5380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.825079632207759e-06,
+      "loss": 0.9678,
+      "step": 5381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.821780729737974e-06,
+      "loss": 1.0263,
+      "step": 5382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.818481957290864e-06,
+      "loss": 0.7836,
+      "step": 5383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.815183315230464e-06,
+      "loss": 0.7718,
+      "step": 5384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811884803920792e-06,
+      "loss": 1.0374,
+      "step": 5385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.808586423725861e-06,
+      "loss": 0.7277,
+      "step": 5386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805288175009659e-06,
+      "loss": 0.9579,
+      "step": 5387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80199005813617e-06,
+      "loss": 0.8378,
+      "step": 5388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798692073469357e-06,
+      "loss": 0.8652,
+      "step": 5389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79539422137317e-06,
+      "loss": 0.9735,
+      "step": 5390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792096502211544e-06,
+      "loss": 1.0884,
+      "step": 5391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7887989163484e-06,
+      "loss": 0.8546,
+      "step": 5392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785501464147638e-06,
+      "loss": 0.8689,
+      "step": 5393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782204145973162e-06,
+      "loss": 0.9172,
+      "step": 5394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778906962188838e-06,
+      "loss": 1.0257,
+      "step": 5395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77560991315853e-06,
+      "loss": 1.2764,
+      "step": 5396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772312999246086e-06,
+      "loss": 1.2453,
+      "step": 5397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769016220815335e-06,
+      "loss": 0.9474,
+      "step": 5398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765719578230096e-06,
+      "loss": 0.7701,
+      "step": 5399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.762423071854171e-06,
+      "loss": 1.0841,
+      "step": 5400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759126702051349e-06,
+      "loss": 0.8583,
+      "step": 5401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755830469185397e-06,
+      "loss": 1.0218,
+      "step": 5402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752534373620071e-06,
+      "loss": 1.1499,
+      "step": 5403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749238415719115e-06,
+      "loss": 0.9213,
+      "step": 5404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.745942595846257e-06,
+      "loss": 1.0525,
+      "step": 5405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742646914365205e-06,
+      "loss": 0.7239,
+      "step": 5406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739351371639655e-06,
+      "loss": 0.8861,
+      "step": 5407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736055968033289e-06,
+      "loss": 1.1808,
+      "step": 5408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732760703909774e-06,
+      "loss": 1.2235,
+      "step": 5409
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72946557963275e-06,
+      "loss": 0.8866,
+      "step": 5410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.726170595565864e-06,
+      "loss": 0.84,
+      "step": 5411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.722875752072726e-06,
+      "loss": 0.7798,
+      "step": 5412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.719581049516945e-06,
+      "loss": 0.6662,
+      "step": 5413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.716286488262103e-06,
+      "loss": 0.8232,
+      "step": 5414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.712992068671774e-06,
+      "loss": 0.7033,
+      "step": 5415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.709697791109517e-06,
+      "loss": 1.2088,
+      "step": 5416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70640365593887e-06,
+      "loss": 0.8155,
+      "step": 5417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.703109663523362e-06,
+      "loss": 0.8377,
+      "step": 5418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6998158142265e-06,
+      "loss": 0.7848,
+      "step": 5419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696522108411776e-06,
+      "loss": 0.8192,
+      "step": 5420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.693228546442665e-06,
+      "loss": 1.192,
+      "step": 5421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.689935128682637e-06,
+      "loss": 0.9057,
+      "step": 5422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686641855495131e-06,
+      "loss": 0.6917,
+      "step": 5423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.683348727243581e-06,
+      "loss": 1.1388,
+      "step": 5424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680055744291398e-06,
+      "loss": 1.0622,
+      "step": 5425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676762907001983e-06,
+      "loss": 0.9441,
+      "step": 5426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.673470215738714e-06,
+      "loss": 0.8935,
+      "step": 5427
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67017767086496e-06,
+      "loss": 0.9463,
+      "step": 5428
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.666885272744073e-06,
+      "loss": 0.9317,
+      "step": 5429
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66359302173938e-06,
+      "loss": 1.0657,
+      "step": 5430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660300918214202e-06,
+      "loss": 0.9085,
+      "step": 5431
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657008962531836e-06,
+      "loss": 0.8195,
+      "step": 5432
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653717155055572e-06,
+      "loss": 0.8745,
+      "step": 5433
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650425496148676e-06,
+      "loss": 0.9066,
+      "step": 5434
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647133986174398e-06,
+      "loss": 1.0586,
+      "step": 5435
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643842625495978e-06,
+      "loss": 0.9462,
+      "step": 5436
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640551414476625e-06,
+      "loss": 1.0439,
+      "step": 5437
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637260353479554e-06,
+      "loss": 1.0809,
+      "step": 5438
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633969442867947e-06,
+      "loss": 0.8977,
+      "step": 5439
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630678683004968e-06,
+      "loss": 0.9502,
+      "step": 5440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627388074253773e-06,
+      "loss": 1.3448,
+      "step": 5441
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624097616977499e-06,
+      "loss": 0.9032,
+      "step": 5442
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620807311539258e-06,
+      "loss": 0.7652,
+      "step": 5443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.617517158302164e-06,
+      "loss": 1.4969,
+      "step": 5444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.614227157629298e-06,
+      "loss": 1.0066,
+      "step": 5445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.610937309883728e-06,
+      "loss": 0.8464,
+      "step": 5446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.607647615428504e-06,
+      "loss": 1.1016,
+      "step": 5447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.604358074626662e-06,
+      "loss": 0.7527,
+      "step": 5448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.601068687841222e-06,
+      "loss": 0.9207,
+      "step": 5449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.597779455435184e-06,
+      "loss": 0.6218,
+      "step": 5450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594490377771532e-06,
+      "loss": 0.8988,
+      "step": 5451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591201455213232e-06,
+      "loss": 0.9354,
+      "step": 5452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587912688123238e-06,
+      "loss": 0.7254,
+      "step": 5453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584624076864474e-06,
+      "loss": 1.0259,
+      "step": 5454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581335621799864e-06,
+      "loss": 0.8328,
+      "step": 5455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.578047323292307e-06,
+      "loss": 0.7209,
+      "step": 5456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574759181704677e-06,
+      "loss": 0.8036,
+      "step": 5457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57147119739984e-06,
+      "loss": 0.7436,
+      "step": 5458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.568183370740646e-06,
+      "loss": 0.8128,
+      "step": 5459
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564895702089917e-06,
+      "loss": 1.1221,
+      "step": 5460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561608191810472e-06,
+      "loss": 0.9566,
+      "step": 5461
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558320840265099e-06,
+      "loss": 0.9544,
+      "step": 5462
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55503364781658e-06,
+      "loss": 0.7828,
+      "step": 5463
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55174661482767e-06,
+      "loss": 1.0136,
+      "step": 5464
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.548459741661106e-06,
+      "loss": 0.9628,
+      "step": 5465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.545173028679624e-06,
+      "loss": 1.0999,
+      "step": 5466
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541886476245922e-06,
+      "loss": 1.2726,
+      "step": 5467
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538600084722687e-06,
+      "loss": 0.7513,
+      "step": 5468
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535313854472593e-06,
+      "loss": 0.8946,
+      "step": 5469
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532027785858293e-06,
+      "loss": 1.1179,
+      "step": 5470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528741879242414e-06,
+      "loss": 0.9917,
+      "step": 5471
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525456134987585e-06,
+      "loss": 1.1754,
+      "step": 5472
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5221705534564e-06,
+      "loss": 0.9458,
+      "step": 5473
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518885135011439e-06,
+      "loss": 0.9396,
+      "step": 5474
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515599880015264e-06,
+      "loss": 0.7343,
+      "step": 5475
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51231478883042e-06,
+      "loss": 1.0744,
+      "step": 5476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.50902986181944e-06,
+      "loss": 0.8226,
+      "step": 5477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.505745099344827e-06,
+      "loss": 1.0459,
+      "step": 5478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.502460501769075e-06,
+      "loss": 0.8015,
+      "step": 5479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.499176069454656e-06,
+      "loss": 1.2122,
+      "step": 5480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.495891802764022e-06,
+      "loss": 1.0121,
+      "step": 5481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.492607702059606e-06,
+      "loss": 0.9434,
+      "step": 5482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.489323767703836e-06,
+      "loss": 0.9044,
+      "step": 5483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486040000059105e-06,
+      "loss": 1.043,
+      "step": 5484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482756399487793e-06,
+      "loss": 1.0796,
+      "step": 5485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479472966352265e-06,
+      "loss": 1.1345,
+      "step": 5486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476189701014862e-06,
+      "loss": 0.9295,
+      "step": 5487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472906603837914e-06,
+      "loss": 1.1644,
+      "step": 5488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469623675183725e-06,
+      "loss": 1.1562,
+      "step": 5489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466340915414586e-06,
+      "loss": 0.9953,
+      "step": 5490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463058324892764e-06,
+      "loss": 0.9777,
+      "step": 5491
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45977590398051e-06,
+      "loss": 1.3473,
+      "step": 5492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456493653040056e-06,
+      "loss": 1.0,
+      "step": 5493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453211572433618e-06,
+      "loss": 0.9966,
+      "step": 5494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.449929662523389e-06,
+      "loss": 0.9169,
+      "step": 5495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446647923671546e-06,
+      "loss": 0.8566,
+      "step": 5496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443366356240248e-06,
+      "loss": 1.1015,
+      "step": 5497
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.440084960591626e-06,
+      "loss": 1.828,
+      "step": 5498
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436803737087807e-06,
+      "loss": 1.4566,
+      "step": 5499
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43352268609089e-06,
+      "loss": 0.7919,
+      "step": 5500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.430241807962953e-06,
+      "loss": 0.8222,
+      "step": 5501
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426961103066058e-06,
+      "loss": 0.805,
+      "step": 5502
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423680571762251e-06,
+      "loss": 0.8844,
+      "step": 5503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420400214413553e-06,
+      "loss": 1.0521,
+      "step": 5504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41712003138197e-06,
+      "loss": 0.8622,
+      "step": 5505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.413840023029488e-06,
+      "loss": 0.9914,
+      "step": 5506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.410560189718073e-06,
+      "loss": 0.8183,
+      "step": 5507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407280531809673e-06,
+      "loss": 1.1902,
+      "step": 5508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.404001049666211e-06,
+      "loss": 1.0157,
+      "step": 5509
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.400721743649596e-06,
+      "loss": 1.0873,
+      "step": 5510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.39744261412172e-06,
+      "loss": 0.8526,
+      "step": 5511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.394163661444451e-06,
+      "loss": 0.8733,
+      "step": 5512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.390884885979638e-06,
+      "loss": 0.9771,
+      "step": 5513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387606288089111e-06,
+      "loss": 0.9062,
+      "step": 5514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384327868134678e-06,
+      "loss": 0.9722,
+      "step": 5515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381049626478137e-06,
+      "loss": 0.9647,
+      "step": 5516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377771563481252e-06,
+      "loss": 0.9556,
+      "step": 5517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37449367950578e-06,
+      "loss": 1.187,
+      "step": 5518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.371215974913446e-06,
+      "loss": 0.9096,
+      "step": 5519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367938450065968e-06,
+      "loss": 0.7742,
+      "step": 5520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.364661105325032e-06,
+      "loss": 0.8014,
+      "step": 5521
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361383941052316e-06,
+      "loss": 0.8692,
+      "step": 5522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35810695760947e-06,
+      "loss": 0.9675,
+      "step": 5523
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354830155358127e-06,
+      "loss": 1.1569,
+      "step": 5524
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3515535346599e-06,
+      "loss": 0.9372,
+      "step": 5525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348277095876374e-06,
+      "loss": 0.9738,
+      "step": 5526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345000839369133e-06,
+      "loss": 1.2295,
+      "step": 5527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341724765499722e-06,
+      "loss": 1.0106,
+      "step": 5528
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.338448874629671e-06,
+      "loss": 0.9288,
+      "step": 5529
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335173167120499e-06,
+      "loss": 1.463,
+      "step": 5530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331897643333691e-06,
+      "loss": 1.0434,
+      "step": 5531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328622303630717e-06,
+      "loss": 1.0412,
+      "step": 5532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325347148373035e-06,
+      "loss": 0.8602,
+      "step": 5533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322072177922074e-06,
+      "loss": 0.7452,
+      "step": 5534
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318797392639242e-06,
+      "loss": 0.7686,
+      "step": 5535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315522792885926e-06,
+      "loss": 1.0665,
+      "step": 5536
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312248379023498e-06,
+      "loss": 1.2208,
+      "step": 5537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308974151413308e-06,
+      "loss": 0.9318,
+      "step": 5538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305700110416681e-06,
+      "loss": 1.2269,
+      "step": 5539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302426256394928e-06,
+      "loss": 0.9751,
+      "step": 5540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299152589709336e-06,
+      "loss": 0.9773,
+      "step": 5541
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.295879110721171e-06,
+      "loss": 0.7293,
+      "step": 5542
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.292605819791671e-06,
+      "loss": 0.9539,
+      "step": 5543
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.28933271728207e-06,
+      "loss": 0.7119,
+      "step": 5544
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.286059803553572e-06,
+      "loss": 1.0231,
+      "step": 5545
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.282787078967356e-06,
+      "loss": 1.1093,
+      "step": 5546
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.279514543884586e-06,
+      "loss": 0.8546,
+      "step": 5547
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2762421986664e-06,
+      "loss": 1.0027,
+      "step": 5548
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272970043673923e-06,
+      "loss": 1.0467,
+      "step": 5549
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269698079268255e-06,
+      "loss": 0.9417,
+      "step": 5550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266426305810472e-06,
+      "loss": 0.9642,
+      "step": 5551
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263154723661631e-06,
+      "loss": 0.8458,
+      "step": 5552
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25988333318277e-06,
+      "loss": 1.0886,
+      "step": 5553
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256612134734898e-06,
+      "loss": 0.792,
+      "step": 5554
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25334112867902e-06,
+      "loss": 1.2539,
+      "step": 5555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2500703153761e-06,
+      "loss": 1.0873,
+      "step": 5556
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246799695187092e-06,
+      "loss": 0.8187,
+      "step": 5557
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243529268472926e-06,
+      "loss": 0.9701,
+      "step": 5558
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240259035594509e-06,
+      "loss": 1.0757,
+      "step": 5559
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.236988996912732e-06,
+      "loss": 1.0198,
+      "step": 5560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23371915278846e-06,
+      "loss": 0.9101,
+      "step": 5561
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.230449503582538e-06,
+      "loss": 0.9499,
+      "step": 5562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227180049655785e-06,
+      "loss": 0.9859,
+      "step": 5563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.223910791369005e-06,
+      "loss": 0.9976,
+      "step": 5564
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.220641729082975e-06,
+      "loss": 1.1499,
+      "step": 5565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21737286315846e-06,
+      "loss": 1.0996,
+      "step": 5566
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21410419395619e-06,
+      "loss": 1.3281,
+      "step": 5567
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210835721836882e-06,
+      "loss": 0.9695,
+      "step": 5568
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207567447161231e-06,
+      "loss": 0.9941,
+      "step": 5569
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204299370289905e-06,
+      "loss": 1.1725,
+      "step": 5570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20103149158355e-06,
+      "loss": 1.107,
+      "step": 5571
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.197763811402805e-06,
+      "loss": 0.9533,
+      "step": 5572
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.194496330108265e-06,
+      "loss": 0.9628,
+      "step": 5573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.19122904806052e-06,
+      "loss": 0.7801,
+      "step": 5574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.187961965620128e-06,
+      "loss": 1.232,
+      "step": 5575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.184695083147628e-06,
+      "loss": 0.9369,
+      "step": 5576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.181428401003542e-06,
+      "loss": 0.7047,
+      "step": 5577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.178161919548362e-06,
+      "loss": 0.8065,
+      "step": 5578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174895639142564e-06,
+      "loss": 0.7508,
+      "step": 5579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171629560146597e-06,
+      "loss": 1.0866,
+      "step": 5580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16836368292089e-06,
+      "loss": 0.785,
+      "step": 5581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.165098007825848e-06,
+      "loss": 0.7835,
+      "step": 5582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16183253522186e-06,
+      "loss": 0.8666,
+      "step": 5583
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.158567265469284e-06,
+      "loss": 0.9649,
+      "step": 5584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155302198928463e-06,
+      "loss": 1.167,
+      "step": 5585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15203733595971e-06,
+      "loss": 0.6827,
+      "step": 5586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148772676923321e-06,
+      "loss": 0.7915,
+      "step": 5587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.145508222179572e-06,
+      "loss": 0.7815,
+      "step": 5588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142243972088711e-06,
+      "loss": 0.9363,
+      "step": 5589
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138979927010964e-06,
+      "loss": 0.9878,
+      "step": 5590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135716087306533e-06,
+      "loss": 0.9224,
+      "step": 5591
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132452453335604e-06,
+      "loss": 0.9529,
+      "step": 5592
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.129189025458332e-06,
+      "loss": 0.8509,
+      "step": 5593
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125925804034858e-06,
+      "loss": 0.7617,
+      "step": 5594
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122662789425293e-06,
+      "loss": 0.7984,
+      "step": 5595
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119399981989729e-06,
+      "loss": 0.9243,
+      "step": 5596
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.116137382088234e-06,
+      "loss": 1.0361,
+      "step": 5597
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112874990080848e-06,
+      "loss": 1.4764,
+      "step": 5598
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1096128063276e-06,
+      "loss": 0.9418,
+      "step": 5599
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106350831188487e-06,
+      "loss": 0.8992,
+      "step": 5600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.103089065023484e-06,
+      "loss": 1.0613,
+      "step": 5601
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099827508192542e-06,
+      "loss": 0.7522,
+      "step": 5602
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096566161055595e-06,
+      "loss": 1.1557,
+      "step": 5603
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093305023972546e-06,
+      "loss": 1.2867,
+      "step": 5604
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090044097303282e-06,
+      "loss": 0.8813,
+      "step": 5605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.086783381407666e-06,
+      "loss": 0.8551,
+      "step": 5606
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.083522876645526e-06,
+      "loss": 1.1555,
+      "step": 5607
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.080262583376682e-06,
+      "loss": 0.8252,
+      "step": 5608
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.077002501960921e-06,
+      "loss": 0.8833,
+      "step": 5609
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.073742632758013e-06,
+      "loss": 0.9968,
+      "step": 5610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.070482976127703e-06,
+      "loss": 1.0395,
+      "step": 5611
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.067223532429709e-06,
+      "loss": 0.7854,
+      "step": 5612
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063964302023725e-06,
+      "loss": 0.891,
+      "step": 5613
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060705285269432e-06,
+      "loss": 0.8212,
+      "step": 5614
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057446482526468e-06,
+      "loss": 1.0132,
+      "step": 5615
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05418789415447e-06,
+      "loss": 1.1807,
+      "step": 5616
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050929520513034e-06,
+      "loss": 1.192,
+      "step": 5617
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047671361961742e-06,
+      "loss": 0.9205,
+      "step": 5618
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.044413418860146e-06,
+      "loss": 0.9773,
+      "step": 5619
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041155691567777e-06,
+      "loss": 1.1027,
+      "step": 5620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037898180444145e-06,
+      "loss": 1.058,
+      "step": 5621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034640885848732e-06,
+      "loss": 0.9225,
+      "step": 5622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031383808140998e-06,
+      "loss": 0.954,
+      "step": 5623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02812694768038e-06,
+      "loss": 0.9342,
+      "step": 5624
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024870304826285e-06,
+      "loss": 1.0104,
+      "step": 5625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021613879938102e-06,
+      "loss": 0.9292,
+      "step": 5626
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0183576733752e-06,
+      "loss": 0.798,
+      "step": 5627
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.015101685496912e-06,
+      "loss": 0.945,
+      "step": 5628
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011845916662557e-06,
+      "loss": 0.8079,
+      "step": 5629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008590367231423e-06,
+      "loss": 0.9216,
+      "step": 5630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005335037562782e-06,
+      "loss": 0.6954,
+      "step": 5631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002079928015868e-06,
+      "loss": 1.2198,
+      "step": 5632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99882503894991e-06,
+      "loss": 0.8768,
+      "step": 5633
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995570370724096e-06,
+      "loss": 0.7593,
+      "step": 5634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.992315923697597e-06,
+      "loss": 1.0775,
+      "step": 5635
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.989061698229556e-06,
+      "loss": 0.8327,
+      "step": 5636
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985807694679096e-06,
+      "loss": 1.024,
+      "step": 5637
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.982553913405314e-06,
+      "loss": 0.9321,
+      "step": 5638
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.979300354767282e-06,
+      "loss": 1.087,
+      "step": 5639
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.976047019124045e-06,
+      "loss": 1.1316,
+      "step": 5640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.972793906834629e-06,
+      "loss": 0.805,
+      "step": 5641
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.969541018258028e-06,
+      "loss": 1.0381,
+      "step": 5642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.966288353753213e-06,
+      "loss": 0.8835,
+      "step": 5643
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.963035913679145e-06,
+      "loss": 0.8855,
+      "step": 5644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959783698394735e-06,
+      "loss": 1.1034,
+      "step": 5645
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956531708258887e-06,
+      "loss": 0.9801,
+      "step": 5646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.953279943630475e-06,
+      "loss": 0.8375,
+      "step": 5647
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950028404868345e-06,
+      "loss": 1.1959,
+      "step": 5648
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946777092331328e-06,
+      "loss": 1.091,
+      "step": 5649
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943526006378218e-06,
+      "loss": 0.7269,
+      "step": 5650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940275147367796e-06,
+      "loss": 0.756,
+      "step": 5651
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.9370245156588e-06,
+      "loss": 0.8815,
+      "step": 5652
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933774111609964e-06,
+      "loss": 0.751,
+      "step": 5653
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93052393557998e-06,
+      "loss": 0.9822,
+      "step": 5654
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927273987927529e-06,
+      "loss": 0.7676,
+      "step": 5655
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924024269011255e-06,
+      "loss": 0.9447,
+      "step": 5656
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.920774779189785e-06,
+      "loss": 0.7708,
+      "step": 5657
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917525518821716e-06,
+      "loss": 1.0457,
+      "step": 5658
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914276488265614e-06,
+      "loss": 0.8666,
+      "step": 5659
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911027687880037e-06,
+      "loss": 0.7058,
+      "step": 5660
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907779118023506e-06,
+      "loss": 0.965,
+      "step": 5661
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90453077905451e-06,
+      "loss": 0.9382,
+      "step": 5662
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.901282671331527e-06,
+      "loss": 0.9619,
+      "step": 5663
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898034795212999e-06,
+      "loss": 0.7766,
+      "step": 5664
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.894787151057346e-06,
+      "loss": 0.8387,
+      "step": 5665
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891539739222967e-06,
+      "loss": 0.8203,
+      "step": 5666
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888292560068227e-06,
+      "loss": 1.1193,
+      "step": 5667
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885045613951474e-06,
+      "loss": 0.9841,
+      "step": 5668
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.881798901231017e-06,
+      "loss": 1.0975,
+      "step": 5669
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878552422265152e-06,
+      "loss": 0.8728,
+      "step": 5670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875306177412149e-06,
+      "loss": 0.9302,
+      "step": 5671
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.872060167030243e-06,
+      "loss": 1.051,
+      "step": 5672
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.86881439147765e-06,
+      "loss": 1.106,
+      "step": 5673
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.86556885111256e-06,
+      "loss": 0.9343,
+      "step": 5674
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.862323546293132e-06,
+      "loss": 0.9263,
+      "step": 5675
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.859078477377504e-06,
+      "loss": 0.8748,
+      "step": 5676
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855833644723787e-06,
+      "loss": 1.0029,
+      "step": 5677
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852589048690068e-06,
+      "loss": 1.0382,
+      "step": 5678
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849344689634397e-06,
+      "loss": 0.8758,
+      "step": 5679
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846100567914814e-06,
+      "loss": 1.1467,
+      "step": 5680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84285668388932e-06,
+      "loss": 0.7159,
+      "step": 5681
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839613037915896e-06,
+      "loss": 0.9373,
+      "step": 5682
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836369630352496e-06,
+      "loss": 0.9863,
+      "step": 5683
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833126461557048e-06,
+      "loss": 1.0111,
+      "step": 5684
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829883531887454e-06,
+      "loss": 1.4663,
+      "step": 5685
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.826640841701581e-06,
+      "loss": 0.9141,
+      "step": 5686
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82339839135728e-06,
+      "loss": 1.0014,
+      "step": 5687
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82015618121238e-06,
+      "loss": 1.0786,
+      "step": 5688
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816914211624666e-06,
+      "loss": 1.057,
+      "step": 5689
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813672482951911e-06,
+      "loss": 0.9811,
+      "step": 5690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810430995551856e-06,
+      "loss": 0.7645,
+      "step": 5691
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807189749782218e-06,
+      "loss": 0.9088,
+      "step": 5692
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80394874600068e-06,
+      "loss": 0.9109,
+      "step": 5693
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800707984564908e-06,
+      "loss": 1.0259,
+      "step": 5694
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79746746583254e-06,
+      "loss": 1.0277,
+      "step": 5695
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794227190161178e-06,
+      "loss": 1.0932,
+      "step": 5696
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790987157908405e-06,
+      "loss": 1.1841,
+      "step": 5697
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.787747369431775e-06,
+      "loss": 0.8529,
+      "step": 5698
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78450782508882e-06,
+      "loss": 0.9387,
+      "step": 5699
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.781268525237037e-06,
+      "loss": 0.9363,
+      "step": 5700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778029470233902e-06,
+      "loss": 0.8778,
+      "step": 5701
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 1.1778,
+      "step": 5702
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771552096203328e-06,
+      "loss": 1.1102,
+      "step": 5703
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.768313777890698e-06,
+      "loss": 0.7894,
+      "step": 5704
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.765075705856348e-06,
+      "loss": 0.9291,
+      "step": 5705
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.761837880457603e-06,
+      "loss": 0.9166,
+      "step": 5706
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.758600302051777e-06,
+      "loss": 0.9337,
+      "step": 5707
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.755362970996156e-06,
+      "loss": 1.0527,
+      "step": 5708
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.752125887647992e-06,
+      "loss": 0.8269,
+      "step": 5709
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748889052364518e-06,
+      "loss": 0.8129,
+      "step": 5710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74565246550294e-06,
+      "loss": 0.9767,
+      "step": 5711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742416127420423e-06,
+      "loss": 1.0376,
+      "step": 5712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.739180038474121e-06,
+      "loss": 1.1384,
+      "step": 5713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735944199021149e-06,
+      "loss": 0.8712,
+      "step": 5714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732708609418598e-06,
+      "loss": 0.9541,
+      "step": 5715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.729473270023538e-06,
+      "loss": 1.1193,
+      "step": 5716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726238181193002e-06,
+      "loss": 0.9132,
+      "step": 5717
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723003343283999e-06,
+      "loss": 0.8539,
+      "step": 5718
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71976875665351e-06,
+      "loss": 0.9383,
+      "step": 5719
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716534421658488e-06,
+      "loss": 1.0701,
+      "step": 5720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71330033865586e-06,
+      "loss": 1.0209,
+      "step": 5721
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710066508002528e-06,
+      "loss": 0.7878,
+      "step": 5722
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706832930055357e-06,
+      "loss": 1.0159,
+      "step": 5723
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70359960517119e-06,
+      "loss": 0.9699,
+      "step": 5724
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700366533706843e-06,
+      "loss": 0.7476,
+      "step": 5725
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697133716019097e-06,
+      "loss": 0.9518,
+      "step": 5726
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693901152464716e-06,
+      "loss": 1.1802,
+      "step": 5727
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690668843400431e-06,
+      "loss": 1.1115,
+      "step": 5728
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687436789182942e-06,
+      "loss": 0.8583,
+      "step": 5729
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684204990168926e-06,
+      "loss": 1.0797,
+      "step": 5730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680973446715018e-06,
+      "loss": 1.0435,
+      "step": 5731
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67774215917785e-06,
+      "loss": 1.3626,
+      "step": 5732
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674511127914007e-06,
+      "loss": 0.9892,
+      "step": 5733
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671280353280047e-06,
+      "loss": 0.9784,
+      "step": 5734
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.668049835632505e-06,
+      "loss": 1.0843,
+      "step": 5735
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.664819575327885e-06,
+      "loss": 0.9494,
+      "step": 5736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.66158957272266e-06,
+      "loss": 0.9381,
+      "step": 5737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.658359828173283e-06,
+      "loss": 1.1494,
+      "step": 5738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.655130342036174e-06,
+      "loss": 0.9617,
+      "step": 5739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.651901114667722e-06,
+      "loss": 0.7962,
+      "step": 5740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.648672146424285e-06,
+      "loss": 0.667,
+      "step": 5741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6454434376622e-06,
+      "loss": 1.1389,
+      "step": 5742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642214988737772e-06,
+      "loss": 0.9962,
+      "step": 5743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638986800007278e-06,
+      "loss": 1.0806,
+      "step": 5744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635758871826964e-06,
+      "loss": 1.0847,
+      "step": 5745
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63253120455305e-06,
+      "loss": 1.0029,
+      "step": 5746
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629303798541726e-06,
+      "loss": 1.5179,
+      "step": 5747
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626076654149147e-06,
+      "loss": 1.0249,
+      "step": 5748
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622849771731456e-06,
+      "loss": 1.0245,
+      "step": 5749
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619623151644752e-06,
+      "loss": 0.8996,
+      "step": 5750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616396794245106e-06,
+      "loss": 0.8122,
+      "step": 5751
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613170699888568e-06,
+      "loss": 1.1889,
+      "step": 5752
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609944868931152e-06,
+      "loss": 0.986,
+      "step": 5753
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606719301728843e-06,
+      "loss": 0.9687,
+      "step": 5754
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603493998637605e-06,
+      "loss": 1.1895,
+      "step": 5755
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600268960013364e-06,
+      "loss": 0.9198,
+      "step": 5756
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597044186212021e-06,
+      "loss": 0.9507,
+      "step": 5757
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593819677589445e-06,
+      "loss": 1.2029,
+      "step": 5758
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590595434501474e-06,
+      "loss": 0.9066,
+      "step": 5759
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58737145730393e-06,
+      "loss": 0.9762,
+      "step": 5760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5841477463525895e-06,
+      "loss": 0.8378,
+      "step": 5761
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580924302003205e-06,
+      "loss": 1.0649,
+      "step": 5762
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5777011246115025e-06,
+      "loss": 1.141,
+      "step": 5763
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5744782145331785e-06,
+      "loss": 0.7618,
+      "step": 5764
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571255572123889e-06,
+      "loss": 1.0874,
+      "step": 5765
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56803319773928e-06,
+      "loss": 0.9641,
+      "step": 5766
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564811091734956e-06,
+      "loss": 0.6891,
+      "step": 5767
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561589254466488e-06,
+      "loss": 0.8888,
+      "step": 5768
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5583676862894255e-06,
+      "loss": 0.8715,
+      "step": 5769
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.555146387559284e-06,
+      "loss": 1.0707,
+      "step": 5770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.551925358631553e-06,
+      "loss": 0.9522,
+      "step": 5771
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.548704599861689e-06,
+      "loss": 1.0012,
+      "step": 5772
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.54548411160512e-06,
+      "loss": 1.1172,
+      "step": 5773
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.542263894217246e-06,
+      "loss": 0.8527,
+      "step": 5774
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.539043948053428e-06,
+      "loss": 1.4244,
+      "step": 5775
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535824273469006e-06,
+      "loss": 1.6705,
+      "step": 5776
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.532604870819296e-06,
+      "loss": 0.7678,
+      "step": 5777
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529385740459566e-06,
+      "loss": 0.9545,
+      "step": 5778
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52616688274507e-06,
+      "loss": 1.0727,
+      "step": 5779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5229482980310235e-06,
+      "loss": 1.0556,
+      "step": 5780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5197299866726106e-06,
+      "loss": 0.8649,
+      "step": 5781
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516511949024996e-06,
+      "loss": 0.8674,
+      "step": 5782
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.513294185443304e-06,
+      "loss": 0.8448,
+      "step": 5783
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510076696282633e-06,
+      "loss": 1.0259,
+      "step": 5784
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5068594818980455e-06,
+      "loss": 1.0437,
+      "step": 5785
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503642542644581e-06,
+      "loss": 0.9901,
+      "step": 5786
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5004258788772435e-06,
+      "loss": 1.2312,
+      "step": 5787
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497209490951012e-06,
+      "loss": 0.871,
+      "step": 5788
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4939933792208295e-06,
+      "loss": 0.9906,
+      "step": 5789
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490777544041613e-06,
+      "loss": 0.9371,
+      "step": 5790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487561985768245e-06,
+      "loss": 0.8833,
+      "step": 5791
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484346704755576e-06,
+      "loss": 0.9081,
+      "step": 5792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481131701358435e-06,
+      "loss": 1.0595,
+      "step": 5793
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477916975931615e-06,
+      "loss": 0.9732,
+      "step": 5794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4747025288298725e-06,
+      "loss": 1.091,
+      "step": 5795
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471488360407942e-06,
+      "loss": 1.0265,
+      "step": 5796
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46827447102052e-06,
+      "loss": 1.2386,
+      "step": 5797
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.465060861022277e-06,
+      "loss": 0.8845,
+      "step": 5798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461847530767857e-06,
+      "loss": 1.0137,
+      "step": 5799
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.458634480611864e-06,
+      "loss": 0.7994,
+      "step": 5800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.455421710908876e-06,
+      "loss": 1.2028,
+      "step": 5801
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.452209222013439e-06,
+      "loss": 0.9077,
+      "step": 5802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.448997014280066e-06,
+      "loss": 0.9482,
+      "step": 5803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.445785088063239e-06,
+      "loss": 0.8377,
+      "step": 5804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.442573443717416e-06,
+      "loss": 1.0422,
+      "step": 5805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.439362081597018e-06,
+      "loss": 0.9402,
+      "step": 5806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4361510020564336e-06,
+      "loss": 0.9856,
+      "step": 5807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432940205450023e-06,
+      "loss": 1.026,
+      "step": 5808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429729692132114e-06,
+      "loss": 0.942,
+      "step": 5809
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426519462457004e-06,
+      "loss": 0.98,
+      "step": 5810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42330951677896e-06,
+      "loss": 1.1701,
+      "step": 5811
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4200998554522185e-06,
+      "loss": 0.9462,
+      "step": 5812
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4168904788309745e-06,
+      "loss": 0.9057,
+      "step": 5813
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413681387269406e-06,
+      "loss": 0.9792,
+      "step": 5814
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410472581121648e-06,
+      "loss": 1.1687,
+      "step": 5815
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.407264060741814e-06,
+      "loss": 1.3065,
+      "step": 5816
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40405582648398e-06,
+      "loss": 0.9423,
+      "step": 5817
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.400847878702191e-06,
+      "loss": 0.9237,
+      "step": 5818
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3976402177504625e-06,
+      "loss": 0.8052,
+      "step": 5819
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394432843982768e-06,
+      "loss": 0.947,
+      "step": 5820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391225757753069e-06,
+      "loss": 0.9406,
+      "step": 5821
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3880189594152826e-06,
+      "loss": 0.9336,
+      "step": 5822
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384812449323291e-06,
+      "loss": 1.3157,
+      "step": 5823
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.381606227830952e-06,
+      "loss": 1.3258,
+      "step": 5824
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378400295292089e-06,
+      "loss": 1.1539,
+      "step": 5825
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.37519465206049e-06,
+      "loss": 0.9428,
+      "step": 5826
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.37198929848992e-06,
+      "loss": 0.8016,
+      "step": 5827
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368784234934104e-06,
+      "loss": 0.8145,
+      "step": 5828
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36557946174674e-06,
+      "loss": 1.0095,
+      "step": 5829
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.362374979281488e-06,
+      "loss": 1.0241,
+      "step": 5830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359170787891975e-06,
+      "loss": 0.9796,
+      "step": 5831
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3559668879318135e-06,
+      "loss": 0.783,
+      "step": 5832
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35276327975456e-06,
+      "loss": 0.8172,
+      "step": 5833
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.349559963713753e-06,
+      "loss": 0.9921,
+      "step": 5834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.346356940162895e-06,
+      "loss": 0.9705,
+      "step": 5835
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.343154209455459e-06,
+      "loss": 0.9701,
+      "step": 5836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.339951771944873e-06,
+      "loss": 1.2152,
+      "step": 5837
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.336749627984554e-06,
+      "loss": 1.2384,
+      "step": 5838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.333547777927875e-06,
+      "loss": 0.9131,
+      "step": 5839
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330346222128171e-06,
+      "loss": 0.8468,
+      "step": 5840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327144960938752e-06,
+      "loss": 1.0695,
+      "step": 5841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323943994712895e-06,
+      "loss": 0.8914,
+      "step": 5842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3207433238038425e-06,
+      "loss": 1.1259,
+      "step": 5843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317542948564806e-06,
+      "loss": 0.9538,
+      "step": 5844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314342869348966e-06,
+      "loss": 0.7977,
+      "step": 5845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.311143086509467e-06,
+      "loss": 1.102,
+      "step": 5846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307943600399419e-06,
+      "loss": 1.0323,
+      "step": 5847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304744411371898e-06,
+      "loss": 0.816,
+      "step": 5848
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301545519779965e-06,
+      "loss": 1.0028,
+      "step": 5849
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298346925976623e-06,
+      "loss": 0.9303,
+      "step": 5850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295148630314857e-06,
+      "loss": 0.8554,
+      "step": 5851
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291950633147616e-06,
+      "loss": 0.8726,
+      "step": 5852
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288752934827812e-06,
+      "loss": 0.8563,
+      "step": 5853
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285555535708333e-06,
+      "loss": 0.6849,
+      "step": 5854
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282358436142028e-06,
+      "loss": 0.9712,
+      "step": 5855
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.279161636481715e-06,
+      "loss": 1.0158,
+      "step": 5856
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275965137080173e-06,
+      "loss": 0.9498,
+      "step": 5857
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.272768938290155e-06,
+      "loss": 0.946,
+      "step": 5858
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269573040464375e-06,
+      "loss": 0.8828,
+      "step": 5859
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266377443955523e-06,
+      "loss": 1.0895,
+      "step": 5860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263182149116246e-06,
+      "loss": 0.9717,
+      "step": 5861
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259987156299162e-06,
+      "loss": 1.404,
+      "step": 5862
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256792465856859e-06,
+      "loss": 0.9465,
+      "step": 5863
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.253598078141882e-06,
+      "loss": 0.8217,
+      "step": 5864
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250403993506745e-06,
+      "loss": 0.989,
+      "step": 5865
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.247210212303946e-06,
+      "loss": 0.9009,
+      "step": 5866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.244016734885924e-06,
+      "loss": 1.297,
+      "step": 5867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2408235616051e-06,
+      "loss": 0.8575,
+      "step": 5868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.237630692813856e-06,
+      "loss": 1.1093,
+      "step": 5869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.23443812886454e-06,
+      "loss": 1.0626,
+      "step": 5870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.231245870109475e-06,
+      "loss": 1.0489,
+      "step": 5871
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228053916900937e-06,
+      "loss": 1.1611,
+      "step": 5872
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224862269591179e-06,
+      "loss": 0.7137,
+      "step": 5873
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221670928532414e-06,
+      "loss": 0.7369,
+      "step": 5874
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2184798940768205e-06,
+      "loss": 0.863,
+      "step": 5875
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215289166576548e-06,
+      "loss": 0.8275,
+      "step": 5876
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212098746383711e-06,
+      "loss": 1.1717,
+      "step": 5877
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2089086338503914e-06,
+      "loss": 0.9795,
+      "step": 5878
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.205718829328629e-06,
+      "loss": 1.0351,
+      "step": 5879
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2025293331704405e-06,
+      "loss": 0.7885,
+      "step": 5880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199340145727798e-06,
+      "loss": 0.9883,
+      "step": 5881
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.196151267352651e-06,
+      "loss": 1.0627,
+      "step": 5882
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192962698396908e-06,
+      "loss": 0.7862,
+      "step": 5883
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189774439212442e-06,
+      "loss": 1.1347,
+      "step": 5884
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186586490151095e-06,
+      "loss": 1.0418,
+      "step": 5885
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183398851564673e-06,
+      "loss": 1.0675,
+      "step": 5886
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180211523804947e-06,
+      "loss": 0.7848,
+      "step": 5887
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17702450722366e-06,
+      "loss": 1.0351,
+      "step": 5888
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173837802172514e-06,
+      "loss": 1.1323,
+      "step": 5889
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170651409003177e-06,
+      "loss": 0.8972,
+      "step": 5890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167465328067289e-06,
+      "loss": 1.2792,
+      "step": 5891
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.16427955971644e-06,
+      "loss": 0.9586,
+      "step": 5892
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.161094104302211e-06,
+      "loss": 1.1232,
+      "step": 5893
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157908962176123e-06,
+      "loss": 0.8506,
+      "step": 5894
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.154724133689677e-06,
+      "loss": 1.2776,
+      "step": 5895
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151539619194335e-06,
+      "loss": 0.9601,
+      "step": 5896
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1483554190415235e-06,
+      "loss": 0.8349,
+      "step": 5897
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145171533582636e-06,
+      "loss": 0.8474,
+      "step": 5898
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.141987963169032e-06,
+      "loss": 0.8751,
+      "step": 5899
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.138804708152036e-06,
+      "loss": 0.9448,
+      "step": 5900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.135621768882939e-06,
+      "loss": 0.9757,
+      "step": 5901
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.132439145712988e-06,
+      "loss": 0.9491,
+      "step": 5902
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.129256838993406e-06,
+      "loss": 0.8916,
+      "step": 5903
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1260748490753774e-06,
+      "loss": 1.2137,
+      "step": 5904
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122893176310052e-06,
+      "loss": 0.9787,
+      "step": 5905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119711821048543e-06,
+      "loss": 0.9283,
+      "step": 5906
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116530783641931e-06,
+      "loss": 1.2049,
+      "step": 5907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.11335006444126e-06,
+      "loss": 0.848,
+      "step": 5908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110169663797532e-06,
+      "loss": 0.9941,
+      "step": 5909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106989582061732e-06,
+      "loss": 0.8177,
+      "step": 5910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.103809819584795e-06,
+      "loss": 0.8551,
+      "step": 5911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100630376717621e-06,
+      "loss": 0.9921,
+      "step": 5912
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097451253811081e-06,
+      "loss": 1.1236,
+      "step": 5913
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0942724512160045e-06,
+      "loss": 0.8806,
+      "step": 5914
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091093969283195e-06,
+      "loss": 1.2367,
+      "step": 5915
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.08791580836341e-06,
+      "loss": 0.836,
+      "step": 5916
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084737968807379e-06,
+      "loss": 1.0225,
+      "step": 5917
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.081560450965794e-06,
+      "loss": 1.0328,
+      "step": 5918
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078383255189305e-06,
+      "loss": 0.87,
+      "step": 5919
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.075206381828534e-06,
+      "loss": 0.8915,
+      "step": 5920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0720298312340705e-06,
+      "loss": 0.9811,
+      "step": 5921
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.06885360375646e-06,
+      "loss": 1.6288,
+      "step": 5922
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065677699746217e-06,
+      "loss": 1.0772,
+      "step": 5923
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0625021195538176e-06,
+      "loss": 0.7261,
+      "step": 5924
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059326863529707e-06,
+      "loss": 0.8832,
+      "step": 5925
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056151932024282e-06,
+      "loss": 0.9984,
+      "step": 5926
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052977325387925e-06,
+      "loss": 0.8863,
+      "step": 5927
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049803043970967e-06,
+      "loss": 0.8289,
+      "step": 5928
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046629088123703e-06,
+      "loss": 0.7536,
+      "step": 5929
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043455458196398e-06,
+      "loss": 0.8987,
+      "step": 5930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040282154539275e-06,
+      "loss": 0.9712,
+      "step": 5931
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.03710917750253e-06,
+      "loss": 0.7629,
+      "step": 5932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.033936527436318e-06,
+      "loss": 1.0396,
+      "step": 5933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.030764204690752e-06,
+      "loss": 0.8694,
+      "step": 5934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.027592209615922e-06,
+      "loss": 0.9184,
+      "step": 5935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.024420542561867e-06,
+      "loss": 1.1058,
+      "step": 5936
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.021249203878595e-06,
+      "loss": 1.1182,
+      "step": 5937
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018078193916092e-06,
+      "loss": 0.9425,
+      "step": 5938
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014907513024287e-06,
+      "loss": 0.8113,
+      "step": 5939
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011737161553082e-06,
+      "loss": 0.7143,
+      "step": 5940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008567139852343e-06,
+      "loss": 0.9324,
+      "step": 5941
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.005397448271893e-06,
+      "loss": 1.086,
+      "step": 5942
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0022280871615355e-06,
+      "loss": 0.9915,
+      "step": 5943
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999059056871017e-06,
+      "loss": 1.0071,
+      "step": 5944
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.995890357750063e-06,
+      "loss": 0.7496,
+      "step": 5945
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992721990148349e-06,
+      "loss": 1.3423,
+      "step": 5946
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989553954415524e-06,
+      "loss": 0.7273,
+      "step": 5947
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986386250901197e-06,
+      "loss": 1.0528,
+      "step": 5948
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983218879954943e-06,
+      "loss": 0.937,
+      "step": 5949
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.980051841926296e-06,
+      "loss": 0.7238,
+      "step": 5950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976885137164757e-06,
+      "loss": 0.9182,
+      "step": 5951
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973718766019788e-06,
+      "loss": 0.8651,
+      "step": 5952
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970552728840807e-06,
+      "loss": 1.195,
+      "step": 5953
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9673870259772145e-06,
+      "loss": 0.7784,
+      "step": 5954
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964221657778361e-06,
+      "loss": 0.8326,
+      "step": 5955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961056624593553e-06,
+      "loss": 0.8679,
+      "step": 5956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957891926772074e-06,
+      "loss": 1.105,
+      "step": 5957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954727564663166e-06,
+      "loss": 1.01,
+      "step": 5958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951563538616029e-06,
+      "loss": 1.1522,
+      "step": 5959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.948399848979832e-06,
+      "loss": 0.8603,
+      "step": 5960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945236496103707e-06,
+      "loss": 0.9497,
+      "step": 5961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942073480336746e-06,
+      "loss": 0.9493,
+      "step": 5962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938910802028002e-06,
+      "loss": 1.0768,
+      "step": 5963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9357484615264884e-06,
+      "loss": 0.8049,
+      "step": 5964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.932586459181199e-06,
+      "loss": 0.9639,
+      "step": 5965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.929424795341067e-06,
+      "loss": 1.2881,
+      "step": 5966
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.926263470355002e-06,
+      "loss": 0.8534,
+      "step": 5967
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.923102484571871e-06,
+      "loss": 0.7724,
+      "step": 5968
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919941838340507e-06,
+      "loss": 0.9103,
+      "step": 5969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916781532009702e-06,
+      "loss": 1.1992,
+      "step": 5970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913621565928217e-06,
+      "loss": 1.0094,
+      "step": 5971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.910461940444769e-06,
+      "loss": 0.9975,
+      "step": 5972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907302655908036e-06,
+      "loss": 1.0502,
+      "step": 5973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.904143712666663e-06,
+      "loss": 0.853,
+      "step": 5974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900985111069256e-06,
+      "loss": 0.6553,
+      "step": 5975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8978268514643845e-06,
+      "loss": 1.0012,
+      "step": 5976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894668934200578e-06,
+      "loss": 1.0845,
+      "step": 5977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89151135962633e-06,
+      "loss": 0.78,
+      "step": 5978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888354128090096e-06,
+      "loss": 0.9674,
+      "step": 5979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.885197239940291e-06,
+      "loss": 1.1253,
+      "step": 5980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882040695525292e-06,
+      "loss": 0.8309,
+      "step": 5981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878884495193448e-06,
+      "loss": 0.8715,
+      "step": 5982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875728639293056e-06,
+      "loss": 0.9478,
+      "step": 5983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872573128172381e-06,
+      "loss": 1.0161,
+      "step": 5984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.869417962179654e-06,
+      "loss": 0.9384,
+      "step": 5985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.866263141663063e-06,
+      "loss": 0.7947,
+      "step": 5986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863108666970755e-06,
+      "loss": 0.9819,
+      "step": 5987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859954538450846e-06,
+      "loss": 1.0416,
+      "step": 5988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.856800756451415e-06,
+      "loss": 1.0468,
+      "step": 5989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853647321320491e-06,
+      "loss": 0.8968,
+      "step": 5990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850494233406076e-06,
+      "loss": 0.768,
+      "step": 5991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.847341493056124e-06,
+      "loss": 0.896,
+      "step": 5992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844189100618565e-06,
+      "loss": 0.953,
+      "step": 5993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841037056441278e-06,
+      "loss": 1.3021,
+      "step": 5994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837885360872107e-06,
+      "loss": 1.069,
+      "step": 5995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.83473401425886e-06,
+      "loss": 0.6034,
+      "step": 5996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8315830169493056e-06,
+      "loss": 0.9402,
+      "step": 5997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.828432369291164e-06,
+      "loss": 0.8356,
+      "step": 5998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8252820716321385e-06,
+      "loss": 0.9754,
+      "step": 5999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.822132124319874e-06,
+      "loss": 1.1317,
+      "step": 6000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.818982527701983e-06,
+      "loss": 0.8818,
+      "step": 6001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.815833282126043e-06,
+      "loss": 0.9687,
+      "step": 6002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.812684387939585e-06,
+      "loss": 0.9856,
+      "step": 6003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.809535845490112e-06,
+      "loss": 0.6862,
+      "step": 6004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806387655125079e-06,
+      "loss": 0.8458,
+      "step": 6005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803239817191907e-06,
+      "loss": 0.9126,
+      "step": 6006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800092332037978e-06,
+      "loss": 0.8855,
+      "step": 6007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796945200010627e-06,
+      "loss": 1.1761,
+      "step": 6008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793798421457158e-06,
+      "loss": 0.621,
+      "step": 6009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.79065199672484e-06,
+      "loss": 1.1222,
+      "step": 6010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7875059261608926e-06,
+      "loss": 1.0517,
+      "step": 6011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.784360210112504e-06,
+      "loss": 0.975,
+      "step": 6012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.781214848926818e-06,
+      "loss": 1.0363,
+      "step": 6013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778069842950942e-06,
+      "loss": 0.7497,
+      "step": 6014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.774925192531947e-06,
+      "loss": 1.3356,
+      "step": 6015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771780898016858e-06,
+      "loss": 0.8278,
+      "step": 6016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768636959752669e-06,
+      "loss": 0.5491,
+      "step": 6017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7654933780863255e-06,
+      "loss": 0.927,
+      "step": 6018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.762350153364738e-06,
+      "loss": 0.9149,
+      "step": 6019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759207285934778e-06,
+      "loss": 1.0812,
+      "step": 6020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75606477614328e-06,
+      "loss": 0.9023,
+      "step": 6021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752922624337037e-06,
+      "loss": 1.0103,
+      "step": 6022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7497808308628e-06,
+      "loss": 0.795,
+      "step": 6023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746639396067284e-06,
+      "loss": 1.1377,
+      "step": 6024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743498320297155e-06,
+      "loss": 0.7925,
+      "step": 6025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.740357603899057e-06,
+      "loss": 1.0563,
+      "step": 6026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737217247219585e-06,
+      "loss": 0.9617,
+      "step": 6027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.734077250605286e-06,
+      "loss": 0.9002,
+      "step": 6028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.730937614402679e-06,
+      "loss": 1.0105,
+      "step": 6029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.72779833895824e-06,
+      "loss": 0.9424,
+      "step": 6030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.724659424618401e-06,
+      "loss": 1.0146,
+      "step": 6031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.721520871729563e-06,
+      "loss": 0.9364,
+      "step": 6032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.718382680638081e-06,
+      "loss": 1.0911,
+      "step": 6033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.715244851690268e-06,
+      "loss": 0.8704,
+      "step": 6034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7121073852324e-06,
+      "loss": 0.6677,
+      "step": 6035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70897028161071e-06,
+      "loss": 0.9174,
+      "step": 6036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705833541171403e-06,
+      "loss": 0.7763,
+      "step": 6037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.702697164260627e-06,
+      "loss": 1.286,
+      "step": 6038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6995611512245e-06,
+      "loss": 0.9856,
+      "step": 6039
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.696425502409098e-06,
+      "loss": 0.7646,
+      "step": 6040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693290218160457e-06,
+      "loss": 1.3472,
+      "step": 6041
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690155298824565e-06,
+      "loss": 1.1095,
+      "step": 6042
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687020744747386e-06,
+      "loss": 0.8072,
+      "step": 6043
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.683886556274834e-06,
+      "loss": 0.9948,
+      "step": 6044
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680752733752778e-06,
+      "loss": 1.0497,
+      "step": 6045
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6776192775270525e-06,
+      "loss": 1.5342,
+      "step": 6046
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6744861879434534e-06,
+      "loss": 1.3447,
+      "step": 6047
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671353465347729e-06,
+      "loss": 1.0499,
+      "step": 6048
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6682211100855974e-06,
+      "loss": 0.9604,
+      "step": 6049
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.665089122502729e-06,
+      "loss": 0.8693,
+      "step": 6050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661957502944756e-06,
+      "loss": 0.8191,
+      "step": 6051
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6588262517572645e-06,
+      "loss": 0.9085,
+      "step": 6052
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655695369285802e-06,
+      "loss": 0.824,
+      "step": 6053
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65256485587589e-06,
+      "loss": 0.9365,
+      "step": 6054
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649434711872988e-06,
+      "loss": 0.818,
+      "step": 6055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.646304937622525e-06,
+      "loss": 1.1956,
+      "step": 6056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.64317553346989e-06,
+      "loss": 0.7832,
+      "step": 6057
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6400464997604265e-06,
+      "loss": 0.8962,
+      "step": 6058
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.636917836839439e-06,
+      "loss": 0.933,
+      "step": 6059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633789545052196e-06,
+      "loss": 1.0265,
+      "step": 6060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6306616247439215e-06,
+      "loss": 1.0924,
+      "step": 6061
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.627534076259794e-06,
+      "loss": 0.7944,
+      "step": 6062
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.624406899944955e-06,
+      "loss": 0.775,
+      "step": 6063
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.621280096144503e-06,
+      "loss": 0.823,
+      "step": 6064
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.618153665203504e-06,
+      "loss": 1.0309,
+      "step": 6065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.61502760746697e-06,
+      "loss": 0.906,
+      "step": 6066
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.611901923279882e-06,
+      "loss": 1.1048,
+      "step": 6067
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608776612987174e-06,
+      "loss": 1.2004,
+      "step": 6068
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60565167693374e-06,
+      "loss": 1.0354,
+      "step": 6069
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6025271154644266e-06,
+      "loss": 1.031,
+      "step": 6070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599402928924059e-06,
+      "loss": 1.0827,
+      "step": 6071
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.596279117657398e-06,
+      "loss": 1.2482,
+      "step": 6072
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593155682009176e-06,
+      "loss": 1.2517,
+      "step": 6073
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5900326223240794e-06,
+      "loss": 1.0775,
+      "step": 6074
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586909938946753e-06,
+      "loss": 1.1417,
+      "step": 6075
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.583787632221803e-06,
+      "loss": 0.9618,
+      "step": 6076
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580665702493794e-06,
+      "loss": 0.8807,
+      "step": 6077
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577544150107246e-06,
+      "loss": 0.6997,
+      "step": 6078
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574422975406636e-06,
+      "loss": 1.031,
+      "step": 6079
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.571302178736404e-06,
+      "loss": 0.7395,
+      "step": 6080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568181760440943e-06,
+      "loss": 0.891,
+      "step": 6081
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.565061720864615e-06,
+      "loss": 0.9499,
+      "step": 6082
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561942060351727e-06,
+      "loss": 1.0125,
+      "step": 6083
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558822779246552e-06,
+      "loss": 1.0447,
+      "step": 6084
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555703877893318e-06,
+      "loss": 1.027,
+      "step": 6085
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552585356636211e-06,
+      "loss": 0.6298,
+      "step": 6086
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549467215819378e-06,
+      "loss": 0.8029,
+      "step": 6087
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546349455786926e-06,
+      "loss": 0.8738,
+      "step": 6088
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543232076882908e-06,
+      "loss": 0.9887,
+      "step": 6089
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540115079451348e-06,
+      "loss": 1.6497,
+      "step": 6090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536998463836222e-06,
+      "loss": 0.9216,
+      "step": 6091
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533882230381463e-06,
+      "loss": 0.8328,
+      "step": 6092
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.530766379430967e-06,
+      "loss": 0.9681,
+      "step": 6093
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.527650911328585e-06,
+      "loss": 0.8996,
+      "step": 6094
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.524535826418121e-06,
+      "loss": 0.8482,
+      "step": 6095
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5214211250433454e-06,
+      "loss": 0.9411,
+      "step": 6096
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.518306807547975e-06,
+      "loss": 0.7162,
+      "step": 6097
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.515192874275701e-06,
+      "loss": 0.6359,
+      "step": 6098
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.512079325570155e-06,
+      "loss": 0.902,
+      "step": 6099
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5089661617749354e-06,
+      "loss": 0.7569,
+      "step": 6100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.505853383233597e-06,
+      "loss": 0.8187,
+      "step": 6101
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.502740990289649e-06,
+      "loss": 0.8846,
+      "step": 6102
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499628983286561e-06,
+      "loss": 0.9888,
+      "step": 6103
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496517362567762e-06,
+      "loss": 0.9687,
+      "step": 6104
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493406128476633e-06,
+      "loss": 0.9665,
+      "step": 6105
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490295281356517e-06,
+      "loss": 1.0338,
+      "step": 6106
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48718482155071e-06,
+      "loss": 1.0522,
+      "step": 6107
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484074749402467e-06,
+      "loss": 1.3465,
+      "step": 6108
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.480965065255e-06,
+      "loss": 1.0152,
+      "step": 6109
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477855769451486e-06,
+      "loss": 0.9198,
+      "step": 6110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474746862335044e-06,
+      "loss": 0.858,
+      "step": 6111
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.471638344248763e-06,
+      "loss": 1.2865,
+      "step": 6112
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468530215535684e-06,
+      "loss": 1.1835,
+      "step": 6113
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4654224765388e-06,
+      "loss": 1.1264,
+      "step": 6114
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462315127601072e-06,
+      "loss": 0.6461,
+      "step": 6115
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459208169065413e-06,
+      "loss": 1.0321,
+      "step": 6116
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456101601274688e-06,
+      "loss": 0.9694,
+      "step": 6117
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452995424571724e-06,
+      "loss": 0.8671,
+      "step": 6118
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.449889639299305e-06,
+      "loss": 1.0302,
+      "step": 6119
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4467842458001685e-06,
+      "loss": 0.8231,
+      "step": 6120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.443679244417014e-06,
+      "loss": 0.8507,
+      "step": 6121
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.440574635492494e-06,
+      "loss": 1.2191,
+      "step": 6122
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437470419369221e-06,
+      "loss": 0.936,
+      "step": 6123
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434366596389754e-06,
+      "loss": 0.8577,
+      "step": 6124
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431263166896617e-06,
+      "loss": 0.9716,
+      "step": 6125
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4281601312323e-06,
+      "loss": 1.1427,
+      "step": 6126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4250574897392285e-06,
+      "loss": 0.7834,
+      "step": 6127
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.421955242759801e-06,
+      "loss": 0.7869,
+      "step": 6128
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.418853390636363e-06,
+      "loss": 0.8041,
+      "step": 6129
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.415751933711225e-06,
+      "loss": 0.9246,
+      "step": 6130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.41265087232664e-06,
+      "loss": 0.8395,
+      "step": 6131
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409550206824836e-06,
+      "loss": 0.7603,
+      "step": 6132
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.406449937547986e-06,
+      "loss": 1.1113,
+      "step": 6133
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4033500648382175e-06,
+      "loss": 0.9284,
+      "step": 6134
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400250589037619e-06,
+      "loss": 0.9755,
+      "step": 6135
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397151510488231e-06,
+      "loss": 0.8038,
+      "step": 6136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3940528295320606e-06,
+      "loss": 0.9322,
+      "step": 6137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390954546511057e-06,
+      "loss": 0.899,
+      "step": 6138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3878566617671345e-06,
+      "loss": 1.0426,
+      "step": 6139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384759175642162e-06,
+      "loss": 0.8871,
+      "step": 6140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.38166208847796e-06,
+      "loss": 0.7426,
+      "step": 6141
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.378565400616306e-06,
+      "loss": 1.004,
+      "step": 6142
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3754691123989466e-06,
+      "loss": 0.8547,
+      "step": 6143
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372373224167565e-06,
+      "loss": 1.1188,
+      "step": 6144
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369277736263808e-06,
+      "loss": 0.9749,
+      "step": 6145
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366182649029283e-06,
+      "loss": 1.0843,
+      "step": 6146
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3630879628055456e-06,
+      "loss": 1.0907,
+      "step": 6147
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359993677934113e-06,
+      "loss": 0.7348,
+      "step": 6148
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356899794756457e-06,
+      "loss": 0.9079,
+      "step": 6149
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353806313614002e-06,
+      "loss": 0.8811,
+      "step": 6150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350713234848129e-06,
+      "loss": 0.8849,
+      "step": 6151
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.347620558800176e-06,
+      "loss": 0.7898,
+      "step": 6152
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344528285811433e-06,
+      "loss": 0.9722,
+      "step": 6153
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.341436416223154e-06,
+      "loss": 0.9754,
+      "step": 6154
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338344950376542e-06,
+      "loss": 1.1976,
+      "step": 6155
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.335253888612756e-06,
+      "loss": 0.9958,
+      "step": 6156
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332163231272911e-06,
+      "loss": 1.2623,
+      "step": 6157
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329072978698074e-06,
+      "loss": 0.9101,
+      "step": 6158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.32598313122927e-06,
+      "loss": 0.6909,
+      "step": 6159
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.322893689207488e-06,
+      "loss": 0.4989,
+      "step": 6160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.319804652973657e-06,
+      "loss": 0.8944,
+      "step": 6161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3167160228686695e-06,
+      "loss": 1.0472,
+      "step": 6162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.313627799233374e-06,
+      "loss": 1.0893,
+      "step": 6163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3105399824085676e-06,
+      "loss": 0.7627,
+      "step": 6164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307452572735014e-06,
+      "loss": 1.2717,
+      "step": 6165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304365570553421e-06,
+      "loss": 0.7339,
+      "step": 6166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.301278976204457e-06,
+      "loss": 0.9801,
+      "step": 6167
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.298192790028742e-06,
+      "loss": 1.3271,
+      "step": 6168
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295107012366852e-06,
+      "loss": 1.1898,
+      "step": 6169
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29202164355932e-06,
+      "loss": 0.7593,
+      "step": 6170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.288936683946634e-06,
+      "loss": 0.7451,
+      "step": 6171
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285852133869235e-06,
+      "loss": 0.9344,
+      "step": 6172
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2827679936675175e-06,
+      "loss": 1.0429,
+      "step": 6173
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.279684263681836e-06,
+      "loss": 1.0723,
+      "step": 6174
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.27660094425249e-06,
+      "loss": 0.9614,
+      "step": 6175
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.273518035719746e-06,
+      "loss": 0.9579,
+      "step": 6176
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270435538423821e-06,
+      "loss": 0.7738,
+      "step": 6177
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267353452704876e-06,
+      "loss": 0.7716,
+      "step": 6178
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2642717789030415e-06,
+      "loss": 0.7815,
+      "step": 6179
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.261190517358395e-06,
+      "loss": 0.9915,
+      "step": 6180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258109668410968e-06,
+      "loss": 0.7297,
+      "step": 6181
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.255029232400751e-06,
+      "loss": 0.7266,
+      "step": 6182
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251949209667687e-06,
+      "loss": 1.1337,
+      "step": 6183
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.24886960055167e-06,
+      "loss": 1.0263,
+      "step": 6184
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245790405392553e-06,
+      "loss": 1.001,
+      "step": 6185
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242711624530137e-06,
+      "loss": 0.9683,
+      "step": 6186
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2396332583041895e-06,
+      "loss": 0.8497,
+      "step": 6187
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236555307054417e-06,
+      "loss": 0.8779,
+      "step": 6188
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233477771120492e-06,
+      "loss": 0.8502,
+      "step": 6189
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.230400650842035e-06,
+      "loss": 1.1731,
+      "step": 6190
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227323946558621e-06,
+      "loss": 0.9539,
+      "step": 6191
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.224247658609781e-06,
+      "loss": 1.1385,
+      "step": 6192
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.221171787335002e-06,
+      "loss": 1.2439,
+      "step": 6193
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.218096333073721e-06,
+      "loss": 0.861,
+      "step": 6194
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.215021296165332e-06,
+      "loss": 0.9779,
+      "step": 6195
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2119466769491784e-06,
+      "loss": 1.0725,
+      "step": 6196
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20887247576456e-06,
+      "loss": 0.6601,
+      "step": 6197
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.205798692950735e-06,
+      "loss": 0.8854,
+      "step": 6198
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.202725328846909e-06,
+      "loss": 0.8228,
+      "step": 6199
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199652383792245e-06,
+      "loss": 1.0236,
+      "step": 6200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196579858125858e-06,
+      "loss": 1.0106,
+      "step": 6201
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1935077521868184e-06,
+      "loss": 0.8344,
+      "step": 6202
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19043606631414e-06,
+      "loss": 0.73,
+      "step": 6203
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187364800846815e-06,
+      "loss": 1.4254,
+      "step": 6204
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184293956123766e-06,
+      "loss": 1.2862,
+      "step": 6205
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181223532483876e-06,
+      "loss": 2.3913,
+      "step": 6206
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178153530265982e-06,
+      "loss": 1.1186,
+      "step": 6207
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1750839498088755e-06,
+      "loss": 0.9315,
+      "step": 6208
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172014791451302e-06,
+      "loss": 1.1537,
+      "step": 6209
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.168946055531959e-06,
+      "loss": 0.9238,
+      "step": 6210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165877742389497e-06,
+      "loss": 1.3463,
+      "step": 6211
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162809852362523e-06,
+      "loss": 1.0136,
+      "step": 6212
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1597423857895884e-06,
+      "loss": 1.169,
+      "step": 6213
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156675343009206e-06,
+      "loss": 0.9418,
+      "step": 6214
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153608724359849e-06,
+      "loss": 1.059,
+      "step": 6215
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.150542530179924e-06,
+      "loss": 0.8041,
+      "step": 6216
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147476760807806e-06,
+      "loss": 0.8915,
+      "step": 6217
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.144411416581819e-06,
+      "loss": 0.9344,
+      "step": 6218
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141346497840242e-06,
+      "loss": 1.0393,
+      "step": 6219
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1382820049212946e-06,
+      "loss": 1.0117,
+      "step": 6220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1352179381631705e-06,
+      "loss": 1.1107,
+      "step": 6221
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132154297904005e-06,
+      "loss": 0.8348,
+      "step": 6222
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129091084481882e-06,
+      "loss": 0.908,
+      "step": 6223
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.126028298234844e-06,
+      "loss": 0.7833,
+      "step": 6224
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.122965939500884e-06,
+      "loss": 1.4758,
+      "step": 6225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.119904008617956e-06,
+      "loss": 0.9785,
+      "step": 6226
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.116842505923955e-06,
+      "loss": 1.2122,
+      "step": 6227
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.113781431756736e-06,
+      "loss": 0.7702,
+      "step": 6228
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110720786454104e-06,
+      "loss": 0.8613,
+      "step": 6229
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107660570353815e-06,
+      "loss": 1.0397,
+      "step": 6230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1046007837935775e-06,
+      "loss": 0.821,
+      "step": 6231
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101541427111066e-06,
+      "loss": 0.9025,
+      "step": 6232
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098482500643887e-06,
+      "loss": 0.8836,
+      "step": 6233
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095424004729611e-06,
+      "loss": 1.0019,
+      "step": 6234
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092365939705761e-06,
+      "loss": 0.8555,
+      "step": 6235
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.089308305909807e-06,
+      "loss": 0.8277,
+      "step": 6236
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08625110367918e-06,
+      "loss": 0.9572,
+      "step": 6237
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.083194333351252e-06,
+      "loss": 0.9748,
+      "step": 6238
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080137995263362e-06,
+      "loss": 1.0928,
+      "step": 6239
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.077082089752786e-06,
+      "loss": 1.3137,
+      "step": 6240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0740266171567605e-06,
+      "loss": 1.0871,
+      "step": 6241
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.070971577812472e-06,
+      "loss": 0.9715,
+      "step": 6242
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.067916972057063e-06,
+      "loss": 0.8766,
+      "step": 6243
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064862800227625e-06,
+      "loss": 1.0856,
+      "step": 6244
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.061809062661199e-06,
+      "loss": 1.1245,
+      "step": 6245
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058755759694787e-06,
+      "loss": 1.0074,
+      "step": 6246
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055702891665325e-06,
+      "loss": 0.983,
+      "step": 6247
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052650458909725e-06,
+      "loss": 0.9504,
+      "step": 6248
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0495984617648384e-06,
+      "loss": 1.1092,
+      "step": 6249
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046546900567463e-06,
+      "loss": 0.8533,
+      "step": 6250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043495775654355e-06,
+      "loss": 0.9478,
+      "step": 6251
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0404450873622256e-06,
+      "loss": 0.9149,
+      "step": 6252
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0373948360277305e-06,
+      "loss": 1.1089,
+      "step": 6253
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.034345021987484e-06,
+      "loss": 1.1323,
+      "step": 6254
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0312956455780495e-06,
+      "loss": 0.879,
+      "step": 6255
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.028246707135941e-06,
+      "loss": 0.8955,
+      "step": 6256
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.025198206997622e-06,
+      "loss": 1.0368,
+      "step": 6257
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.02215014549951e-06,
+      "loss": 0.7973,
+      "step": 6258
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.019102522977982e-06,
+      "loss": 1.111,
+      "step": 6259
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.016055339769353e-06,
+      "loss": 0.7853,
+      "step": 6260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0130085962098974e-06,
+      "loss": 1.0745,
+      "step": 6261
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009962292635838e-06,
+      "loss": 1.0383,
+      "step": 6262
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00691642938335e-06,
+      "loss": 0.8835,
+      "step": 6263
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.003871006788562e-06,
+      "loss": 0.9581,
+      "step": 6264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000826025187552e-06,
+      "loss": 1.3948,
+      "step": 6265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.997781484916352e-06,
+      "loss": 0.89,
+      "step": 6266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9947373863109384e-06,
+      "loss": 0.9758,
+      "step": 6267
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9916937297072464e-06,
+      "loss": 0.862,
+      "step": 6268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988650515441157e-06,
+      "loss": 0.9991,
+      "step": 6269
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985607743848507e-06,
+      "loss": 1.0258,
+      "step": 6270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982565415265083e-06,
+      "loss": 0.8351,
+      "step": 6271
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9795235300266185e-06,
+      "loss": 1.0336,
+      "step": 6272
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9764820884688045e-06,
+      "loss": 0.9904,
+      "step": 6273
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973441090927283e-06,
+      "loss": 0.9396,
+      "step": 6274
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.970400537737632e-06,
+      "loss": 1.0476,
+      "step": 6275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967360429235407e-06,
+      "loss": 0.8883,
+      "step": 6276
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964320765756092e-06,
+      "loss": 1.0835,
+      "step": 6277
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961281547635132e-06,
+      "loss": 1.1108,
+      "step": 6278
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958242775207918e-06,
+      "loss": 1.0486,
+      "step": 6279
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955204448809797e-06,
+      "loss": 1.0872,
+      "step": 6280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952166568776062e-06,
+      "loss": 0.8024,
+      "step": 6281
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949129135441962e-06,
+      "loss": 0.965,
+      "step": 6282
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.946092149142693e-06,
+      "loss": 1.1174,
+      "step": 6283
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9430556102134045e-06,
+      "loss": 1.0836,
+      "step": 6284
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9400195189891885e-06,
+      "loss": 0.8689,
+      "step": 6285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936983875805093e-06,
+      "loss": 1.0741,
+      "step": 6286
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933948680996126e-06,
+      "loss": 0.7016,
+      "step": 6287
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930913934897231e-06,
+      "loss": 0.9787,
+      "step": 6288
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.927879637843308e-06,
+      "loss": 0.9588,
+      "step": 6289
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.924845790169209e-06,
+      "loss": 0.8141,
+      "step": 6290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.921812392209738e-06,
+      "loss": 0.8699,
+      "step": 6291
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.918779444299636e-06,
+      "loss": 0.8738,
+      "step": 6292
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.915746946773616e-06,
+      "loss": 0.9365,
+      "step": 6293
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9127148999663275e-06,
+      "loss": 0.9933,
+      "step": 6294
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9096833042123706e-06,
+      "loss": 0.8482,
+      "step": 6295
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906652159846296e-06,
+      "loss": 0.9853,
+      "step": 6296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.903621467202608e-06,
+      "loss": 1.1847,
+      "step": 6297
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900591226615762e-06,
+      "loss": 0.8453,
+      "step": 6298
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897561438420158e-06,
+      "loss": 0.6268,
+      "step": 6299
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89453210295015e-06,
+      "loss": 0.8912,
+      "step": 6300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891503220540045e-06,
+      "loss": 0.7653,
+      "step": 6301
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.888474791524088e-06,
+      "loss": 0.9133,
+      "step": 6302
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885446816236484e-06,
+      "loss": 0.7022,
+      "step": 6303
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.88241929501139e-06,
+      "loss": 0.9894,
+      "step": 6304
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879392228182908e-06,
+      "loss": 0.8813,
+      "step": 6305
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.876365616085089e-06,
+      "loss": 0.9569,
+      "step": 6306
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873339459051934e-06,
+      "loss": 1.4099,
+      "step": 6307
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8703137574173985e-06,
+      "loss": 1.1557,
+      "step": 6308
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867288511515383e-06,
+      "loss": 0.9386,
+      "step": 6309
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.864263721679741e-06,
+      "loss": 0.7724,
+      "step": 6310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861239388244273e-06,
+      "loss": 0.8184,
+      "step": 6311
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.858215511542729e-06,
+      "loss": 0.9866,
+      "step": 6312
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85519209190881e-06,
+      "loss": 1.0484,
+      "step": 6313
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852169129676164e-06,
+      "loss": 0.9027,
+      "step": 6314
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849146625178395e-06,
+      "loss": 1.0495,
+      "step": 6315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8461245787490505e-06,
+      "loss": 1.161,
+      "step": 6316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843102990721629e-06,
+      "loss": 1.1697,
+      "step": 6317
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840081861429581e-06,
+      "loss": 1.4069,
+      "step": 6318
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.837061191206296e-06,
+      "loss": 1.2648,
+      "step": 6319
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834040980385129e-06,
+      "loss": 1.0868,
+      "step": 6320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.831021229299374e-06,
+      "loss": 0.9782,
+      "step": 6321
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.828001938282275e-06,
+      "loss": 0.9989,
+      "step": 6322
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.824983107667028e-06,
+      "loss": 0.8497,
+      "step": 6323
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.821964737786776e-06,
+      "loss": 0.9584,
+      "step": 6324
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8189468289746075e-06,
+      "loss": 0.8101,
+      "step": 6325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.815929381563576e-06,
+      "loss": 0.9561,
+      "step": 6326
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.812912395886664e-06,
+      "loss": 0.9273,
+      "step": 6327
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809895872276813e-06,
+      "loss": 1.3449,
+      "step": 6328
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806879811066912e-06,
+      "loss": 1.011,
+      "step": 6329
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803864212589796e-06,
+      "loss": 1.1147,
+      "step": 6330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8008490771782586e-06,
+      "loss": 1.0982,
+      "step": 6331
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797834405165035e-06,
+      "loss": 1.0261,
+      "step": 6332
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794820196882811e-06,
+      "loss": 1.0512,
+      "step": 6333
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.791806452664215e-06,
+      "loss": 0.8874,
+      "step": 6334
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788793172841829e-06,
+      "loss": 0.8809,
+      "step": 6335
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785780357748185e-06,
+      "loss": 1.2169,
+      "step": 6336
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782768007715769e-06,
+      "loss": 0.7028,
+      "step": 6337
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779756123077005e-06,
+      "loss": 0.8663,
+      "step": 6338
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776744704164272e-06,
+      "loss": 1.1773,
+      "step": 6339
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773733751309898e-06,
+      "loss": 0.7318,
+      "step": 6340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77072326484615e-06,
+      "loss": 1.1921,
+      "step": 6341
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.767713245105251e-06,
+      "loss": 1.0002,
+      "step": 6342
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76470369241938e-06,
+      "loss": 0.777,
+      "step": 6343
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761694607120655e-06,
+      "loss": 0.8024,
+      "step": 6344
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758685989541142e-06,
+      "loss": 1.0461,
+      "step": 6345
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755677840012859e-06,
+      "loss": 0.9242,
+      "step": 6346
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75267015886777e-06,
+      "loss": 1.0974,
+      "step": 6347
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.749662946437789e-06,
+      "loss": 0.9856,
+      "step": 6348
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746656203054779e-06,
+      "loss": 0.9786,
+      "step": 6349
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743649929050547e-06,
+      "loss": 1.0624,
+      "step": 6350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740644124756854e-06,
+      "loss": 0.9561,
+      "step": 6351
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.737638790505403e-06,
+      "loss": 0.9527,
+      "step": 6352
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73463392662785e-06,
+      "loss": 1.0128,
+      "step": 6353
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7316295334558e-06,
+      "loss": 0.8732,
+      "step": 6354
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7286256113208005e-06,
+      "loss": 0.8665,
+      "step": 6355
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7256221605543515e-06,
+      "loss": 1.0733,
+      "step": 6356
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.722619181487897e-06,
+      "loss": 1.0958,
+      "step": 6357
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.71961667445283e-06,
+      "loss": 1.2182,
+      "step": 6358
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716614639780501e-06,
+      "loss": 0.7799,
+      "step": 6359
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713613077802199e-06,
+      "loss": 1.0288,
+      "step": 6360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7106119888491555e-06,
+      "loss": 0.7098,
+      "step": 6361
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.707611373252557e-06,
+      "loss": 0.8141,
+      "step": 6362
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704611231343542e-06,
+      "loss": 0.9016,
+      "step": 6363
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.701611563453184e-06,
+      "loss": 1.0118,
+      "step": 6364
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698612369912524e-06,
+      "loss": 0.7877,
+      "step": 6365
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695613651052529e-06,
+      "loss": 0.9919,
+      "step": 6366
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69261540720413e-06,
+      "loss": 0.987,
+      "step": 6367
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689617638698197e-06,
+      "loss": 1.034,
+      "step": 6368
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.686620345865541e-06,
+      "loss": 1.0446,
+      "step": 6369
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683623529036941e-06,
+      "loss": 0.9262,
+      "step": 6370
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.680627188543105e-06,
+      "loss": 1.5387,
+      "step": 6371
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677631324714696e-06,
+      "loss": 0.787,
+      "step": 6372
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674635937882325e-06,
+      "loss": 1.0556,
+      "step": 6373
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671641028376547e-06,
+      "loss": 1.0436,
+      "step": 6374
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668646596527866e-06,
+      "loss": 1.0082,
+      "step": 6375
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.665652642666733e-06,
+      "loss": 0.788,
+      "step": 6376
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662659167123547e-06,
+      "loss": 0.925,
+      "step": 6377
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659666170228655e-06,
+      "loss": 0.9266,
+      "step": 6378
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.656673652312348e-06,
+      "loss": 1.1574,
+      "step": 6379
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.653681613704866e-06,
+      "loss": 0.8069,
+      "step": 6380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6506900547363966e-06,
+      "loss": 0.9065,
+      "step": 6381
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647698975737075e-06,
+      "loss": 0.9848,
+      "step": 6382
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.64470837703698e-06,
+      "loss": 1.0391,
+      "step": 6383
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.641718258966142e-06,
+      "loss": 0.9143,
+      "step": 6384
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638728621854537e-06,
+      "loss": 1.0034,
+      "step": 6385
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6357394660320804e-06,
+      "loss": 0.9844,
+      "step": 6386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.632750791828656e-06,
+      "loss": 0.6997,
+      "step": 6387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.629762599574065e-06,
+      "loss": 1.222,
+      "step": 6388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6267748895980755e-06,
+      "loss": 0.6566,
+      "step": 6389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.623787662230398e-06,
+      "loss": 0.7364,
+      "step": 6390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.620800917800682e-06,
+      "loss": 0.9229,
+      "step": 6391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61781465663854e-06,
+      "loss": 1.1349,
+      "step": 6392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.614828879073518e-06,
+      "loss": 1.6985,
+      "step": 6393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611843585435111e-06,
+      "loss": 1.0069,
+      "step": 6394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608858776052766e-06,
+      "loss": 0.8258,
+      "step": 6395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.605874451255866e-06,
+      "loss": 1.1357,
+      "step": 6396
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.602890611373746e-06,
+      "loss": 0.9134,
+      "step": 6397
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599907256735695e-06,
+      "loss": 0.7718,
+      "step": 6398
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596924387670939e-06,
+      "loss": 1.049,
+      "step": 6399
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.593942004508652e-06,
+      "loss": 0.9455,
+      "step": 6400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590960107577958e-06,
+      "loss": 0.7513,
+      "step": 6401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.587978697207928e-06,
+      "loss": 0.7923,
+      "step": 6402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.584997773727564e-06,
+      "loss": 0.7352,
+      "step": 6403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582017337465837e-06,
+      "loss": 1.1041,
+      "step": 6404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579037388751654e-06,
+      "loss": 1.1998,
+      "step": 6405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576057927913862e-06,
+      "loss": 1.0993,
+      "step": 6406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573078955281266e-06,
+      "loss": 0.5462,
+      "step": 6407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570100471182607e-06,
+      "loss": 0.7319,
+      "step": 6408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567122475946578e-06,
+      "loss": 1.0413,
+      "step": 6409
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564144969901815e-06,
+      "loss": 0.986,
+      "step": 6410
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561167953376905e-06,
+      "loss": 0.9165,
+      "step": 6411
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558191426700375e-06,
+      "loss": 1.191,
+      "step": 6412
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555215390200699e-06,
+      "loss": 0.961,
+      "step": 6413
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552239844206295e-06,
+      "loss": 0.7136,
+      "step": 6414
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549264789045543e-06,
+      "loss": 0.8317,
+      "step": 6415
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5462902250467445e-06,
+      "loss": 0.8667,
+      "step": 6416
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543316152538161e-06,
+      "loss": 1.4958,
+      "step": 6417
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.540342571847997e-06,
+      "loss": 0.7865,
+      "step": 6418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5373694833044e-06,
+      "loss": 0.9248,
+      "step": 6419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.534396887235471e-06,
+      "loss": 0.9999,
+      "step": 6420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.531424783969253e-06,
+      "loss": 1.0572,
+      "step": 6421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.52845317383373e-06,
+      "loss": 0.9595,
+      "step": 6422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.525482057156833e-06,
+      "loss": 0.9074,
+      "step": 6423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.52251143426644e-06,
+      "loss": 0.7853,
+      "step": 6424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5195413054903724e-06,
+      "loss": 0.6969,
+      "step": 6425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516571671156407e-06,
+      "loss": 0.8896,
+      "step": 6426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.513602531592257e-06,
+      "loss": 1.2809,
+      "step": 6427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.51063388712558e-06,
+      "loss": 0.9558,
+      "step": 6428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.507665738083986e-06,
+      "loss": 0.8116,
+      "step": 6429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5046980847950135e-06,
+      "loss": 1.0673,
+      "step": 6430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501730927586171e-06,
+      "loss": 1.0785,
+      "step": 6431
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4987642667848975e-06,
+      "loss": 0.9935,
+      "step": 6432
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4957981027185745e-06,
+      "loss": 1.2407,
+      "step": 6433
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49283243571454e-06,
+      "loss": 0.6908,
+      "step": 6434
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489867266100067e-06,
+      "loss": 0.958,
+      "step": 6435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486902594202379e-06,
+      "loss": 0.8746,
+      "step": 6436
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483938420348642e-06,
+      "loss": 1.2768,
+      "step": 6437
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48097474486597e-06,
+      "loss": 0.769,
+      "step": 6438
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478011568081417e-06,
+      "loss": 0.9951,
+      "step": 6439
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475048890321988e-06,
+      "loss": 1.2546,
+      "step": 6440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472086711914626e-06,
+      "loss": 0.8558,
+      "step": 6441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469125033186231e-06,
+      "loss": 0.7265,
+      "step": 6442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4661638544636335e-06,
+      "loss": 0.793,
+      "step": 6443
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463203176073615e-06,
+      "loss": 0.8067,
+      "step": 6444
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460242998342904e-06,
+      "loss": 1.7543,
+      "step": 6445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457283321598169e-06,
+      "loss": 0.6422,
+      "step": 6446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.454324146166026e-06,
+      "loss": 1.16,
+      "step": 6447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451365472373038e-06,
+      "loss": 1.2235,
+      "step": 6448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448407300545715e-06,
+      "loss": 0.8748,
+      "step": 6449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445449631010498e-06,
+      "loss": 0.7906,
+      "step": 6450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442492464093782e-06,
+      "loss": 0.7379,
+      "step": 6451
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.439535800121907e-06,
+      "loss": 0.7429,
+      "step": 6452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.43657963942116e-06,
+      "loss": 0.7733,
+      "step": 6453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.433623982317766e-06,
+      "loss": 0.8787,
+      "step": 6454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.430668829137899e-06,
+      "loss": 0.8624,
+      "step": 6455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.427714180207677e-06,
+      "loss": 1.0768,
+      "step": 6456
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.424760035853156e-06,
+      "loss": 1.0307,
+      "step": 6457
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4218063964003385e-06,
+      "loss": 0.9392,
+      "step": 6458
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418853262175183e-06,
+      "loss": 0.7827,
+      "step": 6459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41590063350358e-06,
+      "loss": 1.0483,
+      "step": 6460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4129485107113665e-06,
+      "loss": 0.832,
+      "step": 6461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409996894124327e-06,
+      "loss": 0.7752,
+      "step": 6462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4070457840681854e-06,
+      "loss": 1.0597,
+      "step": 6463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4040951808686115e-06,
+      "loss": 0.9727,
+      "step": 6464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401145084851223e-06,
+      "loss": 1.0485,
+      "step": 6465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398195496341576e-06,
+      "loss": 0.8857,
+      "step": 6466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.395246415665174e-06,
+      "loss": 1.048,
+      "step": 6467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3922978431474624e-06,
+      "loss": 1.0514,
+      "step": 6468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389349779113834e-06,
+      "loss": 0.8795,
+      "step": 6469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386402223889621e-06,
+      "loss": 0.7949,
+      "step": 6470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.383455177800102e-06,
+      "loss": 1.1037,
+      "step": 6471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380508641170499e-06,
+      "loss": 0.9183,
+      "step": 6472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.377562614325978e-06,
+      "loss": 1.0192,
+      "step": 6473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37461709759165e-06,
+      "loss": 0.9212,
+      "step": 6474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.371672091292561e-06,
+      "loss": 0.7664,
+      "step": 6475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368727595753722e-06,
+      "loss": 1.315,
+      "step": 6476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365783611300062e-06,
+      "loss": 0.9638,
+      "step": 6477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.362840138256467e-06,
+      "loss": 0.9658,
+      "step": 6478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3598971769477655e-06,
+      "loss": 1.0055,
+      "step": 6479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3569547276987254e-06,
+      "loss": 0.9159,
+      "step": 6480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354012790834069e-06,
+      "loss": 1.0058,
+      "step": 6481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3510713666784505e-06,
+      "loss": 0.8603,
+      "step": 6482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3481304555564716e-06,
+      "loss": 0.9075,
+      "step": 6483
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.345190057792682e-06,
+      "loss": 1.0684,
+      "step": 6484
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.34225017371156e-06,
+      "loss": 0.9833,
+      "step": 6485
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.339310803637538e-06,
+      "loss": 1.0817,
+      "step": 6486
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.336371947895001e-06,
+      "loss": 0.9782,
+      "step": 6487
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.333433606808261e-06,
+      "loss": 1.5135,
+      "step": 6488
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3304957807015775e-06,
+      "loss": 1.0389,
+      "step": 6489
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.327558469899159e-06,
+      "loss": 1.0016,
+      "step": 6490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.324621674725151e-06,
+      "loss": 1.1036,
+      "step": 6491
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321685395503644e-06,
+      "loss": 0.9063,
+      "step": 6492
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318749632558674e-06,
+      "loss": 0.799,
+      "step": 6493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315814386214215e-06,
+      "loss": 0.9302,
+      "step": 6494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3128796567941886e-06,
+      "loss": 1.1283,
+      "step": 6495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309945444622457e-06,
+      "loss": 0.8571,
+      "step": 6496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307011750022826e-06,
+      "loss": 1.2499,
+      "step": 6497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3040785733190424e-06,
+      "loss": 1.0121,
+      "step": 6498
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301145914834801e-06,
+      "loss": 0.8095,
+      "step": 6499
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298213774893735e-06,
+      "loss": 1.2421,
+      "step": 6500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295282153819419e-06,
+      "loss": 0.8034,
+      "step": 6501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.292351051935372e-06,
+      "loss": 0.5627,
+      "step": 6502
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.289420469565061e-06,
+      "loss": 0.4914,
+      "step": 6503
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286490407031895e-06,
+      "loss": 0.4387,
+      "step": 6504
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283560864659211e-06,
+      "loss": 0.4906,
+      "step": 6505
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2806318427703045e-06,
+      "loss": 0.7366,
+      "step": 6506
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277703341688407e-06,
+      "loss": 0.6652,
+      "step": 6507
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274775361736692e-06,
+      "loss": 0.576,
+      "step": 6508
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2718479032382855e-06,
+      "loss": 0.6289,
+      "step": 6509
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.268920966516241e-06,
+      "loss": 0.5855,
+      "step": 6510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2659945518935676e-06,
+      "loss": 0.5312,
+      "step": 6511
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263068659693202e-06,
+      "loss": 0.6913,
+      "step": 6512
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260143290238033e-06,
+      "loss": 0.6021,
+      "step": 6513
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257218443850896e-06,
+      "loss": 0.6179,
+      "step": 6514
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.254294120854562e-06,
+      "loss": 0.5881,
+      "step": 6515
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.251370321571742e-06,
+      "loss": 0.6105,
+      "step": 6516
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.248447046325095e-06,
+      "loss": 0.4855,
+      "step": 6517
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.245524295437223e-06,
+      "loss": 0.6471,
+      "step": 6518
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.242602069230655e-06,
+      "loss": 0.4283,
+      "step": 6519
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.239680368027885e-06,
+      "loss": 0.6806,
+      "step": 6520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.236759192151336e-06,
+      "loss": 0.5665,
+      "step": 6521
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.233838541923374e-06,
+      "loss": 0.5038,
+      "step": 6522
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.230918417666309e-06,
+      "loss": 0.5994,
+      "step": 6523
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227998819702389e-06,
+      "loss": 0.4756,
+      "step": 6524
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22507974835381e-06,
+      "loss": 0.4909,
+      "step": 6525
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222161203942706e-06,
+      "loss": 0.5383,
+      "step": 6526
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219243186791154e-06,
+      "loss": 0.6114,
+      "step": 6527
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216325697221172e-06,
+      "loss": 0.5997,
+      "step": 6528
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213408735554719e-06,
+      "loss": 0.6589,
+      "step": 6529
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2104923021136944e-06,
+      "loss": 0.5372,
+      "step": 6530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2075763972199556e-06,
+      "loss": 0.7592,
+      "step": 6531
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.204661021195272e-06,
+      "loss": 0.6191,
+      "step": 6532
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201746174361379e-06,
+      "loss": 0.5909,
+      "step": 6533
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.198831857039941e-06,
+      "loss": 0.8543,
+      "step": 6534
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.195918069552572e-06,
+      "loss": 0.5936,
+      "step": 6535
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193004812220817e-06,
+      "loss": 0.4485,
+      "step": 6536
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.190092085366179e-06,
+      "loss": 0.5523,
+      "step": 6537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18717988931009e-06,
+      "loss": 0.5705,
+      "step": 6538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1842682243739206e-06,
+      "loss": 0.6358,
+      "step": 6539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181357090878993e-06,
+      "loss": 1.0253,
+      "step": 6540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.17844648914656e-06,
+      "loss": 0.4996,
+      "step": 6541
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175536419497831e-06,
+      "loss": 0.6272,
+      "step": 6542
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1726268822539415e-06,
+      "loss": 0.621,
+      "step": 6543
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169717877735976e-06,
+      "loss": 0.642,
+      "step": 6544
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166809406264961e-06,
+      "loss": 0.6125,
+      "step": 6545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163901468161856e-06,
+      "loss": 0.5464,
+      "step": 6546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.160994063747566e-06,
+      "loss": 0.607,
+      "step": 6547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158087193342944e-06,
+      "loss": 0.5898,
+      "step": 6548
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155180857268779e-06,
+      "loss": 0.6618,
+      "step": 6549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.152275055845797e-06,
+      "loss": 0.5185,
+      "step": 6550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.149369789394668e-06,
+      "loss": 0.4184,
+      "step": 6551
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.146465058236005e-06,
+      "loss": 0.537,
+      "step": 6552
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.143560862690361e-06,
+      "loss": 0.8453,
+      "step": 6553
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.140657203078227e-06,
+      "loss": 0.6865,
+      "step": 6554
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.137754079720038e-06,
+      "loss": 0.63,
+      "step": 6555
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.134851492936172e-06,
+      "loss": 0.6234,
+      "step": 6556
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131949443046938e-06,
+      "loss": 0.7582,
+      "step": 6557
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.129047930372598e-06,
+      "loss": 0.7634,
+      "step": 6558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1261469552333475e-06,
+      "loss": 0.7088,
+      "step": 6559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123246517949323e-06,
+      "loss": 0.5887,
+      "step": 6560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120346618840605e-06,
+      "loss": 0.7451,
+      "step": 6561
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11744725822721e-06,
+      "loss": 0.5113,
+      "step": 6562
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114548436429097e-06,
+      "loss": 0.5518,
+      "step": 6563
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111650153766172e-06,
+      "loss": 0.5581,
+      "step": 6564
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108752410558276e-06,
+      "loss": 0.8298,
+      "step": 6565
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.105855207125183e-06,
+      "loss": 0.5701,
+      "step": 6566
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102958543786617e-06,
+      "loss": 0.6722,
+      "step": 6567
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100062420862241e-06,
+      "loss": 0.5709,
+      "step": 6568
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0971668386716525e-06,
+      "loss": 0.3983,
+      "step": 6569
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094271797534404e-06,
+      "loss": 0.5407,
+      "step": 6570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.091377297769973e-06,
+      "loss": 0.4282,
+      "step": 6571
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088483339697783e-06,
+      "loss": 0.819,
+      "step": 6572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0855899236372e-06,
+      "loss": 0.4683,
+      "step": 6573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.082697049907522e-06,
+      "loss": 0.5802,
+      "step": 6574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079804718827992e-06,
+      "loss": 0.7398,
+      "step": 6575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076912930717803e-06,
+      "loss": 0.5011,
+      "step": 6576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074021685896073e-06,
+      "loss": 0.3869,
+      "step": 6577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071130984681866e-06,
+      "loss": 0.5668,
+      "step": 6578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068240827394186e-06,
+      "loss": 0.7183,
+      "step": 6579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065351214351979e-06,
+      "loss": 0.5278,
+      "step": 6580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0624621458741274e-06,
+      "loss": 0.7086,
+      "step": 6581
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0595736222794565e-06,
+      "loss": 0.642,
+      "step": 6582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.056685643886727e-06,
+      "loss": 0.4967,
+      "step": 6583
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.053798211014646e-06,
+      "loss": 0.5135,
+      "step": 6584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.050911323981855e-06,
+      "loss": 0.4947,
+      "step": 6585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0480249831069365e-06,
+      "loss": 0.8704,
+      "step": 6586
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.045139188708415e-06,
+      "loss": 0.4762,
+      "step": 6587
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.042253941104751e-06,
+      "loss": 0.6059,
+      "step": 6588
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.039369240614349e-06,
+      "loss": 0.5403,
+      "step": 6589
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.036485087555549e-06,
+      "loss": 0.7439,
+      "step": 6590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0336014822466285e-06,
+      "loss": 0.7072,
+      "step": 6591
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0307184250058185e-06,
+      "loss": 0.5306,
+      "step": 6592
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0278359161512765e-06,
+      "loss": 0.6177,
+      "step": 6593
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024953956001097e-06,
+      "loss": 0.5559,
+      "step": 6594
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022072544873322e-06,
+      "loss": 0.6622,
+      "step": 6595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019191683085931e-06,
+      "loss": 0.643,
+      "step": 6596
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016311370956839e-06,
+      "loss": 0.5735,
+      "step": 6597
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013431608803909e-06,
+      "loss": 0.5046,
+      "step": 6598
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010552396944937e-06,
+      "loss": 0.6147,
+      "step": 6599
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007673735697659e-06,
+      "loss": 0.6793,
+      "step": 6600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.004795625379746e-06,
+      "loss": 0.4248,
+      "step": 6601
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.001918066308811e-06,
+      "loss": 0.769,
+      "step": 6602
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.999041058802416e-06,
+      "loss": 0.6402,
+      "step": 6603
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996164603178051e-06,
+      "loss": 0.7208,
+      "step": 6604
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993288699753146e-06,
+      "loss": 0.4775,
+      "step": 6605
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.990413348845072e-06,
+      "loss": 0.5484,
+      "step": 6606
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987538550771145e-06,
+      "loss": 0.3323,
+      "step": 6607
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984664305848601e-06,
+      "loss": 0.8648,
+      "step": 6608
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.981790614394641e-06,
+      "loss": 0.6811,
+      "step": 6609
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978917476726385e-06,
+      "loss": 0.7583,
+      "step": 6610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9760448931609e-06,
+      "loss": 0.4865,
+      "step": 6611
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973172864015193e-06,
+      "loss": 0.5788,
+      "step": 6612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.970301389606206e-06,
+      "loss": 0.7529,
+      "step": 6613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.96743047025082e-06,
+      "loss": 0.4586,
+      "step": 6614
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.964560106265858e-06,
+      "loss": 1.011,
+      "step": 6615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.961690297968078e-06,
+      "loss": 0.6142,
+      "step": 6616
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.958821045674179e-06,
+      "loss": 0.5082,
+      "step": 6617
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.955952349700799e-06,
+      "loss": 0.5831,
+      "step": 6618
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.953084210364508e-06,
+      "loss": 0.6184,
+      "step": 6619
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.950216627981833e-06,
+      "loss": 0.742,
+      "step": 6620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947349602869214e-06,
+      "loss": 0.5824,
+      "step": 6621
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9444831353430485e-06,
+      "loss": 0.693,
+      "step": 6622
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941617225719665e-06,
+      "loss": 0.5642,
+      "step": 6623
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938751874315326e-06,
+      "loss": 0.926,
+      "step": 6624
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935887081446249e-06,
+      "loss": 0.6048,
+      "step": 6625
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933022847428571e-06,
+      "loss": 0.5711,
+      "step": 6626
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930159172578384e-06,
+      "loss": 0.6549,
+      "step": 6627
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927296057211697e-06,
+      "loss": 0.6553,
+      "step": 6628
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.924433501644477e-06,
+      "loss": 0.6063,
+      "step": 6629
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.921571506192616e-06,
+      "loss": 0.5889,
+      "step": 6630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91871007117196e-06,
+      "loss": 0.6771,
+      "step": 6631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915849196898278e-06,
+      "loss": 0.6955,
+      "step": 6632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912988883687283e-06,
+      "loss": 0.4462,
+      "step": 6633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91012913185463e-06,
+      "loss": 0.6774,
+      "step": 6634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9072699417158976e-06,
+      "loss": 0.6611,
+      "step": 6635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904411313586614e-06,
+      "loss": 0.5432,
+      "step": 6636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.901553247782254e-06,
+      "loss": 0.4697,
+      "step": 6637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898695744618211e-06,
+      "loss": 0.5947,
+      "step": 6638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89583880440983e-06,
+      "loss": 0.4956,
+      "step": 6639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.892982427472387e-06,
+      "loss": 0.9688,
+      "step": 6640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890126614121098e-06,
+      "loss": 0.7494,
+      "step": 6641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.887271364671119e-06,
+      "loss": 0.3911,
+      "step": 6642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884416679437539e-06,
+      "loss": 0.8916,
+      "step": 6643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88156255873539e-06,
+      "loss": 0.5692,
+      "step": 6644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878709002879638e-06,
+      "loss": 0.5411,
+      "step": 6645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875856012185187e-06,
+      "loss": 0.6379,
+      "step": 6646
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.873003586966881e-06,
+      "loss": 0.9338,
+      "step": 6647
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8701517275395e-06,
+      "loss": 0.6036,
+      "step": 6648
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.867300434217761e-06,
+      "loss": 0.5067,
+      "step": 6649
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.864449707316319e-06,
+      "loss": 0.5902,
+      "step": 6650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.861599547149767e-06,
+      "loss": 0.4311,
+      "step": 6651
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.858749954032631e-06,
+      "loss": 0.8908,
+      "step": 6652
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8559009282793866e-06,
+      "loss": 0.7667,
+      "step": 6653
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853052470204438e-06,
+      "loss": 0.6582,
+      "step": 6654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.850204580122121e-06,
+      "loss": 0.7296,
+      "step": 6655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847357258346719e-06,
+      "loss": 0.6419,
+      "step": 6656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844510505192447e-06,
+      "loss": 0.9603,
+      "step": 6657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.841664320973457e-06,
+      "loss": 0.8114,
+      "step": 6658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838818706003847e-06,
+      "loss": 0.5632,
+      "step": 6659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835973660597644e-06,
+      "loss": 0.6013,
+      "step": 6660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83312918506881e-06,
+      "loss": 0.4908,
+      "step": 6661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830285279731255e-06,
+      "loss": 0.5775,
+      "step": 6662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8274419448988056e-06,
+      "loss": 0.534,
+      "step": 6663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82459918088525e-06,
+      "loss": 0.6083,
+      "step": 6664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821756988004299e-06,
+      "loss": 0.5293,
+      "step": 6665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818915366569603e-06,
+      "loss": 0.6416,
+      "step": 6666
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.81607431689475e-06,
+      "loss": 0.5374,
+      "step": 6667
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813233839293265e-06,
+      "loss": 0.3206,
+      "step": 6668
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.81039393407861e-06,
+      "loss": 0.4134,
+      "step": 6669
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807554601564183e-06,
+      "loss": 0.5169,
+      "step": 6670
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.804715842063318e-06,
+      "loss": 0.6436,
+      "step": 6671
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801877655889288e-06,
+      "loss": 0.5051,
+      "step": 6672
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799040043355302e-06,
+      "loss": 0.5337,
+      "step": 6673
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796203004774506e-06,
+      "loss": 0.6033,
+      "step": 6674
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.79336654045998e-06,
+      "loss": 0.48,
+      "step": 6675
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790530650724745e-06,
+      "loss": 0.3673,
+      "step": 6676
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.787695335881755e-06,
+      "loss": 0.4554,
+      "step": 6677
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.784860596243901e-06,
+      "loss": 0.6291,
+      "step": 6678
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7820264321240135e-06,
+      "loss": 0.6084,
+      "step": 6679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.779192843834852e-06,
+      "loss": 0.6931,
+      "step": 6680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.77635983168913e-06,
+      "loss": 0.5991,
+      "step": 6681
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.773527395999474e-06,
+      "loss": 0.5492,
+      "step": 6682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.770695537078462e-06,
+      "loss": 0.5157,
+      "step": 6683
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7678642552386026e-06,
+      "loss": 0.6306,
+      "step": 6684
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765033550792341e-06,
+      "loss": 0.5017,
+      "step": 6685
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762203424052067e-06,
+      "loss": 0.6146,
+      "step": 6686
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759373875330095e-06,
+      "loss": 0.7386,
+      "step": 6687
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.756544904938682e-06,
+      "loss": 0.6163,
+      "step": 6688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753716513190024e-06,
+      "loss": 0.4608,
+      "step": 6689
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.750888700396241e-06,
+      "loss": 0.4826,
+      "step": 6690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7480614668693956e-06,
+      "loss": 0.5356,
+      "step": 6691
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745234812921497e-06,
+      "loss": 0.556,
+      "step": 6692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742408738864476e-06,
+      "loss": 0.7142,
+      "step": 6693
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7395832450102055e-06,
+      "loss": 0.582,
+      "step": 6694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736758331670493e-06,
+      "loss": 0.4679,
+      "step": 6695
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733933999157087e-06,
+      "loss": 0.5745,
+      "step": 6696
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731110247781656e-06,
+      "loss": 0.7933,
+      "step": 6697
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728287077855827e-06,
+      "loss": 0.4674,
+      "step": 6698
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7254644896911485e-06,
+      "loss": 0.6639,
+      "step": 6699
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722642483599106e-06,
+      "loss": 0.6917,
+      "step": 6700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.719821059891124e-06,
+      "loss": 0.4909,
+      "step": 6701
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717000218878562e-06,
+      "loss": 0.5676,
+      "step": 6702
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.714179960872712e-06,
+      "loss": 0.3966,
+      "step": 6703
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711360286184807e-06,
+      "loss": 0.9419,
+      "step": 6704
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708541195126012e-06,
+      "loss": 0.4683,
+      "step": 6705
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.705722688007428e-06,
+      "loss": 0.5969,
+      "step": 6706
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.702904765140093e-06,
+      "loss": 0.7098,
+      "step": 6707
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700087426834975e-06,
+      "loss": 0.3715,
+      "step": 6708
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697270673402993e-06,
+      "loss": 0.7313,
+      "step": 6709
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.694454505154981e-06,
+      "loss": 0.6809,
+      "step": 6710
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6916389224017186e-06,
+      "loss": 0.5696,
+      "step": 6711
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.688823925453923e-06,
+      "loss": 0.6152,
+      "step": 6712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.686009514622238e-06,
+      "loss": 0.5484,
+      "step": 6713
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6831956902172594e-06,
+      "loss": 0.7193,
+      "step": 6714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6803824525495e-06,
+      "loss": 0.5817,
+      "step": 6715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.67756980192942e-06,
+      "loss": 0.5839,
+      "step": 6716
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.674757738667405e-06,
+      "loss": 0.3136,
+      "step": 6717
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.671946263073781e-06,
+      "loss": 0.5542,
+      "step": 6718
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669135375458807e-06,
+      "loss": 0.6057,
+      "step": 6719
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.666325076132688e-06,
+      "loss": 0.5704,
+      "step": 6720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663515365405551e-06,
+      "loss": 0.7093,
+      "step": 6721
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6607062435874605e-06,
+      "loss": 0.7427,
+      "step": 6722
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6578977109884225e-06,
+      "loss": 0.6577,
+      "step": 6723
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655089767918363e-06,
+      "loss": 0.6012,
+      "step": 6724
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652282414687163e-06,
+      "loss": 0.417,
+      "step": 6725
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6494756516046265e-06,
+      "loss": 0.6653,
+      "step": 6726
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6466694789804935e-06,
+      "loss": 0.6109,
+      "step": 6727
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64386389712444e-06,
+      "loss": 0.6507,
+      "step": 6728
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641058906346076e-06,
+      "loss": 0.5433,
+      "step": 6729
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6382545069549475e-06,
+      "loss": 0.445,
+      "step": 6730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635450699260535e-06,
+      "loss": 0.5213,
+      "step": 6731
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6326474835722545e-06,
+      "loss": 0.7946,
+      "step": 6732
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629844860199453e-06,
+      "loss": 0.9043,
+      "step": 6733
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.627042829451416e-06,
+      "loss": 0.497,
+      "step": 6734
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624241391637359e-06,
+      "loss": 0.6212,
+      "step": 6735
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621440547066445e-06,
+      "loss": 0.4756,
+      "step": 6736
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618640296047751e-06,
+      "loss": 0.7229,
+      "step": 6737
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6158406388903055e-06,
+      "loss": 0.6739,
+      "step": 6738
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613041575903062e-06,
+      "loss": 0.7406,
+      "step": 6739
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.610243107394915e-06,
+      "loss": 0.5891,
+      "step": 6740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607445233674683e-06,
+      "loss": 0.6676,
+      "step": 6741
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6046479550511366e-06,
+      "loss": 0.6059,
+      "step": 6742
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.601851271832969e-06,
+      "loss": 0.6196,
+      "step": 6743
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5990551843288e-06,
+      "loss": 0.714,
+      "step": 6744
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.596259692847198e-06,
+      "loss": 0.5215,
+      "step": 6745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.593464797696656e-06,
+      "loss": 0.528,
+      "step": 6746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5906704991856135e-06,
+      "loss": 0.4854,
+      "step": 6747
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5878767976224305e-06,
+      "loss": 0.6993,
+      "step": 6748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5850836933154075e-06,
+      "loss": 0.5362,
+      "step": 6749
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582291186572778e-06,
+      "loss": 0.9146,
+      "step": 6750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.579499277702714e-06,
+      "loss": 0.4755,
+      "step": 6751
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.576707967013306e-06,
+      "loss": 0.8153,
+      "step": 6752
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.573917254812601e-06,
+      "loss": 0.7354,
+      "step": 6753
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571127141408565e-06,
+      "loss": 0.3597,
+      "step": 6754
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5683376271091006e-06,
+      "loss": 0.7323,
+      "step": 6755
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565548712222046e-06,
+      "loss": 0.7354,
+      "step": 6756
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562760397055172e-06,
+      "loss": 0.5208,
+      "step": 6757
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559972681916186e-06,
+      "loss": 0.6719,
+      "step": 6758
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557185567112725e-06,
+      "loss": 0.7643,
+      "step": 6759
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554399052952362e-06,
+      "loss": 0.5373,
+      "step": 6760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551613139742603e-06,
+      "loss": 0.9925,
+      "step": 6761
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.548827827790889e-06,
+      "loss": 0.5753,
+      "step": 6762
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.546043117404595e-06,
+      "loss": 0.3534,
+      "step": 6763
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.543259008891026e-06,
+      "loss": 0.7126,
+      "step": 6764
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.540475502557423e-06,
+      "loss": 0.5902,
+      "step": 6765
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.537692598710962e-06,
+      "loss": 0.7316,
+      "step": 6766
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53491029765875e-06,
+      "loss": 0.5772,
+      "step": 6767
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.532128599707829e-06,
+      "loss": 0.508,
+      "step": 6768
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52934750516517e-06,
+      "loss": 0.6714,
+      "step": 6769
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5265670143376925e-06,
+      "loss": 0.5372,
+      "step": 6770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.523787127532229e-06,
+      "loss": 0.4366,
+      "step": 6771
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521007845055555e-06,
+      "loss": 0.7295,
+      "step": 6772
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.51822916721438e-06,
+      "loss": 0.6355,
+      "step": 6773
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.515451094315343e-06,
+      "loss": 0.5484,
+      "step": 6774
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.512673626665025e-06,
+      "loss": 0.5109,
+      "step": 6775
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509896764569932e-06,
+      "loss": 0.4964,
+      "step": 6776
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.507120508336504e-06,
+      "loss": 0.7521,
+      "step": 6777
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.50434485827112e-06,
+      "loss": 0.6991,
+      "step": 6778
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.501569814680079e-06,
+      "loss": 0.6095,
+      "step": 6779
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.498795377869622e-06,
+      "loss": 0.4817,
+      "step": 6780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4960215481459315e-06,
+      "loss": 0.7236,
+      "step": 6781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.493248325815107e-06,
+      "loss": 0.4779,
+      "step": 6782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.490475711183192e-06,
+      "loss": 0.5218,
+      "step": 6783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.487703704556158e-06,
+      "loss": 0.6314,
+      "step": 6784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484932306239909e-06,
+      "loss": 0.8381,
+      "step": 6785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.482161516540283e-06,
+      "loss": 0.6825,
+      "step": 6786
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479391335763053e-06,
+      "loss": 0.5329,
+      "step": 6787
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4766217642139235e-06,
+      "loss": 0.8684,
+      "step": 6788
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4738528021985294e-06,
+      "loss": 0.5879,
+      "step": 6789
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47108445002244e-06,
+      "loss": 0.6534,
+      "step": 6790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46831670799116e-06,
+      "loss": 0.6604,
+      "step": 6791
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465549576410121e-06,
+      "loss": 0.6705,
+      "step": 6792
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.462783055584693e-06,
+      "loss": 0.493,
+      "step": 6793
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460017145820174e-06,
+      "loss": 0.5056,
+      "step": 6794
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457251847421799e-06,
+      "loss": 0.5288,
+      "step": 6795
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454487160694727e-06,
+      "loss": 0.5657,
+      "step": 6796
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.451723085944064e-06,
+      "loss": 0.8241,
+      "step": 6797
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4489596234748414e-06,
+      "loss": 0.4466,
+      "step": 6798
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446196773592013e-06,
+      "loss": 0.4272,
+      "step": 6799
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443434536600478e-06,
+      "loss": 0.9598,
+      "step": 6800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440672912805063e-06,
+      "loss": 0.5416,
+      "step": 6801
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437911902510526e-06,
+      "loss": 0.4298,
+      "step": 6802
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.435151506021563e-06,
+      "loss": 0.5192,
+      "step": 6803
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432391723642797e-06,
+      "loss": 0.6424,
+      "step": 6804
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429632555678788e-06,
+      "loss": 0.6353,
+      "step": 6805
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4268740024340166e-06,
+      "loss": 0.5323,
+      "step": 6806
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424116064212903e-06,
+      "loss": 0.3094,
+      "step": 6807
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421358741319809e-06,
+      "loss": 0.7546,
+      "step": 6808
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.418602034059015e-06,
+      "loss": 0.5447,
+      "step": 6809
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4158459427347385e-06,
+      "loss": 0.8047,
+      "step": 6810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.413090467651129e-06,
+      "loss": 0.6879,
+      "step": 6811
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.41033560911227e-06,
+      "loss": 0.3948,
+      "step": 6812
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.407581367422164e-06,
+      "loss": 0.4617,
+      "step": 6813
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.404827742884769e-06,
+      "loss": 0.6912,
+      "step": 6814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402074735803955e-06,
+      "loss": 0.4684,
+      "step": 6815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399322346483532e-06,
+      "loss": 0.6224,
+      "step": 6816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.396570575227243e-06,
+      "loss": 0.6182,
+      "step": 6817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.393819422338756e-06,
+      "loss": 0.7226,
+      "step": 6818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391068888121679e-06,
+      "loss": 0.8709,
+      "step": 6819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388318972879545e-06,
+      "loss": 0.6205,
+      "step": 6820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.385569676915824e-06,
+      "loss": 0.4897,
+      "step": 6821
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382821000533913e-06,
+      "loss": 0.7706,
+      "step": 6822
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380072944037145e-06,
+      "loss": 0.5063,
+      "step": 6823
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377325507728777e-06,
+      "loss": 0.5665,
+      "step": 6824
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374578691912015e-06,
+      "loss": 0.658,
+      "step": 6825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.371832496889974e-06,
+      "loss": 0.7872,
+      "step": 6826
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369086922965713e-06,
+      "loss": 0.5452,
+      "step": 6827
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.36634197044222e-06,
+      "loss": 0.5738,
+      "step": 6828
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363597639622419e-06,
+      "loss": 0.475,
+      "step": 6829
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360853930809153e-06,
+      "loss": 0.4954,
+      "step": 6830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3581108443052145e-06,
+      "loss": 0.5029,
+      "step": 6831
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.355368380413316e-06,
+      "loss": 0.777,
+      "step": 6832
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.352626539436098e-06,
+      "loss": 0.7038,
+      "step": 6833
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.349885321676137e-06,
+      "loss": 0.5257,
+      "step": 6834
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.34714472743594e-06,
+      "loss": 0.6644,
+      "step": 6835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3444047570179525e-06,
+      "loss": 0.6788,
+      "step": 6836
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.34166541072454e-06,
+      "loss": 0.5095,
+      "step": 6837
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.338926688858006e-06,
+      "loss": 0.7214,
+      "step": 6838
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3361885917205835e-06,
+      "loss": 0.5989,
+      "step": 6839
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33345111961443e-06,
+      "loss": 0.6654,
+      "step": 6840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330714272841641e-06,
+      "loss": 0.5579,
+      "step": 6841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.327978051704247e-06,
+      "loss": 0.5727,
+      "step": 6842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.325242456504202e-06,
+      "loss": 0.4879,
+      "step": 6843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.322507487543392e-06,
+      "loss": 0.5524,
+      "step": 6844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.319773145123638e-06,
+      "loss": 0.6789,
+      "step": 6845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.317039429546687e-06,
+      "loss": 0.6778,
+      "step": 6846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3143063411142185e-06,
+      "loss": 0.4525,
+      "step": 6847
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.311573880127843e-06,
+      "loss": 0.4851,
+      "step": 6848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308842046889104e-06,
+      "loss": 0.7125,
+      "step": 6849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306110841699471e-06,
+      "loss": 0.6193,
+      "step": 6850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.303380264860348e-06,
+      "loss": 0.3829,
+      "step": 6851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300650316673069e-06,
+      "loss": 0.6262,
+      "step": 6852
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297920997438897e-06,
+      "loss": 0.5305,
+      "step": 6853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295192307459028e-06,
+      "loss": 0.571,
+      "step": 6854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292464247034586e-06,
+      "loss": 0.6013,
+      "step": 6855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.289736816466627e-06,
+      "loss": 0.5996,
+      "step": 6856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.287010016056134e-06,
+      "loss": 0.5695,
+      "step": 6857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.284283846104032e-06,
+      "loss": 0.2974,
+      "step": 6858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281558306911165e-06,
+      "loss": 0.8807,
+      "step": 6859
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278833398778306e-06,
+      "loss": 0.7091,
+      "step": 6860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276109122006166e-06,
+      "loss": 0.5579,
+      "step": 6861
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273385476895383e-06,
+      "loss": 0.5382,
+      "step": 6862
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.270662463746522e-06,
+      "loss": 0.5189,
+      "step": 6863
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267940082860088e-06,
+      "loss": 0.4213,
+      "step": 6864
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.265218334536508e-06,
+      "loss": 0.5367,
+      "step": 6865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26249721907614e-06,
+      "loss": 0.405,
+      "step": 6866
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2597767367792785e-06,
+      "loss": 0.54,
+      "step": 6867
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.25705688794613e-06,
+      "loss": 0.5154,
+      "step": 6868
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.254337672876857e-06,
+      "loss": 0.4542,
+      "step": 6869
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.251619091871534e-06,
+      "loss": 0.5204,
+      "step": 6870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2489011452301705e-06,
+      "loss": 0.6534,
+      "step": 6871
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246183833252707e-06,
+      "loss": 0.5604,
+      "step": 6872
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243467156239014e-06,
+      "loss": 0.6249,
+      "step": 6873
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240751114488889e-06,
+      "loss": 0.5699,
+      "step": 6874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.238035708302063e-06,
+      "loss": 0.5256,
+      "step": 6875
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.235320937978193e-06,
+      "loss": 0.7744,
+      "step": 6876
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.232606803816872e-06,
+      "loss": 0.5958,
+      "step": 6877
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2298933061176165e-06,
+      "loss": 0.584,
+      "step": 6878
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227180445179875e-06,
+      "loss": 0.52,
+      "step": 6879
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.224468221303027e-06,
+      "loss": 0.5887,
+      "step": 6880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2217566347863795e-06,
+      "loss": 0.4322,
+      "step": 6881
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219045685929171e-06,
+      "loss": 0.5053,
+      "step": 6882
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216335375030568e-06,
+      "loss": 0.6535,
+      "step": 6883
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213625702389669e-06,
+      "loss": 0.555,
+      "step": 6884
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210916668305496e-06,
+      "loss": 0.6696,
+      "step": 6885
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208208273077013e-06,
+      "loss": 0.5827,
+      "step": 6886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205500517003103e-06,
+      "loss": 0.7235,
+      "step": 6887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202793400382579e-06,
+      "loss": 0.5305,
+      "step": 6888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200086923514184e-06,
+      "loss": 0.5417,
+      "step": 6889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197381086696595e-06,
+      "loss": 0.5644,
+      "step": 6890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19467589022841e-06,
+      "loss": 0.6808,
+      "step": 6891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19197133440817e-06,
+      "loss": 0.6757,
+      "step": 6892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189267419534333e-06,
+      "loss": 0.5569,
+      "step": 6893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186564145905293e-06,
+      "loss": 0.5723,
+      "step": 6894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183861513819365e-06,
+      "loss": 0.5363,
+      "step": 6895
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181159523574796e-06,
+      "loss": 0.8743,
+      "step": 6896
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178458175469774e-06,
+      "loss": 0.3313,
+      "step": 6897
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1757574698024025e-06,
+      "loss": 0.3928,
+      "step": 6898
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17305740687072e-06,
+      "loss": 0.8409,
+      "step": 6899
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170357986972689e-06,
+      "loss": 0.6822,
+      "step": 6900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1676592104062116e-06,
+      "loss": 0.6023,
+      "step": 6901
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1649610774691005e-06,
+      "loss": 0.6831,
+      "step": 6902
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.16226358845912e-06,
+      "loss": 0.6815,
+      "step": 6903
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.159566743673946e-06,
+      "loss": 0.5767,
+      "step": 6904
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1568705434111925e-06,
+      "loss": 0.606,
+      "step": 6905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1541749879683975e-06,
+      "loss": 0.6158,
+      "step": 6906
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.151480077643031e-06,
+      "loss": 0.5727,
+      "step": 6907
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1487858127324884e-06,
+      "loss": 0.5204,
+      "step": 6908
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.146092193534098e-06,
+      "loss": 0.7644,
+      "step": 6909
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.143399220345114e-06,
+      "loss": 0.5916,
+      "step": 6910
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.14070689346272e-06,
+      "loss": 0.7393,
+      "step": 6911
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.138015213184029e-06,
+      "loss": 0.6767,
+      "step": 6912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.135324179806079e-06,
+      "loss": 0.7594,
+      "step": 6913
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1326337936258465e-06,
+      "loss": 0.8247,
+      "step": 6914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1299440549402235e-06,
+      "loss": 0.6241,
+      "step": 6915
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127254964046039e-06,
+      "loss": 0.7,
+      "step": 6916
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124566521240048e-06,
+      "loss": 0.3955,
+      "step": 6917
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12187872681893e-06,
+      "loss": 0.5965,
+      "step": 6918
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1191915810793045e-06,
+      "loss": 0.601,
+      "step": 6919
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11650508431771e-06,
+      "loss": 0.4424,
+      "step": 6920
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113819236830618e-06,
+      "loss": 0.7064,
+      "step": 6921
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111134038914418e-06,
+      "loss": 0.5281,
+      "step": 6922
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108449490865441e-06,
+      "loss": 0.7398,
+      "step": 6923
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105765592979935e-06,
+      "loss": 0.5162,
+      "step": 6924
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103082345554092e-06,
+      "loss": 0.561,
+      "step": 6925
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100399748884018e-06,
+      "loss": 0.6881,
+      "step": 6926
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.097717803265751e-06,
+      "loss": 0.6695,
+      "step": 6927
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095036508995261e-06,
+      "loss": 0.7168,
+      "step": 6928
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092355866368433e-06,
+      "loss": 0.6883,
+      "step": 6929
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0896758756811e-06,
+      "loss": 0.4548,
+      "step": 6930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08699653722901e-06,
+      "loss": 0.6281,
+      "step": 6931
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.084317851307841e-06,
+      "loss": 0.8582,
+      "step": 6932
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081639818213202e-06,
+      "loss": 0.7338,
+      "step": 6933
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0789624382406265e-06,
+      "loss": 0.6657,
+      "step": 6934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076285711685578e-06,
+      "loss": 0.4349,
+      "step": 6935
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0736096388434465e-06,
+      "loss": 0.438,
+      "step": 6936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0709342200095515e-06,
+      "loss": 0.5748,
+      "step": 6937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068259455479139e-06,
+      "loss": 0.9355,
+      "step": 6938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065585345547383e-06,
+      "loss": 0.7261,
+      "step": 6939
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.062911890509388e-06,
+      "loss": 0.444,
+      "step": 6940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.060239090660181e-06,
+      "loss": 0.4518,
+      "step": 6941
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.05756694629472e-06,
+      "loss": 0.6014,
+      "step": 6942
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.054895457707891e-06,
+      "loss": 0.687,
+      "step": 6943
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.052224625194505e-06,
+      "loss": 0.6961,
+      "step": 6944
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0495544490493045e-06,
+      "loss": 0.6294,
+      "step": 6945
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046884929566952e-06,
+      "loss": 0.7246,
+      "step": 6946
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.044216067042053e-06,
+      "loss": 0.6516,
+      "step": 6947
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.041547861769126e-06,
+      "loss": 0.7125,
+      "step": 6948
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038880314042619e-06,
+      "loss": 0.5774,
+      "step": 6949
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.03621342415691e-06,
+      "loss": 0.5687,
+      "step": 6950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033547192406306e-06,
+      "loss": 0.5052,
+      "step": 6951
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0308816190850355e-06,
+      "loss": 0.4473,
+      "step": 6952
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028216704487265e-06,
+      "loss": 0.5186,
+      "step": 6953
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025552448907081e-06,
+      "loss": 0.6505,
+      "step": 6954
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022888852638493e-06,
+      "loss": 0.3328,
+      "step": 6955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020225915975452e-06,
+      "loss": 0.4861,
+      "step": 6956
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017563639211813e-06,
+      "loss": 0.404,
+      "step": 6957
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014902022641383e-06,
+      "loss": 0.4925,
+      "step": 6958
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012241066557883e-06,
+      "loss": 0.545,
+      "step": 6959
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009580771254963e-06,
+      "loss": 0.6429,
+      "step": 6960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.006921137026199e-06,
+      "loss": 0.6327,
+      "step": 6961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004262164165098e-06,
+      "loss": 0.5939,
+      "step": 6962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001603852965091e-06,
+      "loss": 0.5002,
+      "step": 6963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998946203719535e-06,
+      "loss": 0.5928,
+      "step": 6964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996289216721717e-06,
+      "loss": 0.6631,
+      "step": 6965
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.993632892264849e-06,
+      "loss": 0.7538,
+      "step": 6966
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99097723064207e-06,
+      "loss": 0.4818,
+      "step": 6967
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9883222321464475e-06,
+      "loss": 0.6342,
+      "step": 6968
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985667897070974e-06,
+      "loss": 0.8722,
+      "step": 6969
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983014225708567e-06,
+      "loss": 0.7913,
+      "step": 6970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.980361218352077e-06,
+      "loss": 0.8194,
+      "step": 6971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.977708875294274e-06,
+      "loss": 0.7058,
+      "step": 6972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.975057196827858e-06,
+      "loss": 0.6499,
+      "step": 6973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.972406183245455e-06,
+      "loss": 0.8256,
+      "step": 6974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9697558348396224e-06,
+      "loss": 0.5403,
+      "step": 6975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.967106151902842e-06,
+      "loss": 0.4289,
+      "step": 6976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.964457134727512e-06,
+      "loss": 0.8173,
+      "step": 6977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.961808783605968e-06,
+      "loss": 0.5653,
+      "step": 6978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959161098830469e-06,
+      "loss": 0.6,
+      "step": 6979
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9565140806932045e-06,
+      "loss": 0.649,
+      "step": 6980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953867729486285e-06,
+      "loss": 0.8386,
+      "step": 6981
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951222045501749e-06,
+      "loss": 0.6439,
+      "step": 6982
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948577029031566e-06,
+      "loss": 0.5774,
+      "step": 6983
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945932680367619e-06,
+      "loss": 0.5211,
+      "step": 6984
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9432889998017264e-06,
+      "loss": 0.7805,
+      "step": 6985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940645987625639e-06,
+      "loss": 0.6103,
+      "step": 6986
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.938003644131024e-06,
+      "loss": 0.5953,
+      "step": 6987
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935361969609476e-06,
+      "loss": 0.5441,
+      "step": 6988
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93272096435252e-06,
+      "loss": 0.797,
+      "step": 6989
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930080628651607e-06,
+      "loss": 0.6888,
+      "step": 6990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927440962798102e-06,
+      "loss": 0.4528,
+      "step": 6991
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924801967083316e-06,
+      "loss": 0.3721,
+      "step": 6992
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9221636417984734e-06,
+      "loss": 0.684,
+      "step": 6993
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.919525987234725e-06,
+      "loss": 0.5861,
+      "step": 6994
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.916889003683153e-06,
+      "loss": 0.4526,
+      "step": 6995
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.914252691434761e-06,
+      "loss": 0.5748,
+      "step": 6996
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911617050780479e-06,
+      "loss": 0.5503,
+      "step": 6997
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908982082011163e-06,
+      "loss": 0.5251,
+      "step": 6998
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906347785417599e-06,
+      "loss": 0.588,
+      "step": 6999
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903714161290491e-06,
+      "loss": 0.5,
+      "step": 7000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901081209920477e-06,
+      "loss": 0.4061,
+      "step": 7001
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898448931598113e-06,
+      "loss": 0.8251,
+      "step": 7002
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8958173266138935e-06,
+      "loss": 0.6875,
+      "step": 7003
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893186395258221e-06,
+      "loss": 0.5188,
+      "step": 7004
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.890556137821433e-06,
+      "loss": 0.5994,
+      "step": 7005
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.887926554593796e-06,
+      "loss": 0.5587,
+      "step": 7006
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.885297645865494e-06,
+      "loss": 0.5612,
+      "step": 7007
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.882669411926645e-06,
+      "loss": 0.7215,
+      "step": 7008
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.880041853067289e-06,
+      "loss": 0.6799,
+      "step": 7009
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877414969577391e-06,
+      "loss": 0.7135,
+      "step": 7010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.874788761746836e-06,
+      "loss": 0.5555,
+      "step": 7011
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8721632298654435e-06,
+      "loss": 0.5735,
+      "step": 7012
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8695383742229506e-06,
+      "loss": 0.7622,
+      "step": 7013
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86691419510903e-06,
+      "loss": 0.5076,
+      "step": 7014
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864290692813273e-06,
+      "loss": 0.4904,
+      "step": 7015
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8616678676251936e-06,
+      "loss": 0.6033,
+      "step": 7016
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85904571983424e-06,
+      "loss": 0.8945,
+      "step": 7017
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.856424249729769e-06,
+      "loss": 0.4677,
+      "step": 7018
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853803457601083e-06,
+      "loss": 0.5121,
+      "step": 7019
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851183343737397e-06,
+      "loss": 0.5296,
+      "step": 7020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848563908427855e-06,
+      "loss": 0.8023,
+      "step": 7021
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8459451519615246e-06,
+      "loss": 0.5475,
+      "step": 7022
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8433270746274e-06,
+      "loss": 0.6217,
+      "step": 7023
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840709676714401e-06,
+      "loss": 0.5837,
+      "step": 7024
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838092958511367e-06,
+      "loss": 0.7415,
+      "step": 7025
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835476920307071e-06,
+      "loss": 0.5074,
+      "step": 7026
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.832861562390205e-06,
+      "loss": 0.5383,
+      "step": 7027
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830246885049385e-06,
+      "loss": 0.7378,
+      "step": 7028
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827632888573155e-06,
+      "loss": 0.624,
+      "step": 7029
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825019573249992e-06,
+      "loss": 0.4391,
+      "step": 7030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822406939368277e-06,
+      "loss": 0.8798,
+      "step": 7031
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819794987216333e-06,
+      "loss": 0.5732,
+      "step": 7032
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817183717082403e-06,
+      "loss": 0.505,
+      "step": 7033
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.814573129254654e-06,
+      "loss": 0.4551,
+      "step": 7034
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8119632240211734e-06,
+      "loss": 0.8285,
+      "step": 7035
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.809354001669986e-06,
+      "loss": 0.5705,
+      "step": 7036
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8067454624890322e-06,
+      "loss": 0.6051,
+      "step": 7037
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8041376067661718e-06,
+      "loss": 0.6435,
+      "step": 7038
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.801530434789199e-06,
+      "loss": 0.3707,
+      "step": 7039
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.798923946845825e-06,
+      "loss": 0.542,
+      "step": 7040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.796318143223696e-06,
+      "loss": 0.3566,
+      "step": 7041
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7937130242103738e-06,
+      "loss": 0.4393,
+      "step": 7042
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7911085900933465e-06,
+      "loss": 0.4769,
+      "step": 7043
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7885048411600256e-06,
+      "loss": 0.6674,
+      "step": 7044
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7859017776977536e-06,
+      "loss": 0.4375,
+      "step": 7045
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7832993999937817e-06,
+      "loss": 0.5767,
+      "step": 7046
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7806977083353046e-06,
+      "loss": 0.715,
+      "step": 7047
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7780967030094316e-06,
+      "loss": 0.5158,
+      "step": 7048
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7754963843031956e-06,
+      "loss": 0.6803,
+      "step": 7049
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7728967525035564e-06,
+      "loss": 0.574,
+      "step": 7050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.770297807897396e-06,
+      "loss": 0.6377,
+      "step": 7051
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.767699550771522e-06,
+      "loss": 0.6145,
+      "step": 7052
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.9685,
+      "step": 7053
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7625051001074832e-06,
+      "loss": 0.4536,
+      "step": 7054
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7599089071425523e-06,
+      "loss": 0.559,
+      "step": 7055
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7573134028043778e-06,
+      "loss": 0.4965,
+      "step": 7056
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754718587379387e-06,
+      "loss": 0.5137,
+      "step": 7057
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.752124461153931e-06,
+      "loss": 0.6706,
+      "step": 7058
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7495310244142857e-06,
+      "loss": 0.3957,
+      "step": 7059
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.74693827744665e-06,
+      "loss": 0.5701,
+      "step": 7060
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.744346220537147e-06,
+      "loss": 0.5358,
+      "step": 7061
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741754853971825e-06,
+      "loss": 0.6451,
+      "step": 7062
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7391641780366507e-06,
+      "loss": 0.6975,
+      "step": 7063
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7365741930175283e-06,
+      "loss": 0.6845,
+      "step": 7064
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733984899200267e-06,
+      "loss": 0.4935,
+      "step": 7065
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7313962968706106e-06,
+      "loss": 0.6646,
+      "step": 7066
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7288083863142275e-06,
+      "loss": 0.4291,
+      "step": 7067
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7262211678167026e-06,
+      "loss": 0.6512,
+      "step": 7068
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.723634641663555e-06,
+      "loss": 0.6624,
+      "step": 7069
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7210488081402196e-06,
+      "loss": 0.6075,
+      "step": 7070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.718463667532055e-06,
+      "loss": 0.5461,
+      "step": 7071
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.71587922012435e-06,
+      "loss": 0.5916,
+      "step": 7072
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7132954662023045e-06,
+      "loss": 0.6748,
+      "step": 7073
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.710712406051049e-06,
+      "loss": 0.5946,
+      "step": 7074
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.708130039955644e-06,
+      "loss": 0.655,
+      "step": 7075
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.705548368201065e-06,
+      "loss": 0.4985,
+      "step": 7076
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702967391072213e-06,
+      "loss": 0.3219,
+      "step": 7077
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7003871088539113e-06,
+      "loss": 0.5826,
+      "step": 7078
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.697807521830907e-06,
+      "loss": 0.7867,
+      "step": 7079
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6952286302878736e-06,
+      "loss": 0.7086,
+      "step": 7080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6926504345094037e-06,
+      "loss": 0.3653,
+      "step": 7081
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6900729347800148e-06,
+      "loss": 0.5538,
+      "step": 7082
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6874961313841463e-06,
+      "loss": 0.6313,
+      "step": 7083
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6849200246061646e-06,
+      "loss": 0.6556,
+      "step": 7084
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6823446147303543e-06,
+      "loss": 0.6457,
+      "step": 7085
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.679769902040927e-06,
+      "loss": 0.4939,
+      "step": 7086
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.677195886822015e-06,
+      "loss": 0.4017,
+      "step": 7087
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.674622569357673e-06,
+      "loss": 0.5219,
+      "step": 7088
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6720499499318828e-06,
+      "loss": 0.5651,
+      "step": 7089
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6694780288285415e-06,
+      "loss": 0.5639,
+      "step": 7090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6669068063314807e-06,
+      "loss": 0.4609,
+      "step": 7091
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.664336282724449e-06,
+      "loss": 0.7237,
+      "step": 7092
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6617664582911095e-06,
+      "loss": 0.4689,
+      "step": 7093
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.65919733331506e-06,
+      "loss": 0.4584,
+      "step": 7094
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6566289080798156e-06,
+      "loss": 0.5469,
+      "step": 7095
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6540611828688143e-06,
+      "loss": 0.7209,
+      "step": 7096
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6514941579654227e-06,
+      "loss": 0.6421,
+      "step": 7097
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6489278336529243e-06,
+      "loss": 0.5195,
+      "step": 7098
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.646362210214527e-06,
+      "loss": 0.5235,
+      "step": 7099
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643797287933356e-06,
+      "loss": 0.5348,
+      "step": 7100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6412330670924636e-06,
+      "loss": 0.4712,
+      "step": 7101
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6386695479748313e-06,
+      "loss": 0.6392,
+      "step": 7102
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6361067308633534e-06,
+      "loss": 0.5665,
+      "step": 7103
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6335446160408516e-06,
+      "loss": 0.4731,
+      "step": 7104
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6309832037900672e-06,
+      "loss": 0.6556,
+      "step": 7105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6284224943936695e-06,
+      "loss": 0.4605,
+      "step": 7106
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6258624881342354e-06,
+      "loss": 0.5486,
+      "step": 7107
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6233031852942856e-06,
+      "loss": 0.659,
+      "step": 7108
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6207445861562497e-06,
+      "loss": 0.4596,
+      "step": 7109
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6181866910024823e-06,
+      "loss": 0.7223,
+      "step": 7110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6156295001152596e-06,
+      "loss": 0.8735,
+      "step": 7111
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.613073013776782e-06,
+      "loss": 0.534,
+      "step": 7112
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.610517232269171e-06,
+      "loss": 0.5274,
+      "step": 7113
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6079621558744692e-06,
+      "loss": 0.6705,
+      "step": 7114
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605407784874645e-06,
+      "loss": 0.4956,
+      "step": 7115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6028541195515853e-06,
+      "loss": 0.7137,
+      "step": 7116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6003011601871007e-06,
+      "loss": 0.458,
+      "step": 7117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59774890706292e-06,
+      "loss": 0.6562,
+      "step": 7118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.595197360460708e-06,
+      "loss": 0.587,
+      "step": 7119
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.592646520662031e-06,
+      "loss": 0.6965,
+      "step": 7120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.590096387948391e-06,
+      "loss": 0.5264,
+      "step": 7121
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.587546962601208e-06,
+      "loss": 0.6064,
+      "step": 7122
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.584998244901826e-06,
+      "loss": 0.651,
+      "step": 7123
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.582450235131505e-06,
+      "loss": 0.7302,
+      "step": 7124
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5799029335714374e-06,
+      "loss": 0.756,
+      "step": 7125
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.577356340502732e-06,
+      "loss": 0.4964,
+      "step": 7126
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5748104562064125e-06,
+      "loss": 0.7374,
+      "step": 7127
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5722652809634317e-06,
+      "loss": 0.5521,
+      "step": 7128
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5697208150546625e-06,
+      "loss": 0.533,
+      "step": 7129
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5671770587609044e-06,
+      "loss": 0.5546,
+      "step": 7130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.564634012362873e-06,
+      "loss": 0.4843,
+      "step": 7131
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.562091676141205e-06,
+      "loss": 0.689,
+      "step": 7132
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5595500503764626e-06,
+      "loss": 0.4334,
+      "step": 7133
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5570091353491277e-06,
+      "loss": 0.4872,
+      "step": 7134
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5544689313395963e-06,
+      "loss": 0.5265,
+      "step": 7135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.551929438628202e-06,
+      "loss": 0.434,
+      "step": 7136
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5493906574951887e-06,
+      "loss": 0.6112,
+      "step": 7137
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5468525882207226e-06,
+      "loss": 0.7024,
+      "step": 7138
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.544315231084895e-06,
+      "loss": 0.4518,
+      "step": 7139
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.541778586367716e-06,
+      "loss": 0.5006,
+      "step": 7140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5392426543491155e-06,
+      "loss": 0.5394,
+      "step": 7141
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5367074353089493e-06,
+      "loss": 0.6157,
+      "step": 7142
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534172929526992e-06,
+      "loss": 0.6121,
+      "step": 7143
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5316391372829384e-06,
+      "loss": 0.7448,
+      "step": 7144
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.529106058856406e-06,
+      "loss": 0.6787,
+      "step": 7145
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5265736945269344e-06,
+      "loss": 0.6144,
+      "step": 7146
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524042044573982e-06,
+      "loss": 0.4243,
+      "step": 7147
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5215111092769307e-06,
+      "loss": 0.6135,
+      "step": 7148
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5189808889150813e-06,
+      "loss": 0.5234,
+      "step": 7149
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516451383767657e-06,
+      "loss": 0.5897,
+      "step": 7150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5139225941138e-06,
+      "loss": 0.6432,
+      "step": 7151
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5113945202325817e-06,
+      "loss": 0.4504,
+      "step": 7152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.508867162402987e-06,
+      "loss": 0.5423,
+      "step": 7153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.506340520903917e-06,
+      "loss": 0.5367,
+      "step": 7154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5038145960142046e-06,
+      "loss": 0.5749,
+      "step": 7155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5012893880125965e-06,
+      "loss": 0.611,
+      "step": 7156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498764897177761e-06,
+      "loss": 0.8646,
+      "step": 7157
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4962411237882945e-06,
+      "loss": 0.6741,
+      "step": 7158
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4937180681227067e-06,
+      "loss": 0.641,
+      "step": 7159
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.491195730459428e-06,
+      "loss": 0.592,
+      "step": 7160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4886741110768174e-06,
+      "loss": 0.6803,
+      "step": 7161
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.486153210253137e-06,
+      "loss": 0.2545,
+      "step": 7162
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.483633028266592e-06,
+      "loss": 0.5486,
+      "step": 7163
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4811135653952933e-06,
+      "loss": 0.4574,
+      "step": 7164
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4785948219172793e-06,
+      "loss": 0.562,
+      "step": 7165
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.476076798110504e-06,
+      "loss": 0.4854,
+      "step": 7166
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.473559494252846e-06,
+      "loss": 0.5617,
+      "step": 7167
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4710429106221024e-06,
+      "loss": 0.4099,
+      "step": 7168
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.468527047495991e-06,
+      "loss": 0.5143,
+      "step": 7169
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.466011905152151e-06,
+      "loss": 0.4809,
+      "step": 7170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4634974838681425e-06,
+      "loss": 0.6794,
+      "step": 7171
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4609837839214432e-06,
+      "loss": 0.6161,
+      "step": 7172
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4584708055894543e-06,
+      "loss": 0.537,
+      "step": 7173
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455958549149496e-06,
+      "loss": 0.5326,
+      "step": 7174
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453447014878809e-06,
+      "loss": 0.479,
+      "step": 7175
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4509362030545536e-06,
+      "loss": 0.5951,
+      "step": 7176
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.448426113953812e-06,
+      "loss": 0.4599,
+      "step": 7177
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.445916747853585e-06,
+      "loss": 0.5656,
+      "step": 7178
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.44340810503079e-06,
+      "loss": 0.743,
+      "step": 7179
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.440900185762278e-06,
+      "loss": 0.4646,
+      "step": 7180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4383929903248094e-06,
+      "loss": 0.5074,
+      "step": 7181
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4358865189950596e-06,
+      "loss": 0.5023,
+      "step": 7182
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.433380772049635e-06,
+      "loss": 0.6798,
+      "step": 7183
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.430875749765058e-06,
+      "loss": 0.4736,
+      "step": 7184
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428371452417766e-06,
+      "loss": 0.6921,
+      "step": 7185
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4258678802841296e-06,
+      "loss": 0.6636,
+      "step": 7186
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4233650336404266e-06,
+      "loss": 0.7297,
+      "step": 7187
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4208629127628644e-06,
+      "loss": 0.5322,
+      "step": 7188
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4183615179275576e-06,
+      "loss": 0.6734,
+      "step": 7189
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.415860849410547e-06,
+      "loss": 0.8392,
+      "step": 7190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4133609074878016e-06,
+      "loss": 0.4186,
+      "step": 7191
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.410861692435201e-06,
+      "loss": 0.5889,
+      "step": 7192
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4083632045285462e-06,
+      "loss": 0.6423,
+      "step": 7193
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4058654440435577e-06,
+      "loss": 0.6052,
+      "step": 7194
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4033684112558808e-06,
+      "loss": 0.4317,
+      "step": 7195
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4008721064410655e-06,
+      "loss": 0.7183,
+      "step": 7196
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.398376529874602e-06,
+      "loss": 0.7489,
+      "step": 7197
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3958816818318875e-06,
+      "loss": 0.5064,
+      "step": 7198
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.393387562588242e-06,
+      "loss": 0.5248,
+      "step": 7199
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3908941724189027e-06,
+      "loss": 0.5982,
+      "step": 7200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3884015115990286e-06,
+      "loss": 0.8922,
+      "step": 7201
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3859095804037002e-06,
+      "loss": 0.5286,
+      "step": 7202
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.383418379107913e-06,
+      "loss": 0.5065,
+      "step": 7203
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3809279079865852e-06,
+      "loss": 0.5713,
+      "step": 7204
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3784381673145527e-06,
+      "loss": 0.5688,
+      "step": 7205
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3759491573665726e-06,
+      "loss": 0.6904,
+      "step": 7206
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.373460878417315e-06,
+      "loss": 0.9743,
+      "step": 7207
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.370973330741385e-06,
+      "loss": 0.5166,
+      "step": 7208
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3684865146132883e-06,
+      "loss": 0.7469,
+      "step": 7209
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366000430307459e-06,
+      "loss": 0.6412,
+      "step": 7210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.363515078098252e-06,
+      "loss": 0.7448,
+      "step": 7211
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.361030458259933e-06,
+      "loss": 0.8309,
+      "step": 7212
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.358546571066701e-06,
+      "loss": 0.4627,
+      "step": 7213
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.356063416792662e-06,
+      "loss": 0.7042,
+      "step": 7214
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.353580995711849e-06,
+      "loss": 0.6315,
+      "step": 7215
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3510993080982046e-06,
+      "loss": 0.7148,
+      "step": 7216
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348618354225597e-06,
+      "loss": 0.5355,
+      "step": 7217
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.34613813436781e-06,
+      "loss": 0.4287,
+      "step": 7218
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3436586487985558e-06,
+      "loss": 0.6197,
+      "step": 7219
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3411798977914554e-06,
+      "loss": 0.3446,
+      "step": 7220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.338701881620052e-06,
+      "loss": 0.7531,
+      "step": 7221
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3362246005578104e-06,
+      "loss": 0.5975,
+      "step": 7222
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.333748054878102e-06,
+      "loss": 0.6289,
+      "step": 7223
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.331272244854237e-06,
+      "loss": 0.6336,
+      "step": 7224
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.32879717075943e-06,
+      "loss": 0.6207,
+      "step": 7225
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3263228328668184e-06,
+      "loss": 0.654,
+      "step": 7226
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3238492314494588e-06,
+      "loss": 0.476,
+      "step": 7227
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3213763667803254e-06,
+      "loss": 0.8113,
+      "step": 7228
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3189042391323124e-06,
+      "loss": 0.5085,
+      "step": 7229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3164328487782326e-06,
+      "loss": 0.4497,
+      "step": 7230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3139621959908152e-06,
+      "loss": 0.4824,
+      "step": 7231
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3114922810427106e-06,
+      "loss": 0.5501,
+      "step": 7232
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.309023104206488e-06,
+      "loss": 0.6307,
+      "step": 7233
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3065546657546322e-06,
+      "loss": 0.5191,
+      "step": 7234
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.304086965959549e-06,
+      "loss": 0.5676,
+      "step": 7235
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.301620005093562e-06,
+      "loss": 0.416,
+      "step": 7236
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299153783428913e-06,
+      "loss": 0.6073,
+      "step": 7237
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2966883012377638e-06,
+      "loss": 0.6764,
+      "step": 7238
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2942235587921912e-06,
+      "loss": 0.5677,
+      "step": 7239
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.29175955636419e-06,
+      "loss": 0.8273,
+      "step": 7240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2892962942256834e-06,
+      "loss": 0.4735,
+      "step": 7241
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2868337726485047e-06,
+      "loss": 0.5925,
+      "step": 7242
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2843719919043982e-06,
+      "loss": 0.821,
+      "step": 7243
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2819109522650394e-06,
+      "loss": 0.5304,
+      "step": 7244
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2794506540020164e-06,
+      "loss": 0.5067,
+      "step": 7245
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2769910973868314e-06,
+      "loss": 0.7946,
+      "step": 7246
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2745322826909177e-06,
+      "loss": 0.5527,
+      "step": 7247
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.272074210185614e-06,
+      "loss": 0.387,
+      "step": 7248
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2696168801421824e-06,
+      "loss": 0.5005,
+      "step": 7249
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2671602928318048e-06,
+      "loss": 0.4164,
+      "step": 7250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2647044485255686e-06,
+      "loss": 0.536,
+      "step": 7251
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2622493474944994e-06,
+      "loss": 0.5652,
+      "step": 7252
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.259794990009527e-06,
+      "loss": 0.5519,
+      "step": 7253
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.257341376341502e-06,
+      "loss": 0.6021,
+      "step": 7254
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2548885067611946e-06,
+      "loss": 0.632,
+      "step": 7255
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.252436381539291e-06,
+      "loss": 0.7485,
+      "step": 7256
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2499850009463963e-06,
+      "loss": 0.5786,
+      "step": 7257
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2475343652530323e-06,
+      "loss": 0.5387,
+      "step": 7258
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2450844747296417e-06,
+      "loss": 0.6427,
+      "step": 7259
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2426353296465807e-06,
+      "loss": 0.7999,
+      "step": 7260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2401869302741263e-06,
+      "loss": 0.6272,
+      "step": 7261
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237739276882472e-06,
+      "loss": 0.6011,
+      "step": 7262
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2352923697417293e-06,
+      "loss": 0.4298,
+      "step": 7263
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.232846209121927e-06,
+      "loss": 0.4416,
+      "step": 7264
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2304007952930116e-06,
+      "loss": 0.5534,
+      "step": 7265
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2279561285248474e-06,
+      "loss": 0.5679,
+      "step": 7266
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.225512209087217e-06,
+      "loss": 0.7292,
+      "step": 7267
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2230690372498153e-06,
+      "loss": 0.7392,
+      "step": 7268
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.220626613282266e-06,
+      "loss": 0.5806,
+      "step": 7269
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.218184937454103e-06,
+      "loss": 0.4652,
+      "step": 7270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.215744010034774e-06,
+      "loss": 0.4452,
+      "step": 7271
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.213303831293647e-06,
+      "loss": 0.4874,
+      "step": 7272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.210864401500009e-06,
+      "loss": 0.7116,
+      "step": 7273
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2084257209230683e-06,
+      "loss": 0.586,
+      "step": 7274
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2059877898319447e-06,
+      "loss": 0.4074,
+      "step": 7275
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2035506084956745e-06,
+      "loss": 0.5915,
+      "step": 7276
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.201114177183219e-06,
+      "loss": 0.5826,
+      "step": 7277
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.198678496163442e-06,
+      "loss": 0.6153,
+      "step": 7278
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1962435657051362e-06,
+      "loss": 0.5746,
+      "step": 7279
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1938093860770137e-06,
+      "loss": 0.6146,
+      "step": 7280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1913759575476953e-06,
+      "loss": 0.526,
+      "step": 7281
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1889432803857244e-06,
+      "loss": 0.7503,
+      "step": 7282
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1865113548595595e-06,
+      "loss": 0.5653,
+      "step": 7283
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1840801812375755e-06,
+      "loss": 0.8604,
+      "step": 7284
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1816497597880646e-06,
+      "loss": 0.9548,
+      "step": 7285
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.179220090779238e-06,
+      "loss": 0.5558,
+      "step": 7286
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176791174479222e-06,
+      "loss": 0.4714,
+      "step": 7287
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1743630111560597e-06,
+      "loss": 0.5009,
+      "step": 7288
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1719356010777114e-06,
+      "loss": 0.448,
+      "step": 7289
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1695089445120553e-06,
+      "loss": 0.6672,
+      "step": 7290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167083041726886e-06,
+      "loss": 0.4141,
+      "step": 7291
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.164657892989913e-06,
+      "loss": 0.6763,
+      "step": 7292
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1622334985687663e-06,
+      "loss": 0.5454,
+      "step": 7293
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1598098587309888e-06,
+      "loss": 0.5684,
+      "step": 7294
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1573869737440423e-06,
+      "loss": 0.6285,
+      "step": 7295
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.154964843875302e-06,
+      "loss": 0.6499,
+      "step": 7296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.152543469392072e-06,
+      "loss": 0.8728,
+      "step": 7297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1501228505615534e-06,
+      "loss": 0.5058,
+      "step": 7298
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.147702987650879e-06,
+      "loss": 0.5698,
+      "step": 7299
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1452838809270925e-06,
+      "loss": 0.5126,
+      "step": 7300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.142865530657151e-06,
+      "loss": 0.6027,
+      "step": 7301
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.14044793710794e-06,
+      "loss": 0.792,
+      "step": 7302
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1380311005462494e-06,
+      "loss": 0.6994,
+      "step": 7303
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1356150212387924e-06,
+      "loss": 0.5351,
+      "step": 7304
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1331996994521917e-06,
+      "loss": 0.4812,
+      "step": 7305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.130785135452993e-06,
+      "loss": 0.5771,
+      "step": 7306
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.128371329507651e-06,
+      "loss": 0.3898,
+      "step": 7307
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1259582818825507e-06,
+      "loss": 0.5948,
+      "step": 7308
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1235459928439805e-06,
+      "loss": 0.4843,
+      "step": 7309
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1211344626581496e-06,
+      "loss": 0.6361,
+      "step": 7310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1187236915911845e-06,
+      "loss": 0.5527,
+      "step": 7311
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1163136799091187e-06,
+      "loss": 0.4968,
+      "step": 7312
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1139044278779183e-06,
+      "loss": 0.4714,
+      "step": 7313
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1114959357634545e-06,
+      "loss": 0.5911,
+      "step": 7314
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1090882038315163e-06,
+      "loss": 0.4692,
+      "step": 7315
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.106681232347809e-06,
+      "loss": 0.3313,
+      "step": 7316
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.104275021577956e-06,
+      "loss": 0.6274,
+      "step": 7317
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1018695717874947e-06,
+      "loss": 0.6697,
+      "step": 7318
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0994648832418782e-06,
+      "loss": 0.5312,
+      "step": 7319
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0970609562064768e-06,
+      "loss": 0.6816,
+      "step": 7320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.094657790946578e-06,
+      "loss": 0.6072,
+      "step": 7321
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.092255387727382e-06,
+      "loss": 0.4094,
+      "step": 7322
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0898537468140033e-06,
+      "loss": 0.5532,
+      "step": 7323
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0874528684714867e-06,
+      "loss": 0.5672,
+      "step": 7324
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.08505275296477e-06,
+      "loss": 0.5867,
+      "step": 7325
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0826534005587238e-06,
+      "loss": 0.5187,
+      "step": 7326
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0802548115181284e-06,
+      "loss": 0.6482,
+      "step": 7327
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.07785698610768e-06,
+      "loss": 0.3432,
+      "step": 7328
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.075459924591989e-06,
+      "loss": 0.2878,
+      "step": 7329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0730636272355897e-06,
+      "loss": 0.6337,
+      "step": 7330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0706680943029254e-06,
+      "loss": 0.6648,
+      "step": 7331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.06827332605835e-06,
+      "loss": 0.5758,
+      "step": 7332
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.065879322766142e-06,
+      "loss": 0.63,
+      "step": 7333
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.063486084690489e-06,
+      "loss": 0.5953,
+      "step": 7334
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0610936120955037e-06,
+      "loss": 0.5314,
+      "step": 7335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.058701905245205e-06,
+      "loss": 0.5212,
+      "step": 7336
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0563109644035293e-06,
+      "loss": 0.512,
+      "step": 7337
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0539207898343304e-06,
+      "loss": 0.6211,
+      "step": 7338
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.05153138180138e-06,
+      "loss": 0.5294,
+      "step": 7339
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0491427405683517e-06,
+      "loss": 0.6596,
+      "step": 7340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0467548663988544e-06,
+      "loss": 0.5676,
+      "step": 7341
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0443677595563982e-06,
+      "loss": 0.594,
+      "step": 7342
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.041981420304415e-06,
+      "loss": 0.6422,
+      "step": 7343
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0395958489062484e-06,
+      "loss": 0.645,
+      "step": 7344
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037211045625158e-06,
+      "loss": 0.4269,
+      "step": 7345
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0348270107243204e-06,
+      "loss": 0.5327,
+      "step": 7346
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0324437444668276e-06,
+      "loss": 0.8945,
+      "step": 7347
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0300612471156832e-06,
+      "loss": 0.5335,
+      "step": 7348
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027679518933809e-06,
+      "loss": 0.6809,
+      "step": 7349
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0252985601840424e-06,
+      "loss": 0.5757,
+      "step": 7350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0229183711291333e-06,
+      "loss": 0.5214,
+      "step": 7351
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0205389520317485e-06,
+      "loss": 0.4789,
+      "step": 7352
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0181603031544694e-06,
+      "loss": 0.8032,
+      "step": 7353
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.015782424759792e-06,
+      "loss": 0.6213,
+      "step": 7354
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0134053171101286e-06,
+      "loss": 0.6627,
+      "step": 7355
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0110289804678037e-06,
+      "loss": 0.7013,
+      "step": 7356
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0086534150950576e-06,
+      "loss": 0.5845,
+      "step": 7357
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0062786212540508e-06,
+      "loss": 0.5695,
+      "step": 7358
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0039045992068548e-06,
+      "loss": 0.4029,
+      "step": 7359
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0015313492154497e-06,
+      "loss": 0.5601,
+      "step": 7360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9991588715417386e-06,
+      "loss": 0.5339,
+      "step": 7361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9967871664475346e-06,
+      "loss": 0.7024,
+      "step": 7362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9944162341945727e-06,
+      "loss": 0.7951,
+      "step": 7363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9920460750444958e-06,
+      "loss": 0.811,
+      "step": 7364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9896766892588613e-06,
+      "loss": 0.6728,
+      "step": 7365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9873080770991504e-06,
+      "loss": 0.6614,
+      "step": 7366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.984940238826741e-06,
+      "loss": 0.5466,
+      "step": 7367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9825731747029394e-06,
+      "loss": 0.6925,
+      "step": 7368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9802068849889697e-06,
+      "loss": 0.4274,
+      "step": 7369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9778413699459598e-06,
+      "loss": 0.5909,
+      "step": 7370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975476629834958e-06,
+      "loss": 0.527,
+      "step": 7371
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.973112664916925e-06,
+      "loss": 0.8547,
+      "step": 7372
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9707494754527376e-06,
+      "loss": 0.652,
+      "step": 7373
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.968387061703184e-06,
+      "loss": 0.6325,
+      "step": 7374
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.966025423928972e-06,
+      "loss": 0.5069,
+      "step": 7375
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963664562390719e-06,
+      "loss": 0.5068,
+      "step": 7376
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9613044773489576e-06,
+      "loss": 0.4734,
+      "step": 7377
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958945169064137e-06,
+      "loss": 0.5588,
+      "step": 7378
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.956586637796619e-06,
+      "loss": 0.6785,
+      "step": 7379
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.954228883806679e-06,
+      "loss": 0.7473,
+      "step": 7380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9518719073545086e-06,
+      "loss": 0.644,
+      "step": 7381
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9495157087002126e-06,
+      "loss": 0.5609,
+      "step": 7382
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9471602881038084e-06,
+      "loss": 0.6119,
+      "step": 7383
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9448056458252284e-06,
+      "loss": 0.6139,
+      "step": 7384
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9424517821243224e-06,
+      "loss": 0.5039,
+      "step": 7385
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9400986972608547e-06,
+      "loss": 0.4329,
+      "step": 7386
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9377463914944937e-06,
+      "loss": 0.5155,
+      "step": 7387
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9353948650848297e-06,
+      "loss": 0.4166,
+      "step": 7388
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9330441182913695e-06,
+      "loss": 0.5623,
+      "step": 7389
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.930694151373523e-06,
+      "loss": 0.7668,
+      "step": 7390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9283449645906315e-06,
+      "loss": 0.6549,
+      "step": 7391
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9259965582019344e-06,
+      "loss": 0.5816,
+      "step": 7392
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.923648932466596e-06,
+      "loss": 0.7369,
+      "step": 7393
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9213020876436806e-06,
+      "loss": 0.6574,
+      "step": 7394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9189560239921756e-06,
+      "loss": 0.6127,
+      "step": 7395
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.916610741770989e-06,
+      "loss": 0.7197,
+      "step": 7396
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.914266241238929e-06,
+      "loss": 0.4989,
+      "step": 7397
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9119225226547266e-06,
+      "loss": 0.6675,
+      "step": 7398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9095795862770216e-06,
+      "loss": 0.4358,
+      "step": 7399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9072374323643725e-06,
+      "loss": 0.646,
+      "step": 7400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9048960611752396e-06,
+      "loss": 0.4921,
+      "step": 7401
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.902555472968015e-06,
+      "loss": 0.5022,
+      "step": 7402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.900215668000991e-06,
+      "loss": 0.6351,
+      "step": 7403
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.897876646532377e-06,
+      "loss": 0.5578,
+      "step": 7404
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8955384088202964e-06,
+      "loss": 0.7214,
+      "step": 7405
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.893200955122788e-06,
+      "loss": 0.6477,
+      "step": 7406
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8908642856978e-06,
+      "loss": 0.8115,
+      "step": 7407
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8885284008031965e-06,
+      "loss": 0.5907,
+      "step": 7408
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8861933006967547e-06,
+      "loss": 0.7739,
+      "step": 7409
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8838589856361656e-06,
+      "loss": 0.6253,
+      "step": 7410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.881525455879033e-06,
+      "loss": 0.6359,
+      "step": 7411
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87919271168287e-06,
+      "loss": 0.7799,
+      "step": 7412
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8768607533051175e-06,
+      "loss": 0.5023,
+      "step": 7413
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8745295810031105e-06,
+      "loss": 0.4416,
+      "step": 7414
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872199195034108e-06,
+      "loss": 0.6103,
+      "step": 7415
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.86986959565528e-06,
+      "loss": 0.6421,
+      "step": 7416
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867540783123711e-06,
+      "loss": 0.5604,
+      "step": 7417
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.865212757696395e-06,
+      "loss": 0.6532,
+      "step": 7418
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862885519630246e-06,
+      "loss": 0.5834,
+      "step": 7419
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8605590691820895e-06,
+      "loss": 0.6059,
+      "step": 7420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8582334066086536e-06,
+      "loss": 0.398,
+      "step": 7421
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8559085321665916e-06,
+      "loss": 0.5376,
+      "step": 7422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8535844461124608e-06,
+      "loss": 0.3495,
+      "step": 7423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.851261148702744e-06,
+      "loss": 0.6267,
+      "step": 7424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8489386401938256e-06,
+      "loss": 0.8319,
+      "step": 7425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8466169208420058e-06,
+      "loss": 0.7074,
+      "step": 7426
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8442959909035005e-06,
+      "loss": 0.6509,
+      "step": 7427
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8419758506344373e-06,
+      "loss": 0.5022,
+      "step": 7428
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.839656500290848e-06,
+      "loss": 0.3494,
+      "step": 7429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.837337940128694e-06,
+      "loss": 0.5764,
+      "step": 7430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.835020170403837e-06,
+      "loss": 0.4727,
+      "step": 7431
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.832703191372056e-06,
+      "loss": 0.5731,
+      "step": 7432
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.830387003289041e-06,
+      "loss": 0.7185,
+      "step": 7433
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8280716064103966e-06,
+      "loss": 0.5878,
+      "step": 7434
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8257570009916367e-06,
+      "loss": 0.441,
+      "step": 7435
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.823443187288193e-06,
+      "loss": 0.4063,
+      "step": 7436
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8211301655554046e-06,
+      "loss": 0.6261,
+      "step": 7437
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.818817936048527e-06,
+      "loss": 0.7003,
+      "step": 7438
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8165064990227255e-06,
+      "loss": 1.0464,
+      "step": 7439
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8141958547330804e-06,
+      "loss": 0.4903,
+      "step": 7440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.811886003434584e-06,
+      "loss": 0.498,
+      "step": 7441
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8095769453821388e-06,
+      "loss": 0.5044,
+      "step": 7442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8072686808305625e-06,
+      "loss": 0.6296,
+      "step": 7443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8049612100345834e-06,
+      "loss": 0.3243,
+      "step": 7444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8026545332488418e-06,
+      "loss": 0.5093,
+      "step": 7445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8003486507278954e-06,
+      "loss": 0.6333,
+      "step": 7446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7980435627262083e-06,
+      "loss": 0.6529,
+      "step": 7447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.795739269498163e-06,
+      "loss": 0.5113,
+      "step": 7448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.793435771298043e-06,
+      "loss": 0.5419,
+      "step": 7449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7911330683800543e-06,
+      "loss": 0.6089,
+      "step": 7450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.788831160998311e-06,
+      "loss": 0.4865,
+      "step": 7451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786530049406844e-06,
+      "loss": 0.6557,
+      "step": 7452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7842297338595925e-06,
+      "loss": 0.694,
+      "step": 7453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7819302146104068e-06,
+      "loss": 0.6811,
+      "step": 7454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.779631491913055e-06,
+      "loss": 0.6636,
+      "step": 7455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.777333566021203e-06,
+      "loss": 0.6594,
+      "step": 7456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7750364371884488e-06,
+      "loss": 0.6165,
+      "step": 7457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.772740105668289e-06,
+      "loss": 0.6534,
+      "step": 7458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.770444571714137e-06,
+      "loss": 0.6242,
+      "step": 7459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7681498355793157e-06,
+      "loss": 0.5587,
+      "step": 7460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7658558975170624e-06,
+      "loss": 0.6863,
+      "step": 7461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7635627577805235e-06,
+      "loss": 0.3792,
+      "step": 7462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.761270416622761e-06,
+      "loss": 0.6223,
+      "step": 7463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.758978874296745e-06,
+      "loss": 0.5431,
+      "step": 7464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7566881310553604e-06,
+      "loss": 0.6809,
+      "step": 7465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7543981871514026e-06,
+      "loss": 0.6481,
+      "step": 7466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7521090428375786e-06,
+      "loss": 0.553,
+      "step": 7467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7498206983665076e-06,
+      "loss": 0.5299,
+      "step": 7468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.74753315399072e-06,
+      "loss": 0.4449,
+      "step": 7469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7452464099626585e-06,
+      "loss": 0.4704,
+      "step": 7470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7429604665346775e-06,
+      "loss": 0.625,
+      "step": 7471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7406753239590436e-06,
+      "loss": 0.6031,
+      "step": 7472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7383909824879295e-06,
+      "loss": 0.6973,
+      "step": 7473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7361074423734315e-06,
+      "loss": 0.3378,
+      "step": 7474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.73382470386755e-06,
+      "loss": 0.3869,
+      "step": 7475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7315427672221906e-06,
+      "loss": 0.5602,
+      "step": 7476
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7292616326891807e-06,
+      "loss": 0.5563,
+      "step": 7477
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.726981300520256e-06,
+      "loss": 0.6287,
+      "step": 7478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7247017709670586e-06,
+      "loss": 0.8289,
+      "step": 7479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.722423044281153e-06,
+      "loss": 0.6382,
+      "step": 7480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720145120714006e-06,
+      "loss": 0.6003,
+      "step": 7481
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.717868000517001e-06,
+      "loss": 0.4057,
+      "step": 7482
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715591683941424e-06,
+      "loss": 0.4594,
+      "step": 7483
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7133161712384803e-06,
+      "loss": 0.7405,
+      "step": 7484
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7110414626592883e-06,
+      "loss": 0.7722,
+      "step": 7485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7087675584548724e-06,
+      "loss": 0.6791,
+      "step": 7486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7064944588761686e-06,
+      "loss": 0.3823,
+      "step": 7487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704222164174026e-06,
+      "loss": 0.6099,
+      "step": 7488
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7019506745992086e-06,
+      "loss": 0.5406,
+      "step": 7489
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6996799904023764e-06,
+      "loss": 0.6652,
+      "step": 7490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69741011183412e-06,
+      "loss": 0.668,
+      "step": 7491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6951410391449317e-06,
+      "loss": 0.9405,
+      "step": 7492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6928727725852134e-06,
+      "loss": 0.6455,
+      "step": 7493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.69060531240528e-06,
+      "loss": 0.4803,
+      "step": 7494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6883386588553597e-06,
+      "loss": 0.4661,
+      "step": 7495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6860728121855884e-06,
+      "loss": 0.5897,
+      "step": 7496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6838077726460132e-06,
+      "loss": 0.5804,
+      "step": 7497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.681543540486595e-06,
+      "loss": 0.8399,
+      "step": 7498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6792801159572026e-06,
+      "loss": 0.4771,
+      "step": 7499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.677017499307617e-06,
+      "loss": 0.709,
+      "step": 7500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.674755690787526e-06,
+      "loss": 0.45,
+      "step": 7501
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.672494690646542e-06,
+      "loss": 0.6756,
+      "step": 7502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670234499134169e-06,
+      "loss": 0.7193,
+      "step": 7503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6679751164998346e-06,
+      "loss": 0.4369,
+      "step": 7504
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6657165429928724e-06,
+      "loss": 0.474,
+      "step": 7505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6634587788625245e-06,
+      "loss": 0.5077,
+      "step": 7506
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.661201824357954e-06,
+      "loss": 0.632,
+      "step": 7507
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6589456797282254e-06,
+      "loss": 0.6317,
+      "step": 7508
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.656690345222317e-06,
+      "loss": 0.4134,
+      "step": 7509
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.654435821089113e-06,
+      "loss": 0.6524,
+      "step": 7510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6521821075774137e-06,
+      "loss": 0.6684,
+      "step": 7511
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.649929204935924e-06,
+      "loss": 0.5888,
+      "step": 7512
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6476771134132716e-06,
+      "loss": 0.5979,
+      "step": 7513
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.645425833257983e-06,
+      "loss": 0.6945,
+      "step": 7514
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6431753647184986e-06,
+      "loss": 0.7002,
+      "step": 7515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6409257080431695e-06,
+      "loss": 0.4626,
+      "step": 7516
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6386768634802574e-06,
+      "loss": 0.8173,
+      "step": 7517
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.636428831277933e-06,
+      "loss": 0.6787,
+      "step": 7518
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.63418161168428e-06,
+      "loss": 0.783,
+      "step": 7519
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.631935204947289e-06,
+      "loss": 0.9353,
+      "step": 7520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.629689611314864e-06,
+      "loss": 0.4986,
+      "step": 7521
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.627444831034818e-06,
+      "loss": 0.589,
+      "step": 7522
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6252008643548742e-06,
+      "loss": 0.6398,
+      "step": 7523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6229577115226656e-06,
+      "loss": 0.3588,
+      "step": 7524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.620715372785736e-06,
+      "loss": 0.5991,
+      "step": 7525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6184738483915406e-06,
+      "loss": 0.5008,
+      "step": 7526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.616233138587441e-06,
+      "loss": 0.8018,
+      "step": 7527
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6139932436207126e-06,
+      "loss": 0.3989,
+      "step": 7528
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6117541637385403e-06,
+      "loss": 0.5659,
+      "step": 7529
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6095158991880167e-06,
+      "loss": 0.7892,
+      "step": 7530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6072784502161463e-06,
+      "loss": 0.5098,
+      "step": 7531
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6050418170698455e-06,
+      "loss": 0.4844,
+      "step": 7532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6028059999959356e-06,
+      "loss": 0.5918,
+      "step": 7533
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6005709992411487e-06,
+      "loss": 0.5075,
+      "step": 7534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.598336815052135e-06,
+      "loss": 0.4945,
+      "step": 7535
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5961034476754486e-06,
+      "loss": 0.5918,
+      "step": 7536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.593870897357548e-06,
+      "loss": 0.3363,
+      "step": 7537
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591639164344808e-06,
+      "loss": 0.6137,
+      "step": 7538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5894082488835125e-06,
+      "loss": 0.4614,
+      "step": 7539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5871781512198534e-06,
+      "loss": 0.9527,
+      "step": 7540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.584948871599938e-06,
+      "loss": 0.6104,
+      "step": 7541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5827204102697766e-06,
+      "loss": 0.6561,
+      "step": 7542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.580492767475292e-06,
+      "loss": 0.6476,
+      "step": 7543
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.578265943462318e-06,
+      "loss": 0.3681,
+      "step": 7544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576039938476589e-06,
+      "loss": 0.7225,
+      "step": 7545
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5738147527637636e-06,
+      "loss": 0.5574,
+      "step": 7546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5715903865694003e-06,
+      "loss": 0.4462,
+      "step": 7547
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5693668401389705e-06,
+      "loss": 0.4604,
+      "step": 7548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.567144113717852e-06,
+      "loss": 0.6166,
+      "step": 7549
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564922207551337e-06,
+      "loss": 0.431,
+      "step": 7550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.562701121884623e-06,
+      "loss": 0.8158,
+      "step": 7551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5604808569628193e-06,
+      "loss": 0.3794,
+      "step": 7552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5582614130309424e-06,
+      "loss": 0.6793,
+      "step": 7553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.55604279033392e-06,
+      "loss": 0.6692,
+      "step": 7554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.553824989116589e-06,
+      "loss": 0.3939,
+      "step": 7555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551608009623696e-06,
+      "loss": 0.5324,
+      "step": 7556
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5493918520998963e-06,
+      "loss": 0.3628,
+      "step": 7557
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5471765167897535e-06,
+      "loss": 0.5698,
+      "step": 7558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.544962003937741e-06,
+      "loss": 0.5548,
+      "step": 7559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5427483137882434e-06,
+      "loss": 0.5268,
+      "step": 7560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5405354465855527e-06,
+      "loss": 0.5655,
+      "step": 7561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.538323402573867e-06,
+      "loss": 0.6147,
+      "step": 7562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5361121819973036e-06,
+      "loss": 0.6739,
+      "step": 7563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5339017850998804e-06,
+      "loss": 0.5313,
+      "step": 7564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5316922121255226e-06,
+      "loss": 0.692,
+      "step": 7565
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5294834633180697e-06,
+      "loss": 0.7231,
+      "step": 7566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527275538921267e-06,
+      "loss": 0.7086,
+      "step": 7567
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525068439178775e-06,
+      "loss": 0.5682,
+      "step": 7568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5228621643341566e-06,
+      "loss": 0.9633,
+      "step": 7569
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.520656714630886e-06,
+      "loss": 0.4908,
+      "step": 7570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.518452090312348e-06,
+      "loss": 0.501,
+      "step": 7571
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.51624829162183e-06,
+      "loss": 0.5857,
+      "step": 7572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5140453188025314e-06,
+      "loss": 0.5929,
+      "step": 7573
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511843172097569e-06,
+      "loss": 0.6018,
+      "step": 7574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5096418517499566e-06,
+      "loss": 0.6145,
+      "step": 7575
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5074413580026235e-06,
+      "loss": 0.4101,
+      "step": 7576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5052416910984035e-06,
+      "loss": 0.4855,
+      "step": 7577
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.503042851280043e-06,
+      "loss": 0.6561,
+      "step": 7578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.500844838790195e-06,
+      "loss": 0.7883,
+      "step": 7579
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498647653871422e-06,
+      "loss": 0.6319,
+      "step": 7580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4964512967661948e-06,
+      "loss": 0.3883,
+      "step": 7581
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494255767716892e-06,
+      "loss": 0.6767,
+      "step": 7582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.492061066965803e-06,
+      "loss": 0.5035,
+      "step": 7583
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4898671947551234e-06,
+      "loss": 0.3819,
+      "step": 7584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4876741513269597e-06,
+      "loss": 0.5683,
+      "step": 7585
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.485481936923325e-06,
+      "loss": 0.5703,
+      "step": 7586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4832905517861416e-06,
+      "loss": 0.581,
+      "step": 7587
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.48109999615724e-06,
+      "loss": 0.4643,
+      "step": 7588
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4789102702783607e-06,
+      "loss": 0.5513,
+      "step": 7589
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.476721374391147e-06,
+      "loss": 0.5654,
+      "step": 7590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4745333087371646e-06,
+      "loss": 0.5895,
+      "step": 7591
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4723460735578696e-06,
+      "loss": 0.4322,
+      "step": 7592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4701596690946384e-06,
+      "loss": 0.6453,
+      "step": 7593
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.46797409558875e-06,
+      "loss": 0.4482,
+      "step": 7594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4657893532813924e-06,
+      "loss": 0.6124,
+      "step": 7595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.463605442413669e-06,
+      "loss": 0.8905,
+      "step": 7596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461422363226582e-06,
+      "loss": 0.6873,
+      "step": 7597
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.45924011596105e-06,
+      "loss": 0.5369,
+      "step": 7598
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4570587008578896e-06,
+      "loss": 0.5578,
+      "step": 7599
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454878118157834e-06,
+      "loss": 0.5782,
+      "step": 7600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.452698368101517e-06,
+      "loss": 0.4135,
+      "step": 7601
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.450519450929495e-06,
+      "loss": 0.624,
+      "step": 7602
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4483413668822164e-06,
+      "loss": 0.4909,
+      "step": 7603
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.446164116200047e-06,
+      "loss": 0.5874,
+      "step": 7604
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.443987699123259e-06,
+      "loss": 1.0221,
+      "step": 7605
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4418121158920228e-06,
+      "loss": 0.7061,
+      "step": 7606
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439637366746435e-06,
+      "loss": 0.5151,
+      "step": 7607
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4374634519264872e-06,
+      "loss": 0.7245,
+      "step": 7608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4352903716720822e-06,
+      "loss": 0.6604,
+      "step": 7609
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.433118126223031e-06,
+      "loss": 0.5803,
+      "step": 7610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4309467158190525e-06,
+      "loss": 0.6254,
+      "step": 7611
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4287761406997722e-06,
+      "loss": 0.9215,
+      "step": 7612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4266064011047263e-06,
+      "loss": 0.4696,
+      "step": 7613
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424437497273354e-06,
+      "loss": 0.5478,
+      "step": 7614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4222694294450087e-06,
+      "loss": 0.6624,
+      "step": 7615
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4201021978589458e-06,
+      "loss": 0.7024,
+      "step": 7616
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4179358027543275e-06,
+      "loss": 0.7342,
+      "step": 7617
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415770244370238e-06,
+      "loss": 0.507,
+      "step": 7618
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4136055229456466e-06,
+      "loss": 0.5964,
+      "step": 7619
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.411441638719445e-06,
+      "loss": 0.5601,
+      "step": 7620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409278591930431e-06,
+      "loss": 0.4942,
+      "step": 7621
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.407116382817306e-06,
+      "loss": 0.5413,
+      "step": 7622
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4049550116186803e-06,
+      "loss": 0.4927,
+      "step": 7623
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.402794478573076e-06,
+      "loss": 0.6634,
+      "step": 7624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4006347839189215e-06,
+      "loss": 0.3312,
+      "step": 7625
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3984759278945426e-06,
+      "loss": 0.6873,
+      "step": 7626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.396317910738184e-06,
+      "loss": 0.5076,
+      "step": 7627
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3941607326879925e-06,
+      "loss": 0.7658,
+      "step": 7628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3920043939820283e-06,
+      "loss": 0.316,
+      "step": 7629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3898488948582533e-06,
+      "loss": 0.742,
+      "step": 7630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3876942355545363e-06,
+      "loss": 0.626,
+      "step": 7631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3855404163086558e-06,
+      "loss": 0.5361,
+      "step": 7632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3833874373583e-06,
+      "loss": 0.393,
+      "step": 7633
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3812352989410537e-06,
+      "loss": 0.5197,
+      "step": 7634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379084001294425e-06,
+      "loss": 0.6014,
+      "step": 7635
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3769335446558163e-06,
+      "loss": 0.7154,
+      "step": 7636
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.374783929262544e-06,
+      "loss": 0.5976,
+      "step": 7637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.372635155351828e-06,
+      "loss": 0.5989,
+      "step": 7638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.370487223160797e-06,
+      "loss": 0.4594,
+      "step": 7639
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3683401329264878e-06,
+      "loss": 0.8696,
+      "step": 7640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366193884885841e-06,
+      "loss": 0.5353,
+      "step": 7641
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.364048479275708e-06,
+      "loss": 0.6232,
+      "step": 7642
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3619039163328437e-06,
+      "loss": 0.5682,
+      "step": 7643
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3597601962939144e-06,
+      "loss": 0.6912,
+      "step": 7644
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357617319395489e-06,
+      "loss": 0.7841,
+      "step": 7645
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355475285874046e-06,
+      "loss": 0.7878,
+      "step": 7646
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.35333409596597e-06,
+      "loss": 0.5735,
+      "step": 7647
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.351193749907551e-06,
+      "loss": 0.3982,
+      "step": 7648
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34905424793499e-06,
+      "loss": 0.5507,
+      "step": 7649
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346915590284391e-06,
+      "loss": 0.5039,
+      "step": 7650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3447777771917625e-06,
+      "loss": 0.4949,
+      "step": 7651
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34264080889303e-06,
+      "loss": 0.5718,
+      "step": 7652
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3405046856240175e-06,
+      "loss": 0.7468,
+      "step": 7653
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3383694076204555e-06,
+      "loss": 0.7402,
+      "step": 7654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3362349751179825e-06,
+      "loss": 0.5854,
+      "step": 7655
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3341013883521423e-06,
+      "loss": 0.5413,
+      "step": 7656
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3319686475583927e-06,
+      "loss": 0.5342,
+      "step": 7657
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.329836752972091e-06,
+      "loss": 0.5983,
+      "step": 7658
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3277057048285034e-06,
+      "loss": 0.701,
+      "step": 7659
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3255755033628037e-06,
+      "loss": 0.6513,
+      "step": 7660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3234461488100656e-06,
+      "loss": 0.6632,
+      "step": 7661
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.321317641405274e-06,
+      "loss": 0.5576,
+      "step": 7662
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3191899813833274e-06,
+      "loss": 0.5555,
+      "step": 7663
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3170631689790215e-06,
+      "loss": 0.7693,
+      "step": 7664
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.31493720442706e-06,
+      "loss": 0.7103,
+      "step": 7665
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3128120879620562e-06,
+      "loss": 0.6772,
+      "step": 7666
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3106878198185258e-06,
+      "loss": 0.6338,
+      "step": 7667
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308564400230894e-06,
+      "loss": 0.4483,
+      "step": 7668
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.30644182943349e-06,
+      "loss": 0.3998,
+      "step": 7669
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3043201076605526e-06,
+      "loss": 0.6638,
+      "step": 7670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.302199235146224e-06,
+      "loss": 0.6603,
+      "step": 7671
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.300079212124553e-06,
+      "loss": 0.6827,
+      "step": 7672
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2979600388294963e-06,
+      "loss": 0.6997,
+      "step": 7673
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2958417154949153e-06,
+      "loss": 0.7202,
+      "step": 7674
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.293724242354578e-06,
+      "loss": 0.637,
+      "step": 7675
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291607619642159e-06,
+      "loss": 0.8788,
+      "step": 7676
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2894918475912375e-06,
+      "loss": 0.607,
+      "step": 7677
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2873769264353e-06,
+      "loss": 0.7889,
+      "step": 7678
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.285262856407741e-06,
+      "loss": 0.5424,
+      "step": 7679
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2831496377418617e-06,
+      "loss": 0.5746,
+      "step": 7680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2810372706708604e-06,
+      "loss": 0.5541,
+      "step": 7681
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2789257554278523e-06,
+      "loss": 0.6825,
+      "step": 7682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.276815092245851e-06,
+      "loss": 0.6486,
+      "step": 7683
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2747052813577795e-06,
+      "loss": 0.5503,
+      "step": 7684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272596322996471e-06,
+      "loss": 0.6665,
+      "step": 7685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2704882173946564e-06,
+      "loss": 0.5368,
+      "step": 7686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.26838096478498e-06,
+      "loss": 0.7424,
+      "step": 7687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.266274565399982e-06,
+      "loss": 0.6157,
+      "step": 7688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2641690194721156e-06,
+      "loss": 0.6873,
+      "step": 7689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2620643272337428e-06,
+      "loss": 0.7082,
+      "step": 7690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.259960488917127e-06,
+      "loss": 0.6588,
+      "step": 7691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.257857504754435e-06,
+      "loss": 0.9076,
+      "step": 7692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.255755374977744e-06,
+      "loss": 0.5895,
+      "step": 7693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.253654099819037e-06,
+      "loss": 0.8221,
+      "step": 7694
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2515536795101933e-06,
+      "loss": 0.6848,
+      "step": 7695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2494541142830128e-06,
+      "loss": 0.4462,
+      "step": 7696
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2473554043691915e-06,
+      "loss": 0.6047,
+      "step": 7697
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2452575500003315e-06,
+      "loss": 0.4506,
+      "step": 7698
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2431605514079435e-06,
+      "loss": 0.5903,
+      "step": 7699
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2410644088234423e-06,
+      "loss": 0.5237,
+      "step": 7700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2389691224781472e-06,
+      "loss": 0.7257,
+      "step": 7701
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2368746926032846e-06,
+      "loss": 0.6638,
+      "step": 7702
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234781119429986e-06,
+      "loss": 0.5765,
+      "step": 7703
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2326884031892884e-06,
+      "loss": 0.453,
+      "step": 7704
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2305965441121324e-06,
+      "loss": 0.6354,
+      "step": 7705
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2285055424293643e-06,
+      "loss": 0.5053,
+      "step": 7706
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2264153983717464e-06,
+      "loss": 0.5201,
+      "step": 7707
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2243261121699257e-06,
+      "loss": 0.4245,
+      "step": 7708
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2222376840544714e-06,
+      "loss": 0.7272,
+      "step": 7709
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2201501142558514e-06,
+      "loss": 0.5728,
+      "step": 7710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.21806340300444e-06,
+      "loss": 0.8252,
+      "step": 7711
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.215977550530514e-06,
+      "loss": 0.4616,
+      "step": 7712
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2138925570642632e-06,
+      "loss": 0.5048,
+      "step": 7713
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2118084228357784e-06,
+      "loss": 0.4948,
+      "step": 7714
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.209725148075049e-06,
+      "loss": 0.6159,
+      "step": 7715
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207642733011979e-06,
+      "loss": 0.6436,
+      "step": 7716
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2055611778763684e-06,
+      "loss": 0.7093,
+      "step": 7717
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.203480482897935e-06,
+      "loss": 0.6294,
+      "step": 7718
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2014006483062922e-06,
+      "loss": 0.5591,
+      "step": 7719
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1993216743309586e-06,
+      "loss": 0.6291,
+      "step": 7720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.197243561201362e-06,
+      "loss": 0.5644,
+      "step": 7721
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1951663091468355e-06,
+      "loss": 0.4974,
+      "step": 7722
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.193089918396606e-06,
+      "loss": 0.8786,
+      "step": 7723
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.191014389179822e-06,
+      "loss": 0.558,
+      "step": 7724
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.188939721725526e-06,
+      "loss": 0.7964,
+      "step": 7725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1868659162626694e-06,
+      "loss": 0.5777,
+      "step": 7726
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.184792973020107e-06,
+      "loss": 0.4627,
+      "step": 7727
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1827208922266e-06,
+      "loss": 0.4374,
+      "step": 7728
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1806496741108117e-06,
+      "loss": 0.6956,
+      "step": 7729
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.178579318901313e-06,
+      "loss": 0.6117,
+      "step": 7730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176509826826577e-06,
+      "loss": 0.6142,
+      "step": 7731
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1744411981149848e-06,
+      "loss": 0.76,
+      "step": 7732
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.172373432994821e-06,
+      "loss": 0.6657,
+      "step": 7733
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1703065316942717e-06,
+      "loss": 0.4128,
+      "step": 7734
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1682404944414316e-06,
+      "loss": 0.6926,
+      "step": 7735
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.166175321464299e-06,
+      "loss": 0.8239,
+      "step": 7736
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.164111012990776e-06,
+      "loss": 0.6808,
+      "step": 7737
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.162047569248671e-06,
+      "loss": 0.4013,
+      "step": 7738
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.159984990465691e-06,
+      "loss": 0.5065,
+      "step": 7739
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.15792327686946e-06,
+      "loss": 0.5145,
+      "step": 7740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1558624286874965e-06,
+      "loss": 0.4419,
+      "step": 7741
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153802446147225e-06,
+      "loss": 0.7064,
+      "step": 7742
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1517433294759737e-06,
+      "loss": 0.6372,
+      "step": 7743
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149685078900978e-06,
+      "loss": 0.7264,
+      "step": 7744
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1476276946493747e-06,
+      "loss": 0.6407,
+      "step": 7745
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1455711769482113e-06,
+      "loss": 0.6547,
+      "step": 7746
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.143515526024433e-06,
+      "loss": 0.6028,
+      "step": 7747
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1414607421048926e-06,
+      "loss": 0.5153,
+      "step": 7748
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.139406825416348e-06,
+      "loss": 0.7243,
+      "step": 7749
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137353776185451e-06,
+      "loss": 0.6798,
+      "step": 7750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1353015946387757e-06,
+      "loss": 0.5139,
+      "step": 7751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.133250281002788e-06,
+      "loss": 0.6183,
+      "step": 7752
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.131199835503861e-06,
+      "loss": 0.5063,
+      "step": 7753
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.129150258368271e-06,
+      "loss": 0.6044,
+      "step": 7754
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1271015498222025e-06,
+      "loss": 0.3979,
+      "step": 7755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1250537100917377e-06,
+      "loss": 0.7142,
+      "step": 7756
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.12300673940287e-06,
+      "loss": 0.6001,
+      "step": 7757
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1209606379814896e-06,
+      "loss": 0.8219,
+      "step": 7758
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1189154060533977e-06,
+      "loss": 0.4423,
+      "step": 7759
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116871043844295e-06,
+      "loss": 0.434,
+      "step": 7760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1148275515797866e-06,
+      "loss": 0.4132,
+      "step": 7761
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1127849294853852e-06,
+      "loss": 0.5917,
+      "step": 7762
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1107431777865016e-06,
+      "loss": 0.5095,
+      "step": 7763
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1087022967084568e-06,
+      "loss": 1.0889,
+      "step": 7764
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1066622864764707e-06,
+      "loss": 0.5357,
+      "step": 7765
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.104623147315671e-06,
+      "loss": 0.4221,
+      "step": 7766
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1025848794510818e-06,
+      "loss": 0.5572,
+      "step": 7767
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1005474831076444e-06,
+      "loss": 0.3127,
+      "step": 7768
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.098510958510196e-06,
+      "loss": 0.727,
+      "step": 7769
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0964753058834717e-06,
+      "loss": 0.7188,
+      "step": 7770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0944405254521192e-06,
+      "loss": 0.5704,
+      "step": 7771
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0924066174406878e-06,
+      "loss": 0.6582,
+      "step": 7772
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090373582073627e-06,
+      "loss": 0.6735,
+      "step": 7773
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.088341419575297e-06,
+      "loss": 0.6816,
+      "step": 7774
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0863101301699563e-06,
+      "loss": 0.6814,
+      "step": 7775
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0842797140817704e-06,
+      "loss": 0.658,
+      "step": 7776
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.082250171534802e-06,
+      "loss": 0.5067,
+      "step": 7777
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08022150275302e-06,
+      "loss": 0.6976,
+      "step": 7778
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0781937079603054e-06,
+      "loss": 0.6491,
+      "step": 7779
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0761667873804334e-06,
+      "loss": 0.3579,
+      "step": 7780
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.074140741237084e-06,
+      "loss": 0.7092,
+      "step": 7781
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.072115569753843e-06,
+      "loss": 0.5799,
+      "step": 7782
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0700912731542022e-06,
+      "loss": 0.4645,
+      "step": 7783
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0680678516615426e-06,
+      "loss": 0.7532,
+      "step": 7784
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.06604530549917e-06,
+      "loss": 0.6277,
+      "step": 7785
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.06402363489028e-06,
+      "loss": 0.5083,
+      "step": 7786
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0620028400579727e-06,
+      "loss": 0.6796,
+      "step": 7787
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0599829212252552e-06,
+      "loss": 0.4476,
+      "step": 7788
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0579638786150345e-06,
+      "loss": 0.4945,
+      "step": 7789
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0559457124501246e-06,
+      "loss": 0.7778,
+      "step": 7790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0539284229532398e-06,
+      "loss": 0.5169,
+      "step": 7791
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.051912010346998e-06,
+      "loss": 0.6074,
+      "step": 7792
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.049896474853922e-06,
+      "loss": 0.5014,
+      "step": 7793
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0478818166964356e-06,
+      "loss": 0.491,
+      "step": 7794
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045868036096864e-06,
+      "loss": 0.4842,
+      "step": 7795
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043855133277448e-06,
+      "loss": 0.3604,
+      "step": 7796
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.041843108460313e-06,
+      "loss": 0.6081,
+      "step": 7797
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0398319618675e-06,
+      "loss": 0.5924,
+      "step": 7798
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037821693720947e-06,
+      "loss": 0.727,
+      "step": 7799
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035812304242498e-06,
+      "loss": 1.1086,
+      "step": 7800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0338037936539035e-06,
+      "loss": 0.5775,
+      "step": 7801
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0317961621768114e-06,
+      "loss": 0.3696,
+      "step": 7802
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0297894100327755e-06,
+      "loss": 0.4891,
+      "step": 7803
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0277835374432475e-06,
+      "loss": 0.7085,
+      "step": 7804
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0257785446295876e-06,
+      "loss": 0.5716,
+      "step": 7805
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0237744318130548e-06,
+      "loss": 0.7265,
+      "step": 7806
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0217711992148204e-06,
+      "loss": 0.5781,
+      "step": 7807
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0197688470559464e-06,
+      "loss": 0.5006,
+      "step": 7808
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0177673755574048e-06,
+      "loss": 0.7583,
+      "step": 7809
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0157667849400677e-06,
+      "loss": 0.6287,
+      "step": 7810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0137670754247106e-06,
+      "loss": 0.6478,
+      "step": 7811
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.011768247232012e-06,
+      "loss": 0.6205,
+      "step": 7812
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009770300582553e-06,
+      "loss": 0.7702,
+      "step": 7813
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.007773235696818e-06,
+      "loss": 0.6675,
+      "step": 7814
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.005777052795194e-06,
+      "loss": 0.4016,
+      "step": 7815
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0037817520979687e-06,
+      "loss": 0.5741,
+      "step": 7816
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0017873338253347e-06,
+      "loss": 0.5898,
+      "step": 7817
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.999793798197387e-06,
+      "loss": 0.5719,
+      "step": 7818
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9978011454341217e-06,
+      "loss": 0.573,
+      "step": 7819
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9958093757554397e-06,
+      "loss": 0.5455,
+      "step": 7820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.993818489381143e-06,
+      "loss": 0.6143,
+      "step": 7821
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9918284865309346e-06,
+      "loss": 0.6206,
+      "step": 7822
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9898393674244242e-06,
+      "loss": 0.438,
+      "step": 7823
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.98785113228112e-06,
+      "loss": 0.5739,
+      "step": 7824
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9858637813204352e-06,
+      "loss": 0.7667,
+      "step": 7825
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9838773147616828e-06,
+      "loss": 0.494,
+      "step": 7826
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.981891732824082e-06,
+      "loss": 0.6381,
+      "step": 7827
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9799070357267467e-06,
+      "loss": 0.6575,
+      "step": 7828
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.977923223688707e-06,
+      "loss": 0.9111,
+      "step": 7829
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9759402969288822e-06,
+      "loss": 0.635,
+      "step": 7830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.973958255666102e-06,
+      "loss": 0.5826,
+      "step": 7831
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.971977100119089e-06,
+      "loss": 0.7141,
+      "step": 7832
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9699968305064786e-06,
+      "loss": 0.7299,
+      "step": 7833
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9680174470467993e-06,
+      "loss": 0.7423,
+      "step": 7834
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9660389499584932e-06,
+      "loss": 0.5298,
+      "step": 7835
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.964061339459894e-06,
+      "loss": 0.8293,
+      "step": 7836
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962084615769242e-06,
+      "loss": 0.4663,
+      "step": 7837
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9601087791046825e-06,
+      "loss": 0.6939,
+      "step": 7838
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.958133829684249e-06,
+      "loss": 0.5691,
+      "step": 7839
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.956159767725898e-06,
+      "loss": 0.667,
+      "step": 7840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9541865934474734e-06,
+      "loss": 0.8463,
+      "step": 7841
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9522143070667263e-06,
+      "loss": 0.5722,
+      "step": 7842
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9502429088013086e-06,
+      "loss": 0.5592,
+      "step": 7843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9482723988687734e-06,
+      "loss": 0.6035,
+      "step": 7844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.946302777486578e-06,
+      "loss": 0.4286,
+      "step": 7845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9443340448720814e-06,
+      "loss": 0.706,
+      "step": 7846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94236620124254e-06,
+      "loss": 0.6531,
+      "step": 7847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9403992468151202e-06,
+      "loss": 0.5459,
+      "step": 7848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9384331818068827e-06,
+      "loss": 0.7559,
+      "step": 7849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.936468006434793e-06,
+      "loss": 0.6656,
+      "step": 7850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.934503720915719e-06,
+      "loss": 0.5248,
+      "step": 7851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.932540325466432e-06,
+      "loss": 0.5703,
+      "step": 7852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9305778203035997e-06,
+      "loss": 0.5843,
+      "step": 7853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9286162056437974e-06,
+      "loss": 0.7404,
+      "step": 7854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9266554817034976e-06,
+      "loss": 0.645,
+      "step": 7855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924695648699075e-06,
+      "loss": 0.4451,
+      "step": 7856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922736706846813e-06,
+      "loss": 0.5286,
+      "step": 7857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.920778656362889e-06,
+      "loss": 0.7452,
+      "step": 7858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918821497463382e-06,
+      "loss": 0.6124,
+      "step": 7859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9168652303642755e-06,
+      "loss": 0.5455,
+      "step": 7860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9149098552814504e-06,
+      "loss": 0.4668,
+      "step": 7861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9129553724307e-06,
+      "loss": 0.5574,
+      "step": 7862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.911001782027708e-06,
+      "loss": 0.4742,
+      "step": 7863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.909049084288063e-06,
+      "loss": 0.5309,
+      "step": 7864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.907097279427258e-06,
+      "loss": 0.6678,
+      "step": 7865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9051463676606808e-06,
+      "loss": 0.4419,
+      "step": 7866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9031963492036232e-06,
+      "loss": 0.7413,
+      "step": 7867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9012472242712865e-06,
+      "loss": 0.5427,
+      "step": 7868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8992989930787641e-06,
+      "loss": 0.5095,
+      "step": 7869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8973516558410532e-06,
+      "loss": 0.6362,
+      "step": 7870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8954052127730516e-06,
+      "loss": 0.367,
+      "step": 7871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8934596640895608e-06,
+      "loss": 0.5415,
+      "step": 7872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8915150100052826e-06,
+      "loss": 0.8512,
+      "step": 7873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8895712507348184e-06,
+      "loss": 0.5712,
+      "step": 7874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8876283864926737e-06,
+      "loss": 0.4783,
+      "step": 7875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8856864174932533e-06,
+      "loss": 0.6417,
+      "step": 7876
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8837453439508624e-06,
+      "loss": 0.654,
+      "step": 7877
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8818051660797099e-06,
+      "loss": 0.6677,
+      "step": 7878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8798658840939054e-06,
+      "loss": 0.6287,
+      "step": 7879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8779274982074569e-06,
+      "loss": 0.549,
+      "step": 7880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8759900086342764e-06,
+      "loss": 0.5035,
+      "step": 7881
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.874053415588175e-06,
+      "loss": 0.6046,
+      "step": 7882
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8721177192828688e-06,
+      "loss": 0.6823,
+      "step": 7883
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8701829199319655e-06,
+      "loss": 0.7574,
+      "step": 7884
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.868249017748992e-06,
+      "loss": 0.3315,
+      "step": 7885
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8663160129473535e-06,
+      "loss": 0.675,
+      "step": 7886
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8643839057403712e-06,
+      "loss": 0.6194,
+      "step": 7887
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8624526963412637e-06,
+      "loss": 0.6,
+      "step": 7888
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.860522384963147e-06,
+      "loss": 0.579,
+      "step": 7889
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8585929718190464e-06,
+      "loss": 0.5713,
+      "step": 7890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8566644571218794e-06,
+      "loss": 0.5255,
+      "step": 7891
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8547368410844713e-06,
+      "loss": 0.6073,
+      "step": 7892
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8528101239195394e-06,
+      "loss": 0.8497,
+      "step": 7893
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8508843058397085e-06,
+      "loss": 0.7309,
+      "step": 7894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8489593870575018e-06,
+      "loss": 0.5386,
+      "step": 7895
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8470353677853481e-06,
+      "loss": 0.4773,
+      "step": 7896
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.845112248235571e-06,
+      "loss": 0.9698,
+      "step": 7897
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8431900286203964e-06,
+      "loss": 0.439,
+      "step": 7898
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.841268709151951e-06,
+      "loss": 0.5375,
+      "step": 7899
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393482900422644e-06,
+      "loss": 0.6687,
+      "step": 7900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8374287715032623e-06,
+      "loss": 0.5534,
+      "step": 7901
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8355101537467757e-06,
+      "loss": 0.4593,
+      "step": 7902
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8335924369845326e-06,
+      "loss": 0.6919,
+      "step": 7903
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8316756214281639e-06,
+      "loss": 0.6907,
+      "step": 7904
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8297597072892004e-06,
+      "loss": 0.7011,
+      "step": 7905
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.827844694779073e-06,
+      "loss": 0.6902,
+      "step": 7906
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8259305841091124e-06,
+      "loss": 0.6971,
+      "step": 7907
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8240173754905533e-06,
+      "loss": 0.5396,
+      "step": 7908
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8221050691345255e-06,
+      "loss": 0.8442,
+      "step": 7909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8201936652520636e-06,
+      "loss": 0.4863,
+      "step": 7910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8182831640540989e-06,
+      "loss": 0.525,
+      "step": 7911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8163735657514703e-06,
+      "loss": 0.5026,
+      "step": 7912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8144648705549074e-06,
+      "loss": 0.6794,
+      "step": 7913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8125570786750456e-06,
+      "loss": 0.6669,
+      "step": 7914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8106501903224206e-06,
+      "loss": 0.4479,
+      "step": 7915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8087442057074666e-06,
+      "loss": 0.6181,
+      "step": 7916
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8068391250405171e-06,
+      "loss": 0.5402,
+      "step": 7917
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8049349485318124e-06,
+      "loss": 0.5278,
+      "step": 7918
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8030316763914879e-06,
+      "loss": 0.6136,
+      "step": 7919
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8011293088295767e-06,
+      "loss": 0.6012,
+      "step": 7920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7992278460560152e-06,
+      "loss": 0.5513,
+      "step": 7921
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7973272882806381e-06,
+      "loss": 0.6134,
+      "step": 7922
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7954276357131873e-06,
+      "loss": 0.7385,
+      "step": 7923
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7935288885632962e-06,
+      "loss": 0.5595,
+      "step": 7924
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.791631047040502e-06,
+      "loss": 0.5317,
+      "step": 7925
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.789734111354242e-06,
+      "loss": 0.5403,
+      "step": 7926
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7878380817138542e-06,
+      "loss": 0.4181,
+      "step": 7927
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7859429583285693e-06,
+      "loss": 0.6137,
+      "step": 7928
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7840487414075303e-06,
+      "loss": 0.3322,
+      "step": 7929
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7821554311597723e-06,
+      "loss": 0.5118,
+      "step": 7930
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7802630277942312e-06,
+      "loss": 0.5705,
+      "step": 7931
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7783715315197447e-06,
+      "loss": 0.5718,
+      "step": 7932
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7764809425450491e-06,
+      "loss": 0.6165,
+      "step": 7933
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7745912610787797e-06,
+      "loss": 0.7131,
+      "step": 7934
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.772702487329474e-06,
+      "loss": 0.515,
+      "step": 7935
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7708146215055678e-06,
+      "loss": 0.6164,
+      "step": 7936
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7689276638153963e-06,
+      "loss": 0.6692,
+      "step": 7937
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7670416144671964e-06,
+      "loss": 0.5745,
+      "step": 7938
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7651564736691018e-06,
+      "loss": 0.5097,
+      "step": 7939
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7632722416291504e-06,
+      "loss": 0.7127,
+      "step": 7940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7613889185552735e-06,
+      "loss": 0.7141,
+      "step": 7941
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7595065046553085e-06,
+      "loss": 0.6069,
+      "step": 7942
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7576250001369887e-06,
+      "loss": 0.5445,
+      "step": 7943
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7557444052079475e-06,
+      "loss": 0.4986,
+      "step": 7944
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7538647200757175e-06,
+      "loss": 0.462,
+      "step": 7945
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.751985944947735e-06,
+      "loss": 0.6328,
+      "step": 7946
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.750108080031333e-06,
+      "loss": 0.7247,
+      "step": 7947
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7482311255337392e-06,
+      "loss": 0.6246,
+      "step": 7948
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7463550816620879e-06,
+      "loss": 0.4939,
+      "step": 7949
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7444799486234077e-06,
+      "loss": 0.4638,
+      "step": 7950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7426057266246344e-06,
+      "loss": 0.7608,
+      "step": 7951
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7407324158725958e-06,
+      "loss": 0.9049,
+      "step": 7952
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7388600165740222e-06,
+      "loss": 0.5881,
+      "step": 7953
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7369885289355436e-06,
+      "loss": 0.3904,
+      "step": 7954
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7351179531636842e-06,
+      "loss": 0.7784,
+      "step": 7955
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7332482894648728e-06,
+      "loss": 0.6144,
+      "step": 7956
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.73137953804544e-06,
+      "loss": 0.5048,
+      "step": 7957
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.72951169911161e-06,
+      "loss": 0.7091,
+      "step": 7958
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.72764477286951e-06,
+      "loss": 0.7127,
+      "step": 7959
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.725778759525165e-06,
+      "loss": 0.3232,
+      "step": 7960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7239136592844974e-06,
+      "loss": 0.7043,
+      "step": 7961
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.722049472353332e-06,
+      "loss": 0.622,
+      "step": 7962
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7201861989373925e-06,
+      "loss": 0.6796,
+      "step": 7963
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7183238392422996e-06,
+      "loss": 0.5255,
+      "step": 7964
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7164623934735747e-06,
+      "loss": 0.6325,
+      "step": 7965
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.714601861836639e-06,
+      "loss": 0.5368,
+      "step": 7966
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7127422445368102e-06,
+      "loss": 0.4485,
+      "step": 7967
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.710883541779308e-06,
+      "loss": 0.5024,
+      "step": 7968
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7090257537692501e-06,
+      "loss": 0.7207,
+      "step": 7969
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7071688807116526e-06,
+      "loss": 0.7102,
+      "step": 7970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7053129228114318e-06,
+      "loss": 0.6576,
+      "step": 7971
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7034578802733991e-06,
+      "loss": 0.5571,
+      "step": 7972
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7016037533022733e-06,
+      "loss": 0.7962,
+      "step": 7973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6997505421026671e-06,
+      "loss": 0.5903,
+      "step": 7974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6978982468790883e-06,
+      "loss": 0.4437,
+      "step": 7975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6960468678359476e-06,
+      "loss": 0.7634,
+      "step": 7976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.694196405177556e-06,
+      "loss": 0.4714,
+      "step": 7977
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6923468591081194e-06,
+      "loss": 0.4938,
+      "step": 7978
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6904982298317484e-06,
+      "loss": 0.7138,
+      "step": 7979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6886505175524482e-06,
+      "loss": 0.4515,
+      "step": 7980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6868037224741253e-06,
+      "loss": 0.756,
+      "step": 7981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6849578448005776e-06,
+      "loss": 0.6071,
+      "step": 7982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6831128847355083e-06,
+      "loss": 0.5029,
+      "step": 7983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.681268842482523e-06,
+      "loss": 0.67,
+      "step": 7984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.679425718245119e-06,
+      "loss": 0.571,
+      "step": 7985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6775835122266937e-06,
+      "loss": 0.4869,
+      "step": 7986
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6757422246305465e-06,
+      "loss": 0.5947,
+      "step": 7987
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.673901855659873e-06,
+      "loss": 0.6974,
+      "step": 7988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6720624055177626e-06,
+      "loss": 0.4937,
+      "step": 7989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6702238744072142e-06,
+      "loss": 0.4272,
+      "step": 7990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6683862625311165e-06,
+      "loss": 0.6831,
+      "step": 7991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6665495700922606e-06,
+      "loss": 0.7446,
+      "step": 7992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6647137972933348e-06,
+      "loss": 0.4855,
+      "step": 7993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6628789443369265e-06,
+      "loss": 0.5993,
+      "step": 7994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6610450114255205e-06,
+      "loss": 0.4572,
+      "step": 7995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.659211998761503e-06,
+      "loss": 0.6581,
+      "step": 7996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.657379906547153e-06,
+      "loss": 0.5625,
+      "step": 7997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6555487349846545e-06,
+      "loss": 0.6846,
+      "step": 7998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6537184842760867e-06,
+      "loss": 0.4178,
+      "step": 7999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6518891546234229e-06,
+      "loss": 0.6926,
+      "step": 8000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.650060746228548e-06,
+      "loss": 0.8625,
+      "step": 8001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6482332592932293e-06,
+      "loss": 0.5527,
+      "step": 8002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6464066940191414e-06,
+      "loss": 0.5552,
+      "step": 8003
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6445810506078552e-06,
+      "loss": 0.5335,
+      "step": 8004
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6427563292608417e-06,
+      "loss": 0.4969,
+      "step": 8005
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6409325301794633e-06,
+      "loss": 0.5858,
+      "step": 8006
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.639109653564992e-06,
+      "loss": 0.66,
+      "step": 8007
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6372876996185928e-06,
+      "loss": 0.5416,
+      "step": 8008
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6354666685413212e-06,
+      "loss": 0.6336,
+      "step": 8009
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6336465605341411e-06,
+      "loss": 0.6638,
+      "step": 8010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6318273757979087e-06,
+      "loss": 0.515,
+      "step": 8011
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6300091145333841e-06,
+      "loss": 0.9295,
+      "step": 8012
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6281917769412204e-06,
+      "loss": 0.6581,
+      "step": 8013
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6263753632219714e-06,
+      "loss": 0.6324,
+      "step": 8014
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6245598735760858e-06,
+      "loss": 0.597,
+      "step": 8015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6227453082039157e-06,
+      "loss": 0.3745,
+      "step": 8016
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6209316673057008e-06,
+      "loss": 0.3914,
+      "step": 8017
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6191189510815942e-06,
+      "loss": 0.4676,
+      "step": 8018
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6173071597316336e-06,
+      "loss": 0.5452,
+      "step": 8019
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6154962934557616e-06,
+      "loss": 0.5993,
+      "step": 8020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6136863524538171e-06,
+      "loss": 0.7795,
+      "step": 8021
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6118773369255346e-06,
+      "loss": 0.6588,
+      "step": 8022
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6100692470705504e-06,
+      "loss": 0.5975,
+      "step": 8023
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6082620830883967e-06,
+      "loss": 0.5499,
+      "step": 8024
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6064558451785018e-06,
+      "loss": 0.6434,
+      "step": 8025
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.604650533540194e-06,
+      "loss": 0.5893,
+      "step": 8026
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6028461483726998e-06,
+      "loss": 0.6154,
+      "step": 8027
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.601042689875143e-06,
+      "loss": 0.6172,
+      "step": 8028
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5992401582465434e-06,
+      "loss": 0.489,
+      "step": 8029
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5974385536858194e-06,
+      "loss": 0.632,
+      "step": 8030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.595637876391789e-06,
+      "loss": 0.4972,
+      "step": 8031
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5938381265631665e-06,
+      "loss": 0.6738,
+      "step": 8032
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5920393043985593e-06,
+      "loss": 0.662,
+      "step": 8033
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5902414100964835e-06,
+      "loss": 0.421,
+      "step": 8034
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5884444438553437e-06,
+      "loss": 0.4346,
+      "step": 8035
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5866484058734455e-06,
+      "loss": 0.3279,
+      "step": 8036
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5848532963489882e-06,
+      "loss": 0.7606,
+      "step": 8037
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5830591154800724e-06,
+      "loss": 0.7712,
+      "step": 8038
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5812658634646939e-06,
+      "loss": 0.5595,
+      "step": 8039
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.579473540500751e-06,
+      "loss": 0.8497,
+      "step": 8040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.577682146786035e-06,
+      "loss": 0.4959,
+      "step": 8041
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5758916825182346e-06,
+      "loss": 0.5,
+      "step": 8042
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5741021478949402e-06,
+      "loss": 0.9457,
+      "step": 8043
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5723135431136283e-06,
+      "loss": 0.8575,
+      "step": 8044
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5705258683716896e-06,
+      "loss": 0.7604,
+      "step": 8045
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5687391238663985e-06,
+      "loss": 0.6814,
+      "step": 8046
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.566953309794933e-06,
+      "loss": 0.6729,
+      "step": 8047
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5651684263543665e-06,
+      "loss": 0.3836,
+      "step": 8048
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5633844737416714e-06,
+      "loss": 0.5965,
+      "step": 8049
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.561601452153716e-06,
+      "loss": 0.5019,
+      "step": 8050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5598193617872637e-06,
+      "loss": 0.6173,
+      "step": 8051
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5580382028389807e-06,
+      "loss": 0.6405,
+      "step": 8052
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5562579755054264e-06,
+      "loss": 0.6301,
+      "step": 8053
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5544786799830569e-06,
+      "loss": 0.5428,
+      "step": 8054
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5527003164682242e-06,
+      "loss": 0.5468,
+      "step": 8055
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.55092288515719e-06,
+      "loss": 0.6443,
+      "step": 8056
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5491463862460932e-06,
+      "loss": 0.5117,
+      "step": 8057
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5473708199309834e-06,
+      "loss": 0.6,
+      "step": 8058
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5455961864078027e-06,
+      "loss": 0.7216,
+      "step": 8059
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5438224858723928e-06,
+      "loss": 0.6295,
+      "step": 8060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5420497185204875e-06,
+      "loss": 0.5761,
+      "step": 8061
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.540277884547725e-06,
+      "loss": 0.4578,
+      "step": 8062
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.538506984149638e-06,
+      "loss": 0.5473,
+      "step": 8063
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.536737017521649e-06,
+      "loss": 0.5871,
+      "step": 8064
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5349679848590858e-06,
+      "loss": 0.6412,
+      "step": 8065
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5331998863571707e-06,
+      "loss": 0.6902,
+      "step": 8066
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5314327222110181e-06,
+      "loss": 0.4837,
+      "step": 8067
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5296664926156523e-06,
+      "loss": 0.6039,
+      "step": 8068
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5279011977659797e-06,
+      "loss": 0.4809,
+      "step": 8069
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5261368378568153e-06,
+      "loss": 0.7822,
+      "step": 8070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.524373413082859e-06,
+      "loss": 0.7059,
+      "step": 8071
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5226109236387155e-06,
+      "loss": 0.7317,
+      "step": 8072
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5208493697188876e-06,
+      "loss": 0.506,
+      "step": 8073
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.519088751517771e-06,
+      "loss": 0.6815,
+      "step": 8074
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5173290692296583e-06,
+      "loss": 0.7152,
+      "step": 8075
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5155703230487395e-06,
+      "loss": 0.3954,
+      "step": 8076
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.513812513169105e-06,
+      "loss": 0.5227,
+      "step": 8077
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5120556397847307e-06,
+      "loss": 0.3177,
+      "step": 8078
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5102997030895028e-06,
+      "loss": 0.5782,
+      "step": 8079
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5085447032771982e-06,
+      "loss": 0.6053,
+      "step": 8080
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5067906405414879e-06,
+      "loss": 0.3694,
+      "step": 8081
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5050375150759433e-06,
+      "loss": 0.3783,
+      "step": 8082
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.50328532707403e-06,
+      "loss": 0.6207,
+      "step": 8083
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.501534076729112e-06,
+      "loss": 0.6005,
+      "step": 8084
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4997837642344492e-06,
+      "loss": 0.5678,
+      "step": 8085
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4980343897831961e-06,
+      "loss": 0.565,
+      "step": 8086
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4962859535684071e-06,
+      "loss": 0.6255,
+      "step": 8087
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.49453845578303e-06,
+      "loss": 0.589,
+      "step": 8088
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4927918966199095e-06,
+      "loss": 0.583,
+      "step": 8089
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4910462762717938e-06,
+      "loss": 0.6131,
+      "step": 8090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4893015949313129e-06,
+      "loss": 0.725,
+      "step": 8091
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4875578527910061e-06,
+      "loss": 0.8057,
+      "step": 8092
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4858150500433033e-06,
+      "loss": 0.5059,
+      "step": 8093
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4840731868805302e-06,
+      "loss": 0.7015,
+      "step": 8094
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4823322634949133e-06,
+      "loss": 0.8211,
+      "step": 8095
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4805922800785733e-06,
+      "loss": 0.7206,
+      "step": 8096
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4788532368235264e-06,
+      "loss": 0.6777,
+      "step": 8097
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4771151339216817e-06,
+      "loss": 0.4768,
+      "step": 8098
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4753779715648498e-06,
+      "loss": 0.5461,
+      "step": 8099
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4736417499447343e-06,
+      "loss": 0.4066,
+      "step": 8100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4719064692529394e-06,
+      "loss": 0.4808,
+      "step": 8101
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4701721296809613e-06,
+      "loss": 0.8603,
+      "step": 8102
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.468438731420192e-06,
+      "loss": 0.4642,
+      "step": 8103
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4667062746619232e-06,
+      "loss": 0.5117,
+      "step": 8104
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4649747595973397e-06,
+      "loss": 0.5121,
+      "step": 8105
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4632441864175218e-06,
+      "loss": 0.5672,
+      "step": 8106
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4615145553134501e-06,
+      "loss": 0.5542,
+      "step": 8107
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4597858664759956e-06,
+      "loss": 0.5186,
+      "step": 8108
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.458058120095931e-06,
+      "loss": 0.5555,
+      "step": 8109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4563313163639193e-06,
+      "loss": 0.5335,
+      "step": 8110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4546054554705247e-06,
+      "loss": 0.4639,
+      "step": 8111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4528805376062028e-06,
+      "loss": 0.528,
+      "step": 8112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.45115656296131e-06,
+      "loss": 0.6465,
+      "step": 8113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4494335317260945e-06,
+      "loss": 0.38,
+      "step": 8114
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4477114440907002e-06,
+      "loss": 0.4686,
+      "step": 8115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4459903002451692e-06,
+      "loss": 0.627,
+      "step": 8116
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4442701003794435e-06,
+      "loss": 0.7579,
+      "step": 8117
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4425508446833514e-06,
+      "loss": 0.4554,
+      "step": 8118
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4408325333466211e-06,
+      "loss": 0.6508,
+      "step": 8119
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4391151665588798e-06,
+      "loss": 0.7586,
+      "step": 8120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4373987445096471e-06,
+      "loss": 0.3952,
+      "step": 8121
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4356832673883359e-06,
+      "loss": 0.4584,
+      "step": 8122
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4339687353842647e-06,
+      "loss": 0.6919,
+      "step": 8123
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4322551486866364e-06,
+      "loss": 0.5479,
+      "step": 8124
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4305425074845592e-06,
+      "loss": 0.7831,
+      "step": 8125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4288308119670247e-06,
+      "loss": 0.7248,
+      "step": 8126
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.427120062322932e-06,
+      "loss": 0.7218,
+      "step": 8127
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.425410258741068e-06,
+      "loss": 0.5212,
+      "step": 8128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4237014014101235e-06,
+      "loss": 0.6968,
+      "step": 8129
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4219934905186762e-06,
+      "loss": 0.7262,
+      "step": 8130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4202865262552045e-06,
+      "loss": 0.5551,
+      "step": 8131
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.418580508808084e-06,
+      "loss": 0.5113,
+      "step": 8132
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4168754383655724e-06,
+      "loss": 0.6051,
+      "step": 8133
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.415171315115844e-06,
+      "loss": 0.857,
+      "step": 8134
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.413468139246953e-06,
+      "loss": 0.6304,
+      "step": 8135
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.411765910946854e-06,
+      "loss": 0.6058,
+      "step": 8136
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4100646304033972e-06,
+      "loss": 0.6459,
+      "step": 8137
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4083642978043278e-06,
+      "loss": 0.606,
+      "step": 8138
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4066649133372867e-06,
+      "loss": 0.5788,
+      "step": 8139
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4049664771898096e-06,
+      "loss": 0.6745,
+      "step": 8140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4032689895493278e-06,
+      "loss": 0.811,
+      "step": 8141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.401572450603168e-06,
+      "loss": 0.5628,
+      "step": 8142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3998768605385527e-06,
+      "loss": 0.7021,
+      "step": 8143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3981822195425987e-06,
+      "loss": 0.6289,
+      "step": 8144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3964885278023177e-06,
+      "loss": 0.8441,
+      "step": 8145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.394795785504618e-06,
+      "loss": 0.5189,
+      "step": 8146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3931039928363033e-06,
+      "loss": 0.7612,
+      "step": 8147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.391413149984071e-06,
+      "loss": 0.7664,
+      "step": 8148
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3897232571345143e-06,
+      "loss": 0.4867,
+      "step": 8149
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3880343144741204e-06,
+      "loss": 0.6878,
+      "step": 8150
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.386346322189277e-06,
+      "loss": 0.6211,
+      "step": 8151
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.384659280466264e-06,
+      "loss": 0.5494,
+      "step": 8152
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.382973189491248e-06,
+      "loss": 0.6586,
+      "step": 8153
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3812880494503022e-06,
+      "loss": 0.4923,
+      "step": 8154
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3796038605293894e-06,
+      "loss": 0.6923,
+      "step": 8155
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3779206229143704e-06,
+      "loss": 0.5539,
+      "step": 8156
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3762383367909992e-06,
+      "loss": 0.4823,
+      "step": 8157
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.374557002344924e-06,
+      "loss": 0.7127,
+      "step": 8158
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3728766197616906e-06,
+      "loss": 0.5941,
+      "step": 8159
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.371197189226735e-06,
+      "loss": 0.9068,
+      "step": 8160
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3695187109253906e-06,
+      "loss": 0.619,
+      "step": 8161
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3678411850428908e-06,
+      "loss": 0.514,
+      "step": 8162
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3661646117643558e-06,
+      "loss": 0.6859,
+      "step": 8163
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.364488991274806e-06,
+      "loss": 0.5475,
+      "step": 8164
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.362814323759154e-06,
+      "loss": 0.4844,
+      "step": 8165
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3611406094022096e-06,
+      "loss": 0.5533,
+      "step": 8166
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3594678483886737e-06,
+      "loss": 0.7278,
+      "step": 8167
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3577960409031455e-06,
+      "loss": 0.6123,
+      "step": 8168
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3561251871301174e-06,
+      "loss": 0.5331,
+      "step": 8169
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3544552872539763e-06,
+      "loss": 0.5284,
+      "step": 8170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.352786341459006e-06,
+      "loss": 0.5538,
+      "step": 8171
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3511183499293824e-06,
+      "loss": 0.371,
+      "step": 8172
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3494513128491772e-06,
+      "loss": 0.9156,
+      "step": 8173
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3477852304023564e-06,
+      "loss": 0.475,
+      "step": 8174
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.346120102772781e-06,
+      "loss": 0.6493,
+      "step": 8175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3444559301442074e-06,
+      "loss": 0.629,
+      "step": 8176
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3427927127002827e-06,
+      "loss": 0.4646,
+      "step": 8177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.341130450624556e-06,
+      "loss": 0.7452,
+      "step": 8178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339469144100467e-06,
+      "loss": 0.7005,
+      "step": 8179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.337808793311346e-06,
+      "loss": 0.4134,
+      "step": 8180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3361493984404216e-06,
+      "loss": 0.6882,
+      "step": 8181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3344909596708177e-06,
+      "loss": 0.3825,
+      "step": 8182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3328334771855512e-06,
+      "loss": 0.3858,
+      "step": 8183
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3311769511675355e-06,
+      "loss": 0.5058,
+      "step": 8184
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3295213817995768e-06,
+      "loss": 0.7354,
+      "step": 8185
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3278667692643764e-06,
+      "loss": 0.6178,
+      "step": 8186
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3262131137445266e-06,
+      "loss": 0.5435,
+      "step": 8187
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.324560415422519e-06,
+      "loss": 0.5331,
+      "step": 8188
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.322908674480734e-06,
+      "loss": 0.562,
+      "step": 8189
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3212578911014561e-06,
+      "loss": 0.6098,
+      "step": 8190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3196080654668542e-06,
+      "loss": 0.5984,
+      "step": 8191
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3179591977589955e-06,
+      "loss": 0.53,
+      "step": 8192
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3163112881598418e-06,
+      "loss": 0.6337,
+      "step": 8193
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3146643368512479e-06,
+      "loss": 0.5841,
+      "step": 8194
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3130183440149635e-06,
+      "loss": 0.5436,
+      "step": 8195
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.311373309832633e-06,
+      "loss": 0.4774,
+      "step": 8196
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3097292344857949e-06,
+      "loss": 0.4671,
+      "step": 8197
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3080861181558812e-06,
+      "loss": 0.4735,
+      "step": 8198
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3064439610242175e-06,
+      "loss": 0.5879,
+      "step": 8199
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.304802763272025e-06,
+      "loss": 0.3805,
+      "step": 8200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3031625250804192e-06,
+      "loss": 0.548,
+      "step": 8201
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3015232466304085e-06,
+      "loss": 0.5214,
+      "step": 8202
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2998849281028959e-06,
+      "loss": 0.5249,
+      "step": 8203
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2982475696786777e-06,
+      "loss": 0.6147,
+      "step": 8204
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2966111715384434e-06,
+      "loss": 0.7349,
+      "step": 8205
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2949757338627844e-06,
+      "loss": 0.6613,
+      "step": 8206
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2933412568321746e-06,
+      "loss": 0.4281,
+      "step": 8207
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2917077406269873e-06,
+      "loss": 0.6072,
+      "step": 8208
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2900751854274918e-06,
+      "loss": 0.6762,
+      "step": 8209
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.288443591413846e-06,
+      "loss": 0.6438,
+      "step": 8210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.286812958766106e-06,
+      "loss": 0.606,
+      "step": 8211
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2851832876642223e-06,
+      "loss": 0.4416,
+      "step": 8212
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2835545782880376e-06,
+      "loss": 0.5329,
+      "step": 8213
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2819268308172895e-06,
+      "loss": 0.4244,
+      "step": 8214
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2803000454316051e-06,
+      "loss": 0.7929,
+      "step": 8215
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2786742223105064e-06,
+      "loss": 0.6584,
+      "step": 8216
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2770493616334189e-06,
+      "loss": 0.3765,
+      "step": 8217
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2754254635796503e-06,
+      "loss": 0.6635,
+      "step": 8218
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2738025283284072e-06,
+      "loss": 0.6855,
+      "step": 8219
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2721805560587875e-06,
+      "loss": 0.4255,
+      "step": 8220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2705595469497888e-06,
+      "loss": 0.5669,
+      "step": 8221
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2689395011802896e-06,
+      "loss": 0.5694,
+      "step": 8222
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2673204189290777e-06,
+      "loss": 0.5312,
+      "step": 8223
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.265702300374826e-06,
+      "loss": 0.5474,
+      "step": 8224
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2640851456961e-06,
+      "loss": 0.5279,
+      "step": 8225
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262468955071363e-06,
+      "loss": 0.6611,
+      "step": 8226
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2608537286789701e-06,
+      "loss": 0.5891,
+      "step": 8227
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2592394666971697e-06,
+      "loss": 0.6784,
+      "step": 8228
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2576261693041037e-06,
+      "loss": 0.7472,
+      "step": 8229
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2560138366778086e-06,
+      "loss": 0.6624,
+      "step": 8230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.254402468996213e-06,
+      "loss": 0.58,
+      "step": 8231
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2527920664371407e-06,
+      "loss": 0.6816,
+      "step": 8232
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.251182629178307e-06,
+      "loss": 0.7269,
+      "step": 8233
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249574157397322e-06,
+      "loss": 0.677,
+      "step": 8234
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2479666512716903e-06,
+      "loss": 0.6223,
+      "step": 8235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2463601109788058e-06,
+      "loss": 0.7498,
+      "step": 8236
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2447545366959623e-06,
+      "loss": 0.5145,
+      "step": 8237
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2431499286003378e-06,
+      "loss": 0.7146,
+      "step": 8238
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2415462868690165e-06,
+      "loss": 0.5548,
+      "step": 8239
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2399436116789643e-06,
+      "loss": 0.6151,
+      "step": 8240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2383419032070476e-06,
+      "loss": 0.6312,
+      "step": 8241
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2367411616300185e-06,
+      "loss": 0.6241,
+      "step": 8242
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.235141387124531e-06,
+      "loss": 0.634,
+      "step": 8243
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.233542579867124e-06,
+      "loss": 0.5822,
+      "step": 8244
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.231944740034241e-06,
+      "loss": 0.6456,
+      "step": 8245
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2303478678022097e-06,
+      "loss": 0.5345,
+      "step": 8246
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.228751963347251e-06,
+      "loss": 0.8048,
+      "step": 8247
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2271570268454857e-06,
+      "loss": 0.4782,
+      "step": 8248
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2255630584729184e-06,
+      "loss": 0.3409,
+      "step": 8249
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2239700584054515e-06,
+      "loss": 0.4507,
+      "step": 8250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2223780268188857e-06,
+      "loss": 0.5776,
+      "step": 8251
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.220786963888907e-06,
+      "loss": 0.5176,
+      "step": 8252
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2191968697910984e-06,
+      "loss": 0.5277,
+      "step": 8253
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.217607744700935e-06,
+      "loss": 0.5742,
+      "step": 8254
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2160195887937853e-06,
+      "loss": 0.6675,
+      "step": 8255
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2144324022449105e-06,
+      "loss": 0.5263,
+      "step": 8256
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2128461852294648e-06,
+      "loss": 0.7063,
+      "step": 8257
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.211260937922495e-06,
+      "loss": 0.5794,
+      "step": 8258
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209676660498943e-06,
+      "loss": 0.6016,
+      "step": 8259
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2080933531336403e-06,
+      "loss": 0.5757,
+      "step": 8260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206511016001314e-06,
+      "loss": 0.4399,
+      "step": 8261
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2049296492765827e-06,
+      "loss": 0.8034,
+      "step": 8262
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2033492531339608e-06,
+      "loss": 0.7058,
+      "step": 8263
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2017698277478507e-06,
+      "loss": 0.5945,
+      "step": 8264
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2001913732925507e-06,
+      "loss": 0.4922,
+      "step": 8265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1986138899422495e-06,
+      "loss": 0.7491,
+      "step": 8266
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1970373778710365e-06,
+      "loss": 0.5588,
+      "step": 8267
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.195461837252886e-06,
+      "loss": 0.32,
+      "step": 8268
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1938872682616642e-06,
+      "loss": 0.4573,
+      "step": 8269
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1923136710711346e-06,
+      "loss": 0.5695,
+      "step": 8270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1907410458549517e-06,
+      "loss": 0.6618,
+      "step": 8271
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.189169392786661e-06,
+      "loss": 0.4747,
+      "step": 8272
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1875987120397081e-06,
+      "loss": 0.5917,
+      "step": 8273
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1860290037874211e-06,
+      "loss": 0.4914,
+      "step": 8274
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1844602682030303e-06,
+      "loss": 0.452,
+      "step": 8275
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1828925054596485e-06,
+      "loss": 0.4173,
+      "step": 8276
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.181325715730286e-06,
+      "loss": 0.6019,
+      "step": 8277
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1797598991878512e-06,
+      "loss": 0.7126,
+      "step": 8278
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.178195056005138e-06,
+      "loss": 0.5352,
+      "step": 8279
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1766311863548352e-06,
+      "loss": 0.6057,
+      "step": 8280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1750682904095245e-06,
+      "loss": 0.5057,
+      "step": 8281
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1735063683416781e-06,
+      "loss": 0.6545,
+      "step": 8282
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1719454203236625e-06,
+      "loss": 0.5822,
+      "step": 8283
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1703854465277375e-06,
+      "loss": 0.7234,
+      "step": 8284
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1688264471260546e-06,
+      "loss": 0.7219,
+      "step": 8285
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1672684222906573e-06,
+      "loss": 0.5908,
+      "step": 8286
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1657113721934798e-06,
+      "loss": 0.5144,
+      "step": 8287
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1641552970063542e-06,
+      "loss": 0.4722,
+      "step": 8288
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1626001969009982e-06,
+      "loss": 0.4342,
+      "step": 8289
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.161046072049028e-06,
+      "loss": 0.6644,
+      "step": 8290
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1594929226219475e-06,
+      "loss": 0.6817,
+      "step": 8291
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1579407487911565e-06,
+      "loss": 0.4384,
+      "step": 8292
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1563895507279431e-06,
+      "loss": 0.6446,
+      "step": 8293
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.154839328603491e-06,
+      "loss": 0.7003,
+      "step": 8294
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1532900825888794e-06,
+      "loss": 0.5319,
+      "step": 8295
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1517418128550706e-06,
+      "loss": 0.4608,
+      "step": 8296
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1501945195729258e-06,
+      "loss": 0.5786,
+      "step": 8297
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1486482029131973e-06,
+      "loss": 0.481,
+      "step": 8298
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1471028630465287e-06,
+      "loss": 0.6397,
+      "step": 8299
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1455585001434555e-06,
+      "loss": 0.7479,
+      "step": 8300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1440151143744082e-06,
+      "loss": 0.5557,
+      "step": 8301
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1424727059097096e-06,
+      "loss": 0.299,
+      "step": 8302
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1409312749195679e-06,
+      "loss": 0.6374,
+      "step": 8303
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.139390821574089e-06,
+      "loss": 0.5035,
+      "step": 8304
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1378513460432695e-06,
+      "loss": 0.6142,
+      "step": 8305
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1363128484970021e-06,
+      "loss": 0.4989,
+      "step": 8306
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1347753291050656e-06,
+      "loss": 0.5609,
+      "step": 8307
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1332387880371342e-06,
+      "loss": 0.7102,
+      "step": 8308
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1317032254627725e-06,
+      "loss": 0.7264,
+      "step": 8309
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1301686415514401e-06,
+      "loss": 0.5736,
+      "step": 8310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1286350364724808e-06,
+      "loss": 0.7261,
+      "step": 8311
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1271024103951411e-06,
+      "loss": 0.4647,
+      "step": 8312
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1255707634885527e-06,
+      "loss": 0.4674,
+      "step": 8313
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1240400959217413e-06,
+      "loss": 0.6724,
+      "step": 8314
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.122510407863624e-06,
+      "loss": 0.55,
+      "step": 8315
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1209816994830093e-06,
+      "loss": 0.4939,
+      "step": 8316
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1194539709485985e-06,
+      "loss": 0.2937,
+      "step": 8317
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1179272224289838e-06,
+      "loss": 0.4555,
+      "step": 8318
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1164014540926516e-06,
+      "loss": 0.4166,
+      "step": 8319
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1148766661079768e-06,
+      "loss": 0.5652,
+      "step": 8320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1133528586432284e-06,
+      "loss": 0.4491,
+      "step": 8321
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1118300318665665e-06,
+      "loss": 0.6476,
+      "step": 8322
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.110308185946043e-06,
+      "loss": 0.848,
+      "step": 8323
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1087873210496026e-06,
+      "loss": 0.7965,
+      "step": 8324
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1072674373450786e-06,
+      "loss": 0.6143,
+      "step": 8325
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1057485350001995e-06,
+      "loss": 0.4947,
+      "step": 8326
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.104230614182581e-06,
+      "loss": 0.7753,
+      "step": 8327
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.102713675059739e-06,
+      "loss": 0.4413,
+      "step": 8328
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.101197717799074e-06,
+      "loss": 0.52,
+      "step": 8329
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0996827425678802e-06,
+      "loss": 0.5149,
+      "step": 8330
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0981687495333393e-06,
+      "loss": 0.5378,
+      "step": 8331
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0966557388625321e-06,
+      "loss": 0.5744,
+      "step": 8332
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0951437107224239e-06,
+      "loss": 0.7482,
+      "step": 8333
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.093632665279878e-06,
+      "loss": 0.7084,
+      "step": 8334
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0921226027016462e-06,
+      "loss": 0.5474,
+      "step": 8335
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0906135231543703e-06,
+      "loss": 0.4523,
+      "step": 8336
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0891054268045886e-06,
+      "loss": 0.658,
+      "step": 8337
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0875983138187197e-06,
+      "loss": 0.4671,
+      "step": 8338
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.086092184363089e-06,
+      "loss": 0.687,
+      "step": 8339
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0845870386039027e-06,
+      "loss": 0.5339,
+      "step": 8340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0830828767072621e-06,
+      "loss": 0.5853,
+      "step": 8341
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0815796988391591e-06,
+      "loss": 0.6421,
+      "step": 8342
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0800775051654766e-06,
+      "loss": 0.6099,
+      "step": 8343
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0785762958519907e-06,
+      "loss": 0.4138,
+      "step": 8344
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0770760710643668e-06,
+      "loss": 0.5447,
+      "step": 8345
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0755768309681635e-06,
+      "loss": 0.5621,
+      "step": 8346
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0740785757288296e-06,
+      "loss": 0.606,
+      "step": 8347
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072581305511705e-06,
+      "loss": 0.4797,
+      "step": 8348
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710850204820188e-06,
+      "loss": 0.5203,
+      "step": 8349
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0695897208049022e-06,
+      "loss": 0.6281,
+      "step": 8350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0680954066453597e-06,
+      "loss": 0.6571,
+      "step": 8351
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0666020781683018e-06,
+      "loss": 0.6117,
+      "step": 8352
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0651097355385243e-06,
+      "loss": 0.724,
+      "step": 8353
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0636183789207145e-06,
+      "loss": 0.8148,
+      "step": 8354
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0621280084794483e-06,
+      "loss": 0.6405,
+      "step": 8355
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0606386243792033e-06,
+      "loss": 0.5691,
+      "step": 8356
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.059150226784338e-06,
+      "loss": 0.7042,
+      "step": 8357
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0576628158591008e-06,
+      "loss": 0.3766,
+      "step": 8358
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0561763917676392e-06,
+      "loss": 0.4821,
+      "step": 8359
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.054690954673987e-06,
+      "loss": 0.4737,
+      "step": 8360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.053206504742067e-06,
+      "loss": 0.7608,
+      "step": 8361
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0517230421357016e-06,
+      "loss": 0.5807,
+      "step": 8362
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0502405670185956e-06,
+      "loss": 0.6442,
+      "step": 8363
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0487590795543502e-06,
+      "loss": 0.4722,
+      "step": 8364
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.047278579906451e-06,
+      "loss": 0.7044,
+      "step": 8365
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0457990682382801e-06,
+      "loss": 0.6205,
+      "step": 8366
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0443205447131123e-06,
+      "loss": 0.5991,
+      "step": 8367
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0428430094941078e-06,
+      "loss": 0.7134,
+      "step": 8368
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0413664627443221e-06,
+      "loss": 0.5511,
+      "step": 8369
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.039890904626698e-06,
+      "loss": 0.5801,
+      "step": 8370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0384163353040733e-06,
+      "loss": 0.5162,
+      "step": 8371
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.036942754939172e-06,
+      "loss": 0.7701,
+      "step": 8372
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.035470163694613e-06,
+      "loss": 0.67,
+      "step": 8373
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.033998561732904e-06,
+      "loss": 0.5558,
+      "step": 8374
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0325279492164441e-06,
+      "loss": 0.5841,
+      "step": 8375
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0310583263075236e-06,
+      "loss": 0.8959,
+      "step": 8376
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0295896931683214e-06,
+      "loss": 0.6903,
+      "step": 8377
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0281220499609112e-06,
+      "loss": 0.5665,
+      "step": 8378
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0266553968472536e-06,
+      "loss": 0.4573,
+      "step": 8379
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0251897339892014e-06,
+      "loss": 0.7694,
+      "step": 8380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0237250615484983e-06,
+      "loss": 0.5069,
+      "step": 8381
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0222613796867787e-06,
+      "loss": 0.6312,
+      "step": 8382
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0207986885655664e-06,
+      "loss": 0.651,
+      "step": 8383
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0193369883462822e-06,
+      "loss": 0.5794,
+      "step": 8384
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0178762791902263e-06,
+      "loss": 0.618,
+      "step": 8385
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0164165612585974e-06,
+      "loss": 0.5953,
+      "step": 8386
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0149578347124845e-06,
+      "loss": 0.7793,
+      "step": 8387
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0135000997128618e-06,
+      "loss": 0.644,
+      "step": 8388
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0120433564206033e-06,
+      "loss": 0.6243,
+      "step": 8389
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0105876049964657e-06,
+      "loss": 0.5479,
+      "step": 8390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0091328456011008e-06,
+      "loss": 0.4821,
+      "step": 8391
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0076790783950452e-06,
+      "loss": 0.5843,
+      "step": 8392
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.006226303538731e-06,
+      "loss": 0.6938,
+      "step": 8393
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0047745211924786e-06,
+      "loss": 0.4481,
+      "step": 8394
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0033237315165033e-06,
+      "loss": 0.7625,
+      "step": 8395
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.001873934670905e-06,
+      "loss": 0.5363,
+      "step": 8396
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0004251308156776e-06,
+      "loss": 0.5444,
+      "step": 8397
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989773201107023e-07,
+      "loss": 0.6617,
+      "step": 8398
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975305027157544e-07,
+      "loss": 0.6034,
+      "step": 8399
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960846787904964e-07,
+      "loss": 0.6928,
+      "step": 8400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946398484944831e-07,
+      "loss": 0.5926,
+      "step": 8401
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.931960119871586e-07,
+      "loss": 0.4405,
+      "step": 8402
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.917531694278592e-07,
+      "loss": 0.5642,
+      "step": 8403
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.6293,
+      "step": 8404
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88870466790124e-07,
+      "loss": 0.6044,
+      "step": 8405
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.8743060702981e-07,
+      "loss": 0.5544,
+      "step": 8406
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.859917418537624e-07,
+      "loss": 0.5584,
+      "step": 8407
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.84553871420768e-07,
+      "loss": 0.6477,
+      "step": 8408
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.831169958895037e-07,
+      "loss": 0.6427,
+      "step": 8409
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.816811154185334e-07,
+      "loss": 0.4429,
+      "step": 8410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.802462301663218e-07,
+      "loss": 0.6378,
+      "step": 8411
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.788123402912075e-07,
+      "loss": 0.4685,
+      "step": 8412
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77379445951432e-07,
+      "loss": 0.5154,
+      "step": 8413
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.759475473051206e-07,
+      "loss": 0.6023,
+      "step": 8414
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.745166445102916e-07,
+      "loss": 0.7849,
+      "step": 8415
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.730867377248522e-07,
+      "loss": 0.6082,
+      "step": 8416
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.716578271066024e-07,
+      "loss": 0.5965,
+      "step": 8417
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.702299128132286e-07,
+      "loss": 0.5751,
+      "step": 8418
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.688029950023104e-07,
+      "loss": 0.5405,
+      "step": 8419
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.673770738313116e-07,
+      "loss": 0.4652,
+      "step": 8420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.659521494575919e-07,
+      "loss": 0.4626,
+      "step": 8421
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.645282220383978e-07,
+      "loss": 0.6926,
+      "step": 8422
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.631052917308704e-07,
+      "loss": 0.7673,
+      "step": 8423
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61683358692036e-07,
+      "loss": 0.3804,
+      "step": 8424
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.602624230788127e-07,
+      "loss": 0.4424,
+      "step": 8425
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.588424850480093e-07,
+      "loss": 0.5614,
+      "step": 8426
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.574235447563185e-07,
+      "loss": 0.7599,
+      "step": 8427
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.560056023603338e-07,
+      "loss": 0.7438,
+      "step": 8428
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.545886580165287e-07,
+      "loss": 0.7163,
+      "step": 8429
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.531727118812717e-07,
+      "loss": 0.682,
+      "step": 8430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.5175776411082e-07,
+      "loss": 0.6653,
+      "step": 8431
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.503438148613208e-07,
+      "loss": 0.4857,
+      "step": 8432
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.489308642888107e-07,
+      "loss": 0.7276,
+      "step": 8433
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.475189125492157e-07,
+      "loss": 0.5769,
+      "step": 8434
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.461079597983513e-07,
+      "loss": 0.6027,
+      "step": 8435
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.446980061919253e-07,
+      "loss": 0.7633,
+      "step": 8436
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.43289051885532e-07,
+      "loss": 0.7569,
+      "step": 8437
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.418810970346571e-07,
+      "loss": 0.6551,
+      "step": 8438
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.404741417946783e-07,
+      "loss": 0.6857,
+      "step": 8439
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.390681863208573e-07,
+      "loss": 0.5981,
+      "step": 8440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.376632307683498e-07,
+      "loss": 0.5966,
+      "step": 8441
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.362592752922006e-07,
+      "loss": 0.6491,
+      "step": 8442
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.348563200473426e-07,
+      "loss": 0.2593,
+      "step": 8443
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334543651885975e-07,
+      "loss": 0.8686,
+      "step": 8444
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.320534108706824e-07,
+      "loss": 0.6237,
+      "step": 8445
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.306534572481996e-07,
+      "loss": 0.4895,
+      "step": 8446
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.292545044756385e-07,
+      "loss": 0.4678,
+      "step": 8447
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.278565527073812e-07,
+      "loss": 0.5821,
+      "step": 8448
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.26459602097699e-07,
+      "loss": 0.8496,
+      "step": 8449
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.250636528007551e-07,
+      "loss": 0.3679,
+      "step": 8450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.236687049705983e-07,
+      "loss": 0.5334,
+      "step": 8451
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.22274758761168e-07,
+      "loss": 0.7025,
+      "step": 8452
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.208818143262965e-07,
+      "loss": 0.6848,
+      "step": 8453
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.194898718196977e-07,
+      "loss": 0.5727,
+      "step": 8454
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.180989313949807e-07,
+      "loss": 0.7145,
+      "step": 8455
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.167089932056461e-07,
+      "loss": 0.5866,
+      "step": 8456
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.153200574050791e-07,
+      "loss": 0.6142,
+      "step": 8457
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.13932124146556e-07,
+      "loss": 0.6927,
+      "step": 8458
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.125451935832419e-07,
+      "loss": 0.6667,
+      "step": 8459
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.111592658681934e-07,
+      "loss": 0.4706,
+      "step": 8460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.097743411543536e-07,
+      "loss": 0.5502,
+      "step": 8461
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.08390419594557e-07,
+      "loss": 0.5505,
+      "step": 8462
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070075013415259e-07,
+      "loss": 0.3482,
+      "step": 8463
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.056255865478725e-07,
+      "loss": 0.6286,
+      "step": 8464
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.042446753660994e-07,
+      "loss": 0.6353,
+      "step": 8465
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.028647679485969e-07,
+      "loss": 0.4746,
+      "step": 8466
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.014858644476442e-07,
+      "loss": 0.5124,
+      "step": 8467
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.00107965015412e-07,
+      "loss": 0.6863,
+      "step": 8468
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.987310698039575e-07,
+      "loss": 0.4288,
+      "step": 8469
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.973551789652279e-07,
+      "loss": 0.506,
+      "step": 8470
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.959802926510597e-07,
+      "loss": 0.6717,
+      "step": 8471
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.946064110131824e-07,
+      "loss": 0.6563,
+      "step": 8472
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.932335342032094e-07,
+      "loss": 0.4934,
+      "step": 8473
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.918616623726429e-07,
+      "loss": 0.5495,
+      "step": 8474
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.904907956728759e-07,
+      "loss": 0.7299,
+      "step": 8475
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.891209342551931e-07,
+      "loss": 0.3764,
+      "step": 8476
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.877520782707627e-07,
+      "loss": 0.8713,
+      "step": 8477
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.863842278706491e-07,
+      "loss": 0.6855,
+      "step": 8478
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.850173832057995e-07,
+      "loss": 0.5589,
+      "step": 8479
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.836515444270555e-07,
+      "loss": 0.6862,
+      "step": 8480
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.822867116851397e-07,
+      "loss": 0.6302,
+      "step": 8481
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.809228851306706e-07,
+      "loss": 0.3215,
+      "step": 8482
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.795600649141511e-07,
+      "loss": 0.5753,
+      "step": 8483
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.78198251185981e-07,
+      "loss": 0.7038,
+      "step": 8484
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.768374440964412e-07,
+      "loss": 0.5783,
+      "step": 8485
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.754776437957025e-07,
+      "loss": 0.8008,
+      "step": 8486
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.741188504338272e-07,
+      "loss": 0.6251,
+      "step": 8487
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.727610641607642e-07,
+      "loss": 0.5592,
+      "step": 8488
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714042851263549e-07,
+      "loss": 0.7686,
+      "step": 8489
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.700485134803238e-07,
+      "loss": 0.6696,
+      "step": 8490
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.686937493722891e-07,
+      "loss": 0.4895,
+      "step": 8491
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.673399929517556e-07,
+      "loss": 0.3614,
+      "step": 8492
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.659872443681183e-07,
+      "loss": 0.7794,
+      "step": 8493
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.646355037706588e-07,
+      "loss": 0.7682,
+      "step": 8494
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.632847713085502e-07,
+      "loss": 0.5536,
+      "step": 8495
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.619350471308519e-07,
+      "loss": 0.5348,
+      "step": 8496
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.605863313865126e-07,
+      "loss": 0.5918,
+      "step": 8497
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.592386242243711e-07,
+      "loss": 0.5094,
+      "step": 8498
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.578919257931529e-07,
+      "loss": 1.0109,
+      "step": 8499
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.565462362414767e-07,
+      "loss": 0.5664,
+      "step": 8500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.552015557178417e-07,
+      "loss": 0.7014,
+      "step": 8501
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.538578843706425e-07,
+      "loss": 0.5462,
+      "step": 8502
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.525152223481603e-07,
+      "loss": 0.6604,
+      "step": 8503
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.511735697985646e-07,
+      "loss": 0.5286,
+      "step": 8504
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.498329268699124e-07,
+      "loss": 0.4968,
+      "step": 8505
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.484932937101531e-07,
+      "loss": 0.4821,
+      "step": 8506
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.47154670467123e-07,
+      "loss": 0.6357,
+      "step": 8507
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.458170572885449e-07,
+      "loss": 0.6432,
+      "step": 8508
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.444804543220309e-07,
+      "loss": 0.621,
+      "step": 8509
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.431448617150794e-07,
+      "loss": 0.617,
+      "step": 8510
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.418102796150851e-07,
+      "loss": 0.6683,
+      "step": 8511
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.404767081693244e-07,
+      "loss": 0.9257,
+      "step": 8512
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.391441475249628e-07,
+      "loss": 0.7256,
+      "step": 8513
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.378125978290564e-07,
+      "loss": 0.7535,
+      "step": 8514
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.364820592285494e-07,
+      "loss": 0.4626,
+      "step": 8515
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.351525318702703e-07,
+      "loss": 0.6277,
+      "step": 8516
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.338240159009425e-07,
+      "loss": 0.423,
+      "step": 8517
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.324965114671746e-07,
+      "loss": 0.6052,
+      "step": 8518
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.311700187154637e-07,
+      "loss": 0.5576,
+      "step": 8519
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.298445377921938e-07,
+      "loss": 0.5263,
+      "step": 8520
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.28520068843639e-07,
+      "loss": 0.7131,
+      "step": 8521
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.271966120159624e-07,
+      "loss": 0.5362,
+      "step": 8522
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.25874167455214e-07,
+      "loss": 0.4648,
+      "step": 8523
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.245527353073323e-07,
+      "loss": 0.8545,
+      "step": 8524
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.232323157181433e-07,
+      "loss": 0.7768,
+      "step": 8525
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.219129088333644e-07,
+      "loss": 0.3742,
+      "step": 8526
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.205945147985972e-07,
+      "loss": 0.9133,
+      "step": 8527
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.192771337593341e-07,
+      "loss": 0.7949,
+      "step": 8528
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.179607658609557e-07,
+      "loss": 0.3929,
+      "step": 8529
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.166454112487287e-07,
+      "loss": 0.6189,
+      "step": 8530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.153310700678096e-07,
+      "loss": 0.6551,
+      "step": 8531
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.140177424632423e-07,
+      "loss": 0.7271,
+      "step": 8532
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.127054285799618e-07,
+      "loss": 0.7123,
+      "step": 8533
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.113941285627869e-07,
+      "loss": 0.5602,
+      "step": 8534
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.100838425564294e-07,
+      "loss": 0.5712,
+      "step": 8535
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.087745707054817e-07,
+      "loss": 0.6194,
+      "step": 8536
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.074663131544313e-07,
+      "loss": 0.3793,
+      "step": 8537
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.061590700476484e-07,
+      "loss": 0.5517,
+      "step": 8538
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.048528415293988e-07,
+      "loss": 0.4761,
+      "step": 8539
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.035476277438292e-07,
+      "loss": 0.6216,
+      "step": 8540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.022434288349767e-07,
+      "loss": 0.7956,
+      "step": 8541
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.009402449467684e-07,
+      "loss": 0.4078,
+      "step": 8542
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.99638076223015e-07,
+      "loss": 0.5507,
+      "step": 8543
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.983369228074156e-07,
+      "loss": 0.4988,
+      "step": 8544
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.970367848435645e-07,
+      "loss": 0.4401,
+      "step": 8545
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.957376624749369e-07,
+      "loss": 0.3755,
+      "step": 8546
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.944395558448959e-07,
+      "loss": 0.4719,
+      "step": 8547
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.931424650966968e-07,
+      "loss": 0.5726,
+      "step": 8548
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.918463903734796e-07,
+      "loss": 0.6025,
+      "step": 8549
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.905513318182722e-07,
+      "loss": 0.718,
+      "step": 8550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.892572895739914e-07,
+      "loss": 0.4764,
+      "step": 8551
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.879642637834417e-07,
+      "loss": 0.8628,
+      "step": 8552
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.866722545893157e-07,
+      "loss": 0.571,
+      "step": 8553
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.853812621341927e-07,
+      "loss": 0.6554,
+      "step": 8554
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.840912865605399e-07,
+      "loss": 0.3952,
+      "step": 8555
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.828023280107155e-07,
+      "loss": 0.6502,
+      "step": 8556
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.815143866269592e-07,
+      "loss": 0.5163,
+      "step": 8557
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.80227462551405e-07,
+      "loss": 0.8098,
+      "step": 8558
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.789415559260705e-07,
+      "loss": 0.5155,
+      "step": 8559
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.776566668928609e-07,
+      "loss": 0.6735,
+      "step": 8560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.763727955935729e-07,
+      "loss": 0.6182,
+      "step": 8561
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.750899421698899e-07,
+      "loss": 0.6437,
+      "step": 8562
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.738081067633762e-07,
+      "loss": 0.5423,
+      "step": 8563
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.725272895154934e-07,
+      "loss": 0.682,
+      "step": 8564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.712474905675838e-07,
+      "loss": 0.6143,
+      "step": 8565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.699687100608799e-07,
+      "loss": 0.5955,
+      "step": 8566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.686909481365034e-07,
+      "loss": 0.6761,
+      "step": 8567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.674142049354627e-07,
+      "loss": 0.515,
+      "step": 8568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.661384805986538e-07,
+      "loss": 0.8678,
+      "step": 8569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.648637752668553e-07,
+      "loss": 0.4136,
+      "step": 8570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.635900890807391e-07,
+      "loss": 0.5332,
+      "step": 8571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.62317422180866e-07,
+      "loss": 0.6387,
+      "step": 8572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.610457747076783e-07,
+      "loss": 1.0639,
+      "step": 8573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.597751468015113e-07,
+      "loss": 0.422,
+      "step": 8574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.585055386025841e-07,
+      "loss": 0.5657,
+      "step": 8575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.572369502510047e-07,
+      "loss": 0.757,
+      "step": 8576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.559693818867686e-07,
+      "loss": 0.5508,
+      "step": 8577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.547028336497586e-07,
+      "loss": 0.7092,
+      "step": 8578
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.534373056797451e-07,
+      "loss": 0.494,
+      "step": 8579
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.521727981163862e-07,
+      "loss": 0.7997,
+      "step": 8580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.509093110992261e-07,
+      "loss": 0.6204,
+      "step": 8581
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.496468447676975e-07,
+      "loss": 0.5766,
+      "step": 8582
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.483853992611201e-07,
+      "loss": 0.3813,
+      "step": 8583
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.471249747187003e-07,
+      "loss": 0.5331,
+      "step": 8584
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.458655712795348e-07,
+      "loss": 0.4153,
+      "step": 8585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.446071890826023e-07,
+      "loss": 0.4593,
+      "step": 8586
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.43349828266775e-07,
+      "loss": 0.6159,
+      "step": 8587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.420934889708054e-07,
+      "loss": 0.4834,
+      "step": 8588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.408381713333423e-07,
+      "loss": 0.716,
+      "step": 8589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.39583875492913e-07,
+      "loss": 0.6451,
+      "step": 8590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.383306015879366e-07,
+      "loss": 0.7859,
+      "step": 8591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.370783497567169e-07,
+      "loss": 0.5552,
+      "step": 8592
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.358271201374479e-07,
+      "loss": 0.5222,
+      "step": 8593
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.34576912868209e-07,
+      "loss": 0.5829,
+      "step": 8594
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.333277280869688e-07,
+      "loss": 0.7546,
+      "step": 8595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.320795659315794e-07,
+      "loss": 0.299,
+      "step": 8596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.308324265397837e-07,
+      "loss": 0.2567,
+      "step": 8597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.295863100492073e-07,
+      "loss": 0.546,
+      "step": 8598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.283412165973669e-07,
+      "loss": 0.3954,
+      "step": 8599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.270971463216659e-07,
+      "loss": 0.7564,
+      "step": 8600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.258540993593943e-07,
+      "loss": 0.6537,
+      "step": 8601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.246120758477282e-07,
+      "loss": 0.4984,
+      "step": 8602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.233710759237311e-07,
+      "loss": 0.7174,
+      "step": 8603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.221310997243547e-07,
+      "loss": 0.8471,
+      "step": 8604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.208921473864338e-07,
+      "loss": 0.5894,
+      "step": 8605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.196542190466982e-07,
+      "loss": 0.5068,
+      "step": 8606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.184173148417561e-07,
+      "loss": 0.7035,
+      "step": 8607
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.171814349081086e-07,
+      "loss": 0.5625,
+      "step": 8608
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.159465793821396e-07,
+      "loss": 0.5466,
+      "step": 8609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.147127484001237e-07,
+      "loss": 0.4809,
+      "step": 8610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.134799420982186e-07,
+      "loss": 0.8651,
+      "step": 8611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.122481606124732e-07,
+      "loss": 0.4158,
+      "step": 8612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110174040788187e-07,
+      "loss": 0.4815,
+      "step": 8613
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.097876726330776e-07,
+      "loss": 0.5873,
+      "step": 8614
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08558966410956e-07,
+      "loss": 0.5324,
+      "step": 8615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.073312855480485e-07,
+      "loss": 0.5122,
+      "step": 8616
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.061046301798358e-07,
+      "loss": 0.6279,
+      "step": 8617
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.048790004416861e-07,
+      "loss": 0.5746,
+      "step": 8618
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.036543964688536e-07,
+      "loss": 0.5015,
+      "step": 8619
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.024308183964801e-07,
+      "loss": 0.4053,
+      "step": 8620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.012082663595932e-07,
+      "loss": 0.6891,
+      "step": 8621
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.999867404931094e-07,
+      "loss": 0.5728,
+      "step": 8622
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.987662409318296e-07,
+      "loss": 0.6521,
+      "step": 8623
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.975467678104442e-07,
+      "loss": 0.5711,
+      "step": 8624
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963283212635252e-07,
+      "loss": 0.6248,
+      "step": 8625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.951109014255352e-07,
+      "loss": 0.6091,
+      "step": 8626
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.938945084308224e-07,
+      "loss": 0.4836,
+      "step": 8627
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.926791424136259e-07,
+      "loss": 0.5811,
+      "step": 8628
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.91464803508064e-07,
+      "loss": 0.4731,
+      "step": 8629
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.902514918481473e-07,
+      "loss": 0.5291,
+      "step": 8630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.89039207567771e-07,
+      "loss": 0.5785,
+      "step": 8631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.878279508007146e-07,
+      "loss": 0.7739,
+      "step": 8632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.866177216806491e-07,
+      "loss": 0.565,
+      "step": 8633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.854085203411287e-07,
+      "loss": 0.4141,
+      "step": 8634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.842003469155955e-07,
+      "loss": 0.6102,
+      "step": 8635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.829932015373785e-07,
+      "loss": 0.6442,
+      "step": 8636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.817870843396912e-07,
+      "loss": 0.6433,
+      "step": 8637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.805819954556347e-07,
+      "loss": 0.5475,
+      "step": 8638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.793779350181984e-07,
+      "loss": 0.745,
+      "step": 8639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.781749031602558e-07,
+      "loss": 0.7768,
+      "step": 8640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.769729000145675e-07,
+      "loss": 0.6332,
+      "step": 8641
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.757719257137807e-07,
+      "loss": 0.6533,
+      "step": 8642
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.745719803904283e-07,
+      "loss": 0.4191,
+      "step": 8643
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.733730641769353e-07,
+      "loss": 0.4953,
+      "step": 8644
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72175177205604e-07,
+      "loss": 0.7209,
+      "step": 8645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.709783196086284e-07,
+      "loss": 0.5011,
+      "step": 8646
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.697824915180873e-07,
+      "loss": 0.7907,
+      "step": 8647
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.685876930659474e-07,
+      "loss": 0.5852,
+      "step": 8648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673939243840599e-07,
+      "loss": 0.5855,
+      "step": 8649
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.662011856041661e-07,
+      "loss": 0.5981,
+      "step": 8650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.650094768578919e-07,
+      "loss": 0.6137,
+      "step": 8651
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.63818798276743e-07,
+      "loss": 0.5309,
+      "step": 8652
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.626291499921211e-07,
+      "loss": 0.5854,
+      "step": 8653
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.614405321353068e-07,
+      "loss": 0.8084,
+      "step": 8654
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.602529448374751e-07,
+      "loss": 0.4668,
+      "step": 8655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.590663882296788e-07,
+      "loss": 0.9647,
+      "step": 8656
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.578808624428635e-07,
+      "loss": 0.5496,
+      "step": 8657
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.566963676078564e-07,
+      "loss": 0.6795,
+      "step": 8658
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.555129038553732e-07,
+      "loss": 0.8853,
+      "step": 8659
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.543304713160115e-07,
+      "loss": 0.5666,
+      "step": 8660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.531490701202648e-07,
+      "loss": 0.5525,
+      "step": 8661
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.519687003985053e-07,
+      "loss": 0.532,
+      "step": 8662
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.507893622809914e-07,
+      "loss": 0.7403,
+      "step": 8663
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.496110558978708e-07,
+      "loss": 0.6226,
+      "step": 8664
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.484337813791741e-07,
+      "loss": 0.7104,
+      "step": 8665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.472575388548208e-07,
+      "loss": 0.5295,
+      "step": 8666
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.460823284546158e-07,
+      "loss": 0.6499,
+      "step": 8667
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.449081503082488e-07,
+      "loss": 0.7543,
+      "step": 8668
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.437350045452973e-07,
+      "loss": 0.6298,
+      "step": 8669
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.425628912952242e-07,
+      "loss": 0.7081,
+      "step": 8670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413918106873774e-07,
+      "loss": 0.3577,
+      "step": 8671
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.402217628509922e-07,
+      "loss": 0.452,
+      "step": 8672
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.390527479151887e-07,
+      "loss": 0.4786,
+      "step": 8673
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37884766008976e-07,
+      "loss": 0.6935,
+      "step": 8674
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.367178172612454e-07,
+      "loss": 0.4958,
+      "step": 8675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.355519018007761e-07,
+      "loss": 0.4579,
+      "step": 8676
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.343870197562307e-07,
+      "loss": 0.5858,
+      "step": 8677
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.332231712561665e-07,
+      "loss": 0.4437,
+      "step": 8678
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.32060356429014e-07,
+      "loss": 0.5264,
+      "step": 8679
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.308985754030983e-07,
+      "loss": 0.5605,
+      "step": 8680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.297378283066269e-07,
+      "loss": 0.3381,
+      "step": 8681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.28578115267694e-07,
+      "loss": 0.458,
+      "step": 8682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.274194364142838e-07,
+      "loss": 0.5343,
+      "step": 8683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.262617918742586e-07,
+      "loss": 0.6925,
+      "step": 8684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.25105181775374e-07,
+      "loss": 0.5239,
+      "step": 8685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.239496062452643e-07,
+      "loss": 0.6975,
+      "step": 8686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.227950654114557e-07,
+      "loss": 0.6144,
+      "step": 8687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.216415594013547e-07,
+      "loss": 0.5818,
+      "step": 8688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.20489088342261e-07,
+      "loss": 1.3142,
+      "step": 8689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.193376523613537e-07,
+      "loss": 0.7403,
+      "step": 8690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.181872515857002e-07,
+      "loss": 0.5609,
+      "step": 8691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.170378861422522e-07,
+      "loss": 0.6886,
+      "step": 8692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.158895561578493e-07,
+      "loss": 0.6082,
+      "step": 8693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147422617592169e-07,
+      "loss": 0.4682,
+      "step": 8694
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.135960030729616e-07,
+      "loss": 0.7251,
+      "step": 8695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.12450780225582e-07,
+      "loss": 0.2993,
+      "step": 8696
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.113065933434592e-07,
+      "loss": 0.4277,
+      "step": 8697
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.101634425528579e-07,
+      "loss": 0.6256,
+      "step": 8698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.090213279799328e-07,
+      "loss": 0.6227,
+      "step": 8699
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.07880249750723e-07,
+      "loss": 0.5925,
+      "step": 8700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.0674020799115e-07,
+      "loss": 0.4418,
+      "step": 8701
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.0674020799115e-07,
+      "loss": 0.7474,
+      "step": 8702
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.056012028270242e-07,
+      "loss": 0.6744,
+      "step": 8703
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.044632343840429e-07,
+      "loss": 0.8587,
+      "step": 8704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.033263027877823e-07,
+      "loss": 0.6872,
+      "step": 8705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.021904081637142e-07,
+      "loss": 0.7228,
+      "step": 8706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.010555506371874e-07,
+      "loss": 0.5408,
+      "step": 8707
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.999217303334404e-07,
+      "loss": 0.7162,
+      "step": 8708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.987889473775955e-07,
+      "loss": 0.416,
+      "step": 8709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.976572018946602e-07,
+      "loss": 0.5665,
+      "step": 8710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.965264940095294e-07,
+      "loss": 0.4754,
+      "step": 8711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.953968238469843e-07,
+      "loss": 0.4406,
+      "step": 8712
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.942681915316895e-07,
+      "loss": 0.5371,
+      "step": 8713
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.931405971881953e-07,
+      "loss": 0.6611,
+      "step": 8714
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.920140409409359e-07,
+      "loss": 0.563,
+      "step": 8715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.908885229142336e-07,
+      "loss": 0.5184,
+      "step": 8716
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.897640432322938e-07,
+      "loss": 0.6134,
+      "step": 8717
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.886406020192126e-07,
+      "loss": 0.5926,
+      "step": 8718
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875181993989643e-07,
+      "loss": 0.7022,
+      "step": 8719
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.863968354954142e-07,
+      "loss": 0.5385,
+      "step": 8720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.852765104323099e-07,
+      "loss": 0.5958,
+      "step": 8721
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.841572243332827e-07,
+      "loss": 0.6221,
+      "step": 8722
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.830389773218547e-07,
+      "loss": 0.5339,
+      "step": 8723
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.819217695214296e-07,
+      "loss": 0.6465,
+      "step": 8724
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.808056010552965e-07,
+      "loss": 0.6699,
+      "step": 8725
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.796904720466312e-07,
+      "loss": 0.5315,
+      "step": 8726
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.785763826184931e-07,
+      "loss": 0.6463,
+      "step": 8727
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.774633328938284e-07,
+      "loss": 0.6534,
+      "step": 8728
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.763513229954687e-07,
+      "loss": 0.5224,
+      "step": 8729
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.752403530461281e-07,
+      "loss": 0.4955,
+      "step": 8730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.741304231684097e-07,
+      "loss": 0.625,
+      "step": 8731
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.730215334848e-07,
+      "loss": 0.7685,
+      "step": 8732
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.719136841176687e-07,
+      "loss": 0.5232,
+      "step": 8733
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.70806875189277e-07,
+      "loss": 0.7331,
+      "step": 8734
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.69701106821764e-07,
+      "loss": 0.7738,
+      "step": 8735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.685963791371573e-07,
+      "loss": 0.6092,
+      "step": 8736
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.674926922573687e-07,
+      "loss": 0.8721,
+      "step": 8737
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.663900463041972e-07,
+      "loss": 0.4561,
+      "step": 8738
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.652884413993243e-07,
+      "loss": 0.7246,
+      "step": 8739
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.641878776643195e-07,
+      "loss": 0.3527,
+      "step": 8740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.630883552206368e-07,
+      "loss": 0.4556,
+      "step": 8741
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.619898741896101e-07,
+      "loss": 0.491,
+      "step": 8742
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.60892434692466e-07,
+      "loss": 0.7628,
+      "step": 8743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.597960368503097e-07,
+      "loss": 0.6237,
+      "step": 8744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.587006807841378e-07,
+      "loss": 0.5781,
+      "step": 8745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.576063666148268e-07,
+      "loss": 0.5712,
+      "step": 8746
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.565130944631403e-07,
+      "loss": 0.8542,
+      "step": 8747
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.554208644497283e-07,
+      "loss": 0.5598,
+      "step": 8748
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.543296766951212e-07,
+      "loss": 0.7061,
+      "step": 8749
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.532395313197369e-07,
+      "loss": 0.7411,
+      "step": 8750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.521504284438816e-07,
+      "loss": 0.4971,
+      "step": 8751
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.510623681877425e-07,
+      "loss": 0.5632,
+      "step": 8752
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.499753506713923e-07,
+      "loss": 0.5817,
+      "step": 8753
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.488893760147896e-07,
+      "loss": 0.6005,
+      "step": 8754
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.478044443377761e-07,
+      "loss": 0.7185,
+      "step": 8755
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.467205557600819e-07,
+      "loss": 0.7321,
+      "step": 8756
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.456377104013166e-07,
+      "loss": 0.4559,
+      "step": 8757
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.445559083809815e-07,
+      "loss": 0.5857,
+      "step": 8758
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.434751498184565e-07,
+      "loss": 0.4446,
+      "step": 8759
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.423954348330107e-07,
+      "loss": 0.5535,
+      "step": 8760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.413167635437943e-07,
+      "loss": 0.8394,
+      "step": 8761
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.402391360698456e-07,
+      "loss": 0.8169,
+      "step": 8762
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.391625525300869e-07,
+      "loss": 0.3644,
+      "step": 8763
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.380870130433247e-07,
+      "loss": 0.6225,
+      "step": 8764
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.370125177282493e-07,
+      "loss": 0.7981,
+      "step": 8765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.359390667034358e-07,
+      "loss": 0.6298,
+      "step": 8766
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.348666600873487e-07,
+      "loss": 0.5344,
+      "step": 8767
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.337952979983341e-07,
+      "loss": 0.6858,
+      "step": 8768
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.327249805546176e-07,
+      "loss": 0.7693,
+      "step": 8769
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.316557078743179e-07,
+      "loss": 0.5525,
+      "step": 8770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.305874800754329e-07,
+      "loss": 0.5812,
+      "step": 8771
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.295202972758473e-07,
+      "loss": 0.6708,
+      "step": 8772
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.284541595933335e-07,
+      "loss": 0.5121,
+      "step": 8773
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.273890671455417e-07,
+      "loss": 0.4384,
+      "step": 8774
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.263250200500137e-07,
+      "loss": 0.4683,
+      "step": 8775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.252620184241697e-07,
+      "loss": 0.6686,
+      "step": 8776
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.242000623853183e-07,
+      "loss": 0.4739,
+      "step": 8777
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.231391520506501e-07,
+      "loss": 0.5702,
+      "step": 8778
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.22079287537246e-07,
+      "loss": 0.3706,
+      "step": 8779
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210204689620657e-07,
+      "loss": 0.6107,
+      "step": 8780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.199626964419557e-07,
+      "loss": 0.5656,
+      "step": 8781
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.189059700936472e-07,
+      "loss": 0.6865,
+      "step": 8782
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.178502900337535e-07,
+      "loss": 0.6111,
+      "step": 8783
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.167956563787768e-07,
+      "loss": 0.6083,
+      "step": 8784
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.157420692450998e-07,
+      "loss": 0.5041,
+      "step": 8785
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146895287489917e-07,
+      "loss": 0.7004,
+      "step": 8786
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.136380350066062e-07,
+      "loss": 0.4827,
+      "step": 8787
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.125875881339815e-07,
+      "loss": 0.5556,
+      "step": 8788
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.115381882470394e-07,
+      "loss": 0.6101,
+      "step": 8789
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.104898354615862e-07,
+      "loss": 0.4656,
+      "step": 8790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.094425298933136e-07,
+      "loss": 0.5799,
+      "step": 8791
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.083962716577984e-07,
+      "loss": 0.5476,
+      "step": 8792
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.07351060870499e-07,
+      "loss": 0.7934,
+      "step": 8793
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.063068976467588e-07,
+      "loss": 0.5931,
+      "step": 8794
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.052637821018114e-07,
+      "loss": 0.6099,
+      "step": 8795
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.042217143507655e-07,
+      "loss": 0.6216,
+      "step": 8796
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.031806945086204e-07,
+      "loss": 0.5904,
+      "step": 8797
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.021407226902575e-07,
+      "loss": 0.5983,
+      "step": 8798
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.01101799010445e-07,
+      "loss": 0.4694,
+      "step": 8799
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.00063923583829e-07,
+      "loss": 0.2903,
+      "step": 8800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9902709652495e-07,
+      "loss": 0.4177,
+      "step": 8801
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.979913179482254e-07,
+      "loss": 0.5793,
+      "step": 8802
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.969565879679594e-07,
+      "loss": 0.419,
+      "step": 8803
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.959229066983373e-07,
+      "loss": 0.6101,
+      "step": 8804
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.948902742534312e-07,
+      "loss": 0.7215,
+      "step": 8805
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.938586907472009e-07,
+      "loss": 0.4691,
+      "step": 8806
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.928281562934856e-07,
+      "loss": 0.5578,
+      "step": 8807
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.917986710060086e-07,
+      "loss": 0.6101,
+      "step": 8808
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.907702349983812e-07,
+      "loss": 0.5628,
+      "step": 8809
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.897428483840971e-07,
+      "loss": 0.6491,
+      "step": 8810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.887165112765291e-07,
+      "loss": 0.416,
+      "step": 8811
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.876912237889442e-07,
+      "loss": 0.5112,
+      "step": 8812
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.866669860344853e-07,
+      "loss": 0.6706,
+      "step": 8813
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.856437981261841e-07,
+      "loss": 0.6183,
+      "step": 8814
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.846216601769538e-07,
+      "loss": 0.4874,
+      "step": 8815
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.836005722995929e-07,
+      "loss": 0.5696,
+      "step": 8816
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825805346067847e-07,
+      "loss": 0.8247,
+      "step": 8817
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.815615472110935e-07,
+      "loss": 0.7229,
+      "step": 8818
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805436102249716e-07,
+      "loss": 0.4853,
+      "step": 8819
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.795267237607537e-07,
+      "loss": 0.4545,
+      "step": 8820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.785108879306577e-07,
+      "loss": 0.6885,
+      "step": 8821
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.77496102846785e-07,
+      "loss": 0.5057,
+      "step": 8822
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.764823686211273e-07,
+      "loss": 0.6187,
+      "step": 8823
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.754696853655516e-07,
+      "loss": 0.5345,
+      "step": 8824
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7445805319181305e-07,
+      "loss": 0.5977,
+      "step": 8825
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7344747221155117e-07,
+      "loss": 0.6413,
+      "step": 8826
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7243794253628796e-07,
+      "loss": 0.6403,
+      "step": 8827
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.714294642774331e-07,
+      "loss": 0.5216,
+      "step": 8828
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7042203754627424e-07,
+      "loss": 0.4412,
+      "step": 8829
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.694156624539903e-07,
+      "loss": 0.4719,
+      "step": 8830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6841033911163456e-07,
+      "loss": 0.8041,
+      "step": 8831
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6740606763015286e-07,
+      "loss": 0.656,
+      "step": 8832
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6640284812037085e-07,
+      "loss": 0.7268,
+      "step": 8833
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6540068069300006e-07,
+      "loss": 0.7096,
+      "step": 8834
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6439956545863527e-07,
+      "loss": 0.6058,
+      "step": 8835
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.633995025277527e-07,
+      "loss": 0.4996,
+      "step": 8836
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6240049201071835e-07,
+      "loss": 0.455,
+      "step": 8837
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.614025340177741e-07,
+      "loss": 0.4771,
+      "step": 8838
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.604056286590497e-07,
+      "loss": 0.5684,
+      "step": 8839
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.594097760445626e-07,
+      "loss": 0.7366,
+      "step": 8840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.584149762842083e-07,
+      "loss": 0.6487,
+      "step": 8841
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.574212294877678e-07,
+      "loss": 0.4729,
+      "step": 8842
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5642853576490786e-07,
+      "loss": 0.7243,
+      "step": 8843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.554368952251753e-07,
+      "loss": 0.632,
+      "step": 8844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5444630797800483e-07,
+      "loss": 0.612,
+      "step": 8845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5345677413271115e-07,
+      "loss": 0.7414,
+      "step": 8846
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5246829379849587e-07,
+      "loss": 0.6047,
+      "step": 8847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.514808670844417e-07,
+      "loss": 0.644,
+      "step": 8848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5049449409951817e-07,
+      "loss": 0.6411,
+      "step": 8849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4950917495257487e-07,
+      "loss": 0.76,
+      "step": 8850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4852490975234807e-07,
+      "loss": 0.2916,
+      "step": 8851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4754169860745546e-07,
+      "loss": 0.4187,
+      "step": 8852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4655954162640127e-07,
+      "loss": 0.3691,
+      "step": 8853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4557843891756993e-07,
+      "loss": 0.5516,
+      "step": 8854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4459839058923146e-07,
+      "loss": 0.675,
+      "step": 8855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.436193967495417e-07,
+      "loss": 0.6639,
+      "step": 8856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4264145750653633e-07,
+      "loss": 0.5616,
+      "step": 8857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.416645729681346e-07,
+      "loss": 0.596,
+      "step": 8858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.406887432421425e-07,
+      "loss": 0.5785,
+      "step": 8859
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3971396843624617e-07,
+      "loss": 0.6029,
+      "step": 8860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3874024865801836e-07,
+      "loss": 0.6507,
+      "step": 8861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3776758401491536e-07,
+      "loss": 0.7222,
+      "step": 8862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.367959746142736e-07,
+      "loss": 0.6437,
+      "step": 8863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.358254205633183e-07,
+      "loss": 0.6635,
+      "step": 8864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3485592196915174e-07,
+      "loss": 0.532,
+      "step": 8865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.338874789387637e-07,
+      "loss": 0.512,
+      "step": 8866
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.329200915790288e-07,
+      "loss": 0.7391,
+      "step": 8867
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3195375999670163e-07,
+      "loss": 0.6643,
+      "step": 8868
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3098848429842355e-07,
+      "loss": 0.2265,
+      "step": 8869
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.30024264590716e-07,
+      "loss": 0.757,
+      "step": 8870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2906110097998723e-07,
+      "loss": 0.4011,
+      "step": 8871
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.280989935725277e-07,
+      "loss": 0.7694,
+      "step": 8872
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.271379424745081e-07,
+      "loss": 0.6029,
+      "step": 8873
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.261779477919892e-07,
+      "loss": 0.5453,
+      "step": 8874
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2521900963090836e-07,
+      "loss": 0.5244,
+      "step": 8875
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2426112809709097e-07,
+      "loss": 0.8365,
+      "step": 8876
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.233043032962447e-07,
+      "loss": 0.5813,
+      "step": 8877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2234853533395846e-07,
+      "loss": 0.7489,
+      "step": 8878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2139382431570786e-07,
+      "loss": 0.4794,
+      "step": 8879
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.204401703468497e-07,
+      "loss": 0.5936,
+      "step": 8880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1948757353262537e-07,
+      "loss": 0.77,
+      "step": 8881
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1853603397815746e-07,
+      "loss": 0.6984,
+      "step": 8882
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.175855517884531e-07,
+      "loss": 0.8358,
+      "step": 8883
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1663612706840605e-07,
+      "loss": 0.5147,
+      "step": 8884
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.156877599227871e-07,
+      "loss": 0.5784,
+      "step": 8885
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.147404504562547e-07,
+      "loss": 0.5858,
+      "step": 8886
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.137941987733496e-07,
+      "loss": 0.4259,
+      "step": 8887
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.12849004978495e-07,
+      "loss": 0.3786,
+      "step": 8888
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.119048691759986e-07,
+      "loss": 0.7336,
+      "step": 8889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1096179147005146e-07,
+      "loss": 0.5864,
+      "step": 8890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1001977196472477e-07,
+      "loss": 0.6186,
+      "step": 8891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.090788107639798e-07,
+      "loss": 0.6964,
+      "step": 8892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0813890797165135e-07,
+      "loss": 0.4981,
+      "step": 8893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.07200063691463e-07,
+      "loss": 0.4631,
+      "step": 8894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.062622780270253e-07,
+      "loss": 0.6517,
+      "step": 8895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.053255510818255e-07,
+      "loss": 0.3944,
+      "step": 8896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.043898829592363e-07,
+      "loss": 0.535,
+      "step": 8897
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0345527376251303e-07,
+      "loss": 0.7272,
+      "step": 8898
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0252172359479647e-07,
+      "loss": 0.5347,
+      "step": 8899
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0158923255910533e-07,
+      "loss": 0.631,
+      "step": 8900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.006578007583484e-07,
+      "loss": 0.6506,
+      "step": 8901
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.997274282953123e-07,
+      "loss": 0.5597,
+      "step": 8902
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.987981152726694e-07,
+      "loss": 0.6324,
+      "step": 8903
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.978698617929744e-07,
+      "loss": 0.8509,
+      "step": 8904
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.969426679586641e-07,
+      "loss": 0.3964,
+      "step": 8905
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9601653387205895e-07,
+      "loss": 0.4692,
+      "step": 8906
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.950914596353639e-07,
+      "loss": 0.6327,
+      "step": 8907
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9416744535066384e-07,
+      "loss": 0.6593,
+      "step": 8908
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.932444911199318e-07,
+      "loss": 0.3848,
+      "step": 8909
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.923225970450173e-07,
+      "loss": 0.8558,
+      "step": 8910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.914017632276579e-07,
+      "loss": 0.4658,
+      "step": 8911
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9048198976947115e-07,
+      "loss": 0.5586,
+      "step": 8912
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.895632767719615e-07,
+      "loss": 0.5552,
+      "step": 8913
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.886456243365111e-07,
+      "loss": 0.6776,
+      "step": 8914
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.87729032564389e-07,
+      "loss": 0.5349,
+      "step": 8915
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.868135015567431e-07,
+      "loss": 0.6403,
+      "step": 8916
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.858990314146116e-07,
+      "loss": 0.5509,
+      "step": 8917
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8498562223890925e-07,
+      "loss": 0.6533,
+      "step": 8918
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.840732741304365e-07,
+      "loss": 0.4571,
+      "step": 8919
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8316198718987287e-07,
+      "loss": 0.5333,
+      "step": 8920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8225176151778564e-07,
+      "loss": 0.5772,
+      "step": 8921
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.813425972146212e-07,
+      "loss": 0.5722,
+      "step": 8922
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.804344943807126e-07,
+      "loss": 0.5742,
+      "step": 8923
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7952745311627405e-07,
+      "loss": 0.519,
+      "step": 8924
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.786214735214011e-07,
+      "loss": 0.6259,
+      "step": 8925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.777165556960749e-07,
+      "loss": 0.5229,
+      "step": 8926
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7681269974015443e-07,
+      "loss": 0.5194,
+      "step": 8927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.759099057533877e-07,
+      "loss": 0.5311,
+      "step": 8928
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7500817383540277e-07,
+      "loss": 0.6935,
+      "step": 8929
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7410750408571005e-07,
+      "loss": 0.8149,
+      "step": 8930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.732078966037034e-07,
+      "loss": 0.5813,
+      "step": 8931
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7230935148865776e-07,
+      "loss": 0.7339,
+      "step": 8932
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.714118688397339e-07,
+      "loss": 0.7475,
+      "step": 8933
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.705154487559737e-07,
+      "loss": 0.4991,
+      "step": 8934
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6962009133630015e-07,
+      "loss": 0.6955,
+      "step": 8935
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.687257966795232e-07,
+      "loss": 0.7076,
+      "step": 8936
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6783256488433063e-07,
+      "loss": 0.5546,
+      "step": 8937
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6694039604929455e-07,
+      "loss": 0.457,
+      "step": 8938
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6604929027287515e-07,
+      "loss": 0.6569,
+      "step": 8939
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6515924765340714e-07,
+      "loss": 0.4627,
+      "step": 8940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.642702682891108e-07,
+      "loss": 0.5516,
+      "step": 8941
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6338235227809105e-07,
+      "loss": 0.5435,
+      "step": 8942
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6249549971833386e-07,
+      "loss": 0.5548,
+      "step": 8943
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6160971070770654e-07,
+      "loss": 0.4845,
+      "step": 8944
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.607249853439632e-07,
+      "loss": 0.7815,
+      "step": 8945
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.598413237247378e-07,
+      "loss": 0.5434,
+      "step": 8946
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5895872594754464e-07,
+      "loss": 0.604,
+      "step": 8947
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5807719210978364e-07,
+      "loss": 0.4773,
+      "step": 8948
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.571967223087369e-07,
+      "loss": 0.6589,
+      "step": 8949
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.563173166415701e-07,
+      "loss": 0.6869,
+      "step": 8950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5543897520533e-07,
+      "loss": 0.5384,
+      "step": 8951
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5456169809694465e-07,
+      "loss": 0.4504,
+      "step": 8952
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.536854854132288e-07,
+      "loss": 0.7425,
+      "step": 8953
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.52810337250874e-07,
+      "loss": 1.0461,
+      "step": 8954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5193625370645854e-07,
+      "loss": 0.3512,
+      "step": 8955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.510632348764431e-07,
+      "loss": 0.7949,
+      "step": 8956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5019128085716947e-07,
+      "loss": 0.6818,
+      "step": 8957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4932039174486176e-07,
+      "loss": 0.4868,
+      "step": 8958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4845056763562757e-07,
+      "loss": 0.7968,
+      "step": 8959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4758180862545674e-07,
+      "loss": 0.5495,
+      "step": 8960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.467141148102204e-07,
+      "loss": 0.5833,
+      "step": 8961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.458474862856742e-07,
+      "loss": 0.5035,
+      "step": 8962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4498192314745493e-07,
+      "loss": 0.692,
+      "step": 8963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.441174254910817e-07,
+      "loss": 0.5904,
+      "step": 8964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.432539934119572e-07,
+      "loss": 0.7643,
+      "step": 8965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.42391627005364e-07,
+      "loss": 0.566,
+      "step": 8966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.415303263664704e-07,
+      "loss": 0.7139,
+      "step": 8967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4067009159032495e-07,
+      "loss": 0.6624,
+      "step": 8968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.398109227718582e-07,
+      "loss": 0.6691,
+      "step": 8969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3895282000588557e-07,
+      "loss": 0.7677,
+      "step": 8970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3809578338710125e-07,
+      "loss": 0.9048,
+      "step": 8971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.372398130100851e-07,
+      "loss": 0.4682,
+      "step": 8972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.363849089692983e-07,
+      "loss": 0.6672,
+      "step": 8973
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3553107135908316e-07,
+      "loss": 0.6453,
+      "step": 8974
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.346783002736642e-07,
+      "loss": 0.4104,
+      "step": 8975
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.338265958071507e-07,
+      "loss": 0.5926,
+      "step": 8976
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.329759580535319e-07,
+      "loss": 0.6032,
+      "step": 8977
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3212638710668043e-07,
+      "loss": 0.4108,
+      "step": 8978
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3127788306035134e-07,
+      "loss": 0.5854,
+      "step": 8979
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3043044600818197e-07,
+      "loss": 0.5226,
+      "step": 8980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.295840760436908e-07,
+      "loss": 0.6241,
+      "step": 8981
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2873877326027757e-07,
+      "loss": 0.5136,
+      "step": 8982
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.278945377512266e-07,
+      "loss": 0.6054,
+      "step": 8983
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2705136960970554e-07,
+      "loss": 0.5805,
+      "step": 8984
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2620926892876104e-07,
+      "loss": 0.5345,
+      "step": 8985
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2536823580132436e-07,
+      "loss": 0.8036,
+      "step": 8986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2452827032020575e-07,
+      "loss": 0.5761,
+      "step": 8987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2368937257810206e-07,
+      "loss": 0.5723,
+      "step": 8988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.228515426675893e-07,
+      "loss": 0.8271,
+      "step": 8989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.220147806811258e-07,
+      "loss": 0.5891,
+      "step": 8990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.21179086711052e-07,
+      "loss": 0.432,
+      "step": 8991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2034446084959316e-07,
+      "loss": 0.6108,
+      "step": 8992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1951090318885433e-07,
+      "loss": 0.5853,
+      "step": 8993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.18678413820821e-07,
+      "loss": 0.4123,
+      "step": 8994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1784699283736507e-07,
+      "loss": 0.5883,
+      "step": 8995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.170166403302366e-07,
+      "loss": 0.7462,
+      "step": 8996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.161873563910689e-07,
+      "loss": 0.6465,
+      "step": 8997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.153591411113799e-07,
+      "loss": 0.6232,
+      "step": 8998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.145319945825653e-07,
+      "loss": 0.5189,
+      "step": 8999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.137059168959078e-07,
+      "loss": 0.7508,
+      "step": 9000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1288090814256654e-07,
+      "loss": 0.5435,
+      "step": 9001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.120569684135877e-07,
+      "loss": 0.687,
+      "step": 9002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.112340977998951e-07,
+      "loss": 0.5074,
+      "step": 9003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1041229639229955e-07,
+      "loss": 0.6265,
+      "step": 9004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.095915642814873e-07,
+      "loss": 0.4327,
+      "step": 9005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.087719015580337e-07,
+      "loss": 0.4458,
+      "step": 9006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.079533083123931e-07,
+      "loss": 0.5696,
+      "step": 9007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.07135784634901e-07,
+      "loss": 0.5636,
+      "step": 9008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.063193306157741e-07,
+      "loss": 0.5349,
+      "step": 9009
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.055039463451115e-07,
+      "loss": 0.6005,
+      "step": 9010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.046896319128978e-07,
+      "loss": 0.4958,
+      "step": 9011
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.038763874089967e-07,
+      "loss": 0.3837,
+      "step": 9012
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0306421292315203e-07,
+      "loss": 0.3583,
+      "step": 9013
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.022531085449931e-07,
+      "loss": 0.6085,
+      "step": 9014
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.014430743640295e-07,
+      "loss": 0.3512,
+      "step": 9015
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0063411046965086e-07,
+      "loss": 0.4948,
+      "step": 9016
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9982621695113345e-07,
+      "loss": 0.416,
+      "step": 9017
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9901939389763047e-07,
+      "loss": 0.6269,
+      "step": 9018
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9821364139817954e-07,
+      "loss": 0.4206,
+      "step": 9019
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.974089595416996e-07,
+      "loss": 0.5316,
+      "step": 9020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9660534841699175e-07,
+      "loss": 0.3949,
+      "step": 9021
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.958028081127395e-07,
+      "loss": 0.8068,
+      "step": 9022
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.950013387175055e-07,
+      "loss": 0.6812,
+      "step": 9023
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.942009403197377e-07,
+      "loss": 0.5353,
+      "step": 9024
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.934016130077644e-07,
+      "loss": 0.5769,
+      "step": 9025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.926033568697939e-07,
+      "loss": 0.8122,
+      "step": 9026
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9180617199391804e-07,
+      "loss": 0.7228,
+      "step": 9027
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9101005846811303e-07,
+      "loss": 0.8352,
+      "step": 9028
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9021501638023197e-07,
+      "loss": 0.324,
+      "step": 9029
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8942104581801136e-07,
+      "loss": 0.7522,
+      "step": 9030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8862814686907217e-07,
+      "loss": 0.5708,
+      "step": 9031
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8783631962091327e-07,
+      "loss": 0.6653,
+      "step": 9032
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.87045564160916e-07,
+      "loss": 0.5647,
+      "step": 9033
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.862558805763471e-07,
+      "loss": 0.5382,
+      "step": 9034
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.854672689543514e-07,
+      "loss": 0.4421,
+      "step": 9035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8467972938195585e-07,
+      "loss": 0.6062,
+      "step": 9036
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.838932619460688e-07,
+      "loss": 0.3018,
+      "step": 9037
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.831078667334819e-07,
+      "loss": 0.6678,
+      "step": 9038
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.82323543830868e-07,
+      "loss": 0.544,
+      "step": 9039
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8154029332478014e-07,
+      "loss": 0.5687,
+      "step": 9040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.807581153016559e-07,
+      "loss": 0.5092,
+      "step": 9041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7997700984781275e-07,
+      "loss": 0.7232,
+      "step": 9042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7919697704944735e-07,
+      "loss": 0.7686,
+      "step": 9043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.784180169926398e-07,
+      "loss": 0.7326,
+      "step": 9044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7764012976335687e-07,
+      "loss": 0.6615,
+      "step": 9045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.768633154474387e-07,
+      "loss": 0.4751,
+      "step": 9046
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.760875741306135e-07,
+      "loss": 0.7149,
+      "step": 9047
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.75312905898486e-07,
+      "loss": 0.5957,
+      "step": 9048
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7453931083654573e-07,
+      "loss": 0.5833,
+      "step": 9049
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.737667890301632e-07,
+      "loss": 0.5787,
+      "step": 9050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.729953405645913e-07,
+      "loss": 0.4784,
+      "step": 9051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.722249655249609e-07,
+      "loss": 0.5283,
+      "step": 9052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.714556639962884e-07,
+      "loss": 0.5325,
+      "step": 9053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.706874360634715e-07,
+      "loss": 0.6674,
+      "step": 9054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.699202818112856e-07,
+      "loss": 0.3934,
+      "step": 9055
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6915420132439085e-07,
+      "loss": 0.5906,
+      "step": 9056
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6838919468733073e-07,
+      "loss": 0.7428,
+      "step": 9057
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.676252619845243e-07,
+      "loss": 0.6371,
+      "step": 9058
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6686240330027755e-07,
+      "loss": 0.4804,
+      "step": 9059
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6610061871877426e-07,
+      "loss": 0.7814,
+      "step": 9060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6533990832408376e-07,
+      "loss": 0.57,
+      "step": 9061
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6458027220015445e-07,
+      "loss": 0.6741,
+      "step": 9062
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.638217104308127e-07,
+      "loss": 0.7066,
+      "step": 9063
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6306422309977375e-07,
+      "loss": 0.5693,
+      "step": 9064
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6230781029062735e-07,
+      "loss": 0.5465,
+      "step": 9065
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6155247208684785e-07,
+      "loss": 0.5606,
+      "step": 9066
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6079820857179305e-07,
+      "loss": 0.6696,
+      "step": 9067
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.600450198286986e-07,
+      "loss": 0.6425,
+      "step": 9068
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5929290594068344e-07,
+      "loss": 0.6617,
+      "step": 9069
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.585418669907458e-07,
+      "loss": 0.6417,
+      "step": 9070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.57791903061767e-07,
+      "loss": 0.539,
+      "step": 9071
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.570430142365099e-07,
+      "loss": 0.8092,
+      "step": 9072
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.562952005976194e-07,
+      "loss": 0.6693,
+      "step": 9073
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.555484622276194e-07,
+      "loss": 0.6257,
+      "step": 9074
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5480279920891747e-07,
+      "loss": 0.486,
+      "step": 9075
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5405821162379996e-07,
+      "loss": 0.5384,
+      "step": 9076
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.533146995544378e-07,
+      "loss": 0.484,
+      "step": 9077
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.525722630828786e-07,
+      "loss": 0.5698,
+      "step": 9078
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.51830902291057e-07,
+      "loss": 0.4508,
+      "step": 9079
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.510906172607852e-07,
+      "loss": 0.3404,
+      "step": 9080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.503514080737568e-07,
+      "loss": 0.5521,
+      "step": 9081
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.496132748115476e-07,
+      "loss": 0.6652,
+      "step": 9082
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4887621755561585e-07,
+      "loss": 0.5827,
+      "step": 9083
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4814023638729756e-07,
+      "loss": 0.5196,
+      "step": 9084
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.474053313878133e-07,
+      "loss": 0.7206,
+      "step": 9085
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.466715026382638e-07,
+      "loss": 0.512,
+      "step": 9086
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.459387502196309e-07,
+      "loss": 0.2918,
+      "step": 9087
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4520707421277657e-07,
+      "loss": 0.4846,
+      "step": 9088
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.444764746984474e-07,
+      "loss": 0.6157,
+      "step": 9089
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4374695175726657e-07,
+      "loss": 0.57,
+      "step": 9090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.430185054697409e-07,
+      "loss": 0.8952,
+      "step": 9091
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4229113591625944e-07,
+      "loss": 0.6272,
+      "step": 9092
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.415648431770912e-07,
+      "loss": 0.5141,
+      "step": 9093
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.408396273323854e-07,
+      "loss": 0.847,
+      "step": 9094
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.401154884621737e-07,
+      "loss": 0.8068,
+      "step": 9095
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.393924266463699e-07,
+      "loss": 0.5085,
+      "step": 9096
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3867044196476673e-07,
+      "loss": 0.5596,
+      "step": 9097
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.379495344970384e-07,
+      "loss": 0.5253,
+      "step": 9098
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3722970432274007e-07,
+      "loss": 0.5952,
+      "step": 9099
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3651095152131154e-07,
+      "loss": 0.4507,
+      "step": 9100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3579327617206825e-07,
+      "loss": 0.5943,
+      "step": 9101
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3507667835421134e-07,
+      "loss": 0.624,
+      "step": 9102
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.343611581468208e-07,
+      "loss": 0.4794,
+      "step": 9103
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3364671562885799e-07,
+      "loss": 0.4546,
+      "step": 9104
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3293335087916314e-07,
+      "loss": 0.6518,
+      "step": 9105
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3222106397646328e-07,
+      "loss": 0.6204,
+      "step": 9106
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3150985499936106e-07,
+      "loss": 0.6238,
+      "step": 9107
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3079972402634265e-07,
+      "loss": 0.5326,
+      "step": 9108
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.300906711357742e-07,
+      "loss": 0.6818,
+      "step": 9109
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2938269640590427e-07,
+      "loss": 0.5392,
+      "step": 9110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2867579991486034e-07,
+      "loss": 0.4581,
+      "step": 9111
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2796998174065332e-07,
+      "loss": 0.6182,
+      "step": 9112
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2726524196117317e-07,
+      "loss": 0.5592,
+      "step": 9113
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.265615806541921e-07,
+      "loss": 0.5876,
+      "step": 9114
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2585899789736243e-07,
+      "loss": 0.7138,
+      "step": 9115
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2515749376821772e-07,
+      "loss": 0.5223,
+      "step": 9116
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.244570683441738e-07,
+      "loss": 0.7557,
+      "step": 9117
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2375772170252553e-07,
+      "loss": 0.7712,
+      "step": 9118
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2305945392044893e-07,
+      "loss": 0.5023,
+      "step": 9119
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2236226507500236e-07,
+      "loss": 0.422,
+      "step": 9120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2166615524312318e-07,
+      "loss": 0.7435,
+      "step": 9121
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2097112450163216e-07,
+      "loss": 0.5883,
+      "step": 9122
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2027717292722904e-07,
+      "loss": 0.7379,
+      "step": 9123
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1958430059649595e-07,
+      "loss": 0.3041,
+      "step": 9124
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1889250758589164e-07,
+      "loss": 0.6714,
+      "step": 9125
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.182017939717629e-07,
+      "loss": 0.656,
+      "step": 9126
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.175121598303309e-07,
+      "loss": 0.6065,
+      "step": 9127
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.168236052377015e-07,
+      "loss": 0.4253,
+      "step": 9128
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.161361302698617e-07,
+      "loss": 0.5188,
+      "step": 9129
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1544973500267629e-07,
+      "loss": 0.5612,
+      "step": 9130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1476441951189252e-07,
+      "loss": 0.5524,
+      "step": 9131
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.140801838731399e-07,
+      "loss": 0.5728,
+      "step": 9132
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1339702816192465e-07,
+      "loss": 0.4503,
+      "step": 9133
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127149524536387e-07,
+      "loss": 0.7844,
+      "step": 9134
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1203395682355299e-07,
+      "loss": 0.5075,
+      "step": 9135
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1135404134681848e-07,
+      "loss": 0.5947,
+      "step": 9136
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1067520609846737e-07,
+      "loss": 0.495,
+      "step": 9137
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0999745115341198e-07,
+      "loss": 0.5255,
+      "step": 9138
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.09320776586448e-07,
+      "loss": 0.6011,
+      "step": 9139
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0864518247224797e-07,
+      "loss": 0.7518,
+      "step": 9140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0797066888536777e-07,
+      "loss": 0.6344,
+      "step": 9141
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.072972359002434e-07,
+      "loss": 0.3244,
+      "step": 9142
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0662488359119214e-07,
+      "loss": 0.5554,
+      "step": 9143
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0595361203241237e-07,
+      "loss": 0.6178,
+      "step": 9144
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0528342129798152e-07,
+      "loss": 0.4646,
+      "step": 9145
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0461431146185819e-07,
+      "loss": 0.5059,
+      "step": 9146
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0394628259788219e-07,
+      "loss": 0.4713,
+      "step": 9147
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.032793347797757e-07,
+      "loss": 0.6202,
+      "step": 9148
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0261346808113757e-07,
+      "loss": 0.5322,
+      "step": 9149
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.019486825754513e-07,
+      "loss": 0.4714,
+      "step": 9150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.012849783360793e-07,
+      "loss": 0.4529,
+      "step": 9151
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0062235543626408e-07,
+      "loss": 0.8662,
+      "step": 9152
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9996081394913047e-07,
+      "loss": 0.6587,
+      "step": 9153
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9930035394768233e-07,
+      "loss": 0.6751,
+      "step": 9154
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9864097550480355e-07,
+      "loss": 0.5997,
+      "step": 9155
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.979826786932626e-07,
+      "loss": 0.6481,
+      "step": 9156
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9732546358570471e-07,
+      "loss": 0.7061,
+      "step": 9157
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.966693302546585e-07,
+      "loss": 0.6267,
+      "step": 9158
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9601427877252944e-07,
+      "loss": 0.638,
+      "step": 9159
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.953603092116063e-07,
+      "loss": 0.3876,
+      "step": 9160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.947074216440592e-07,
+      "loss": 0.6597,
+      "step": 9161
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9405561614193714e-07,
+      "loss": 0.4462,
+      "step": 9162
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9340489277717034e-07,
+      "loss": 0.5789,
+      "step": 9163
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9275525162156915e-07,
+      "loss": 0.5732,
+      "step": 9164
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9210669274682626e-07,
+      "loss": 0.5212,
+      "step": 9165
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9145921622451102e-07,
+      "loss": 0.4053,
+      "step": 9166
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9081282212607854e-07,
+      "loss": 0.6054,
+      "step": 9167
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9016751052285952e-07,
+      "loss": 0.7541,
+      "step": 9168
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8952328148606923e-07,
+      "loss": 0.3953,
+      "step": 9169
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.888801350868008e-07,
+      "loss": 0.4713,
+      "step": 9170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8823807139602857e-07,
+      "loss": 0.5429,
+      "step": 9171
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8759709048460805e-07,
+      "loss": 0.636,
+      "step": 9172
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8695719242327603e-07,
+      "loss": 0.6182,
+      "step": 9173
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.86318377282646e-07,
+      "loss": 0.5274,
+      "step": 9174
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8568064513321715e-07,
+      "loss": 0.4132,
+      "step": 9175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8504399604536426e-07,
+      "loss": 0.3698,
+      "step": 9176
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.844084300893456e-07,
+      "loss": 0.7427,
+      "step": 9177
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8377394733530173e-07,
+      "loss": 0.7812,
+      "step": 9178
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8314054785324886e-07,
+      "loss": 0.5137,
+      "step": 9179
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.825082317130855e-07,
+      "loss": 0.6392,
+      "step": 9180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8187699898459144e-07,
+      "loss": 0.6225,
+      "step": 9181
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8124684973742534e-07,
+      "loss": 0.6549,
+      "step": 9182
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8061778404113162e-07,
+      "loss": 0.6633,
+      "step": 9183
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7998980196512695e-07,
+      "loss": 0.5027,
+      "step": 9184
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7936290357871367e-07,
+      "loss": 0.5517,
+      "step": 9185
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.787370889510742e-07,
+      "loss": 0.6827,
+      "step": 9186
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7811235815127003e-07,
+      "loss": 0.4319,
+      "step": 9187
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7748871124824153e-07,
+      "loss": 0.484,
+      "step": 9188
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.768661483108136e-07,
+      "loss": 0.6165,
+      "step": 9189
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7624466940768915e-07,
+      "loss": 0.5566,
+      "step": 9190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7562427460745103e-07,
+      "loss": 0.4333,
+      "step": 9191
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7500496397856226e-07,
+      "loss": 0.6689,
+      "step": 9192
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.743867375893704e-07,
+      "loss": 0.8095,
+      "step": 9193
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7376959550809535e-07,
+      "loss": 0.6023,
+      "step": 9194
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7315353780284483e-07,
+      "loss": 0.5289,
+      "step": 9195
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7253856454160334e-07,
+      "loss": 0.6361,
+      "step": 9196
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.719246757922366e-07,
+      "loss": 0.6209,
+      "step": 9197
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.713118716224893e-07,
+      "loss": 0.5915,
+      "step": 9198
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7070015209998957e-07,
+      "loss": 0.5221,
+      "step": 9199
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7008951729224344e-07,
+      "loss": 0.4713,
+      "step": 9200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6947996726663585e-07,
+      "loss": 0.6147,
+      "step": 9201
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6887150209043636e-07,
+      "loss": 0.5554,
+      "step": 9202
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.682641218307901e-07,
+      "loss": 0.5185,
+      "step": 9203
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6765782655472572e-07,
+      "loss": 0.6908,
+      "step": 9204
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6705261632914971e-07,
+      "loss": 0.4439,
+      "step": 9205
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6644849122085417e-07,
+      "loss": 0.5389,
+      "step": 9206
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6584545129650355e-07,
+      "loss": 0.7366,
+      "step": 9207
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6524349662264682e-07,
+      "loss": 0.4428,
+      "step": 9208
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6464262726571423e-07,
+      "loss": 0.5217,
+      "step": 9209
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.640428432920138e-07,
+      "loss": 0.5784,
+      "step": 9210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6344414476773596e-07,
+      "loss": 0.6912,
+      "step": 9211
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6284653175895004e-07,
+      "loss": 0.6883,
+      "step": 9212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6225000433160554e-07,
+      "loss": 0.5273,
+      "step": 9213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.616545625515309e-07,
+      "loss": 0.4449,
+      "step": 9214
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.61060206484438e-07,
+      "loss": 0.5254,
+      "step": 9215
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6046693619591659e-07,
+      "loss": 0.5191,
+      "step": 9216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.598747517514365e-07,
+      "loss": 0.5577,
+      "step": 9217
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5928365321634997e-07,
+      "loss": 0.6193,
+      "step": 9218
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.58693640655887e-07,
+      "loss": 0.5183,
+      "step": 9219
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5810471413515993e-07,
+      "loss": 0.497,
+      "step": 9220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.575168737191557e-07,
+      "loss": 0.5771,
+      "step": 9221
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5693011947275016e-07,
+      "loss": 0.6545,
+      "step": 9222
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5634445146069265e-07,
+      "loss": 0.8646,
+      "step": 9223
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.557598697476148e-07,
+      "loss": 0.6482,
+      "step": 9224
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5517637439802835e-07,
+      "loss": 0.4955,
+      "step": 9225
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.545939654763251e-07,
+      "loss": 0.6433,
+      "step": 9226
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.54012643046777e-07,
+      "loss": 0.4056,
+      "step": 9227
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5343240717353713e-07,
+      "loss": 0.6987,
+      "step": 9228
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5285325792063543e-07,
+      "loss": 0.415,
+      "step": 9229
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.522751953519852e-07,
+      "loss": 0.8318,
+      "step": 9230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5169821953137875e-07,
+      "loss": 0.4323,
+      "step": 9231
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5112233052248627e-07,
+      "loss": 0.4832,
+      "step": 9232
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5054752838886466e-07,
+      "loss": 0.6105,
+      "step": 9233
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4997381319394323e-07,
+      "loss": 0.4133,
+      "step": 9234
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.494011850010346e-07,
+      "loss": 0.3421,
+      "step": 9235
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4882964387333164e-07,
+      "loss": 0.6852,
+      "step": 9236
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4825918987390719e-07,
+      "loss": 0.6325,
+      "step": 9237
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.47689823065712e-07,
+      "loss": 0.4712,
+      "step": 9238
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4712154351158137e-07,
+      "loss": 0.8074,
+      "step": 9239
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4655435127422846e-07,
+      "loss": 0.5755,
+      "step": 9240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.459882464162432e-07,
+      "loss": 0.5326,
+      "step": 9241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4542322900009786e-07,
+      "loss": 0.5638,
+      "step": 9242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.44859299088147e-07,
+      "loss": 0.6,
+      "step": 9243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4429645674262306e-07,
+      "loss": 0.7851,
+      "step": 9244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4373470202563856e-07,
+      "loss": 0.4955,
+      "step": 9245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4317403499918614e-07,
+      "loss": 0.6338,
+      "step": 9246
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4261445572513855e-07,
+      "loss": 0.7255,
+      "step": 9247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4205596426524748e-07,
+      "loss": 0.5007,
+      "step": 9248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4149856068114364e-07,
+      "loss": 0.4969,
+      "step": 9249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4094224503434338e-07,
+      "loss": 0.5425,
+      "step": 9250
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.403870173862365e-07,
+      "loss": 0.6719,
+      "step": 9251
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.398328777980973e-07,
+      "loss": 0.3709,
+      "step": 9252
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3927982633107462e-07,
+      "loss": 0.536,
+      "step": 9253
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3872786304620411e-07,
+      "loss": 0.3738,
+      "step": 9254
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3817698800439593e-07,
+      "loss": 0.7618,
+      "step": 9255
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3762720126644258e-07,
+      "loss": 0.5672,
+      "step": 9256
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.370785028930155e-07,
+      "loss": 0.6511,
+      "step": 9257
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3653089294466625e-07,
+      "loss": 0.6086,
+      "step": 9258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3598437148182652e-07,
+      "loss": 0.4514,
+      "step": 9259
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3543893856480915e-07,
+      "loss": 0.6033,
+      "step": 9260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.348945942538049e-07,
+      "loss": 0.6837,
+      "step": 9261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3435133860888348e-07,
+      "loss": 0.6991,
+      "step": 9262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3380917168999808e-07,
+      "loss": 0.5592,
+      "step": 9263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3326809355697744e-07,
+      "loss": 0.6277,
+      "step": 9264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3272810426953497e-07,
+      "loss": 0.5306,
+      "step": 9265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3218920388725853e-07,
+      "loss": 0.8078,
+      "step": 9266
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3165139246962276e-07,
+      "loss": 0.5078,
+      "step": 9267
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3111467007597355e-07,
+      "loss": 0.5596,
+      "step": 9268
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.305790367655435e-07,
+      "loss": 0.6844,
+      "step": 9269
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.300444925974409e-07,
+      "loss": 0.5476,
+      "step": 9270
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2951103763065632e-07,
+      "loss": 0.7561,
+      "step": 9271
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2897867192405932e-07,
+      "loss": 0.5139,
+      "step": 9272
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2844739553640073e-07,
+      "loss": 0.4712,
+      "step": 9273
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2791720852630695e-07,
+      "loss": 0.4885,
+      "step": 9274
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2738811095229008e-07,
+      "loss": 0.5976,
+      "step": 9275
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2686010287273564e-07,
+      "loss": 0.715,
+      "step": 9276
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2633318434591367e-07,
+      "loss": 0.4063,
+      "step": 9277
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2580735542997214e-07,
+      "loss": 0.7522,
+      "step": 9278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2528261618293903e-07,
+      "loss": 0.4573,
+      "step": 9279
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2475896666272135e-07,
+      "loss": 0.4328,
+      "step": 9280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2423640692710847e-07,
+      "loss": 0.5648,
+      "step": 9281
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2371493703376536e-07,
+      "loss": 0.527,
+      "step": 9282
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2319455704024152e-07,
+      "loss": 0.4405,
+      "step": 9283
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2267526700396104e-07,
+      "loss": 0.5319,
+      "step": 9284
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.221570669822314e-07,
+      "loss": 0.5127,
+      "step": 9285
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2163995703223908e-07,
+      "loss": 0.3466,
+      "step": 9286
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2112393721104843e-07,
+      "loss": 0.6555,
+      "step": 9287
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2060900757560724e-07,
+      "loss": 0.6593,
+      "step": 9288
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.200951681827389e-07,
+      "loss": 0.602,
+      "step": 9289
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1958241908914813e-07,
+      "loss": 0.7111,
+      "step": 9290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1907076035142073e-07,
+      "loss": 0.5136,
+      "step": 9291
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1856019202602042e-07,
+      "loss": 0.5148,
+      "step": 9292
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1805071416929103e-07,
+      "loss": 0.6796,
+      "step": 9293
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1754232683745537e-07,
+      "loss": 0.8668,
+      "step": 9294
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1703503008661965e-07,
+      "loss": 0.4926,
+      "step": 9295
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1652882397276245e-07,
+      "loss": 0.4623,
+      "step": 9296
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1602370855174905e-07,
+      "loss": 0.5836,
+      "step": 9297
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1551968387932045e-07,
+      "loss": 0.57,
+      "step": 9298
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1501675001109769e-07,
+      "loss": 0.6259,
+      "step": 9299
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1451490700258416e-07,
+      "loss": 0.4971,
+      "step": 9300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1401415490916002e-07,
+      "loss": 0.4855,
+      "step": 9301
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1351449378608659e-07,
+      "loss": 0.4247,
+      "step": 9302
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.13015923688502e-07,
+      "loss": 0.6117,
+      "step": 9303
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1251844467142781e-07,
+      "loss": 0.4905,
+      "step": 9304
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.120220567897623e-07,
+      "loss": 0.5483,
+      "step": 9305
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1152676009828611e-07,
+      "loss": 0.4502,
+      "step": 9306
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1103255465165663e-07,
+      "loss": 0.5586,
+      "step": 9307
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1053944050441245e-07,
+      "loss": 0.509,
+      "step": 9308
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1004741771097227e-07,
+      "loss": 0.4158,
+      "step": 9309
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0955648632563043e-07,
+      "loss": 0.6715,
+      "step": 9310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0906664640256581e-07,
+      "loss": 0.5539,
+      "step": 9311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0857789799583517e-07,
+      "loss": 0.6356,
+      "step": 9312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0809024115937428e-07,
+      "loss": 0.794,
+      "step": 9313
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0760367594699672e-07,
+      "loss": 0.522,
+      "step": 9314
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0711820241240067e-07,
+      "loss": 0.4854,
+      "step": 9315
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0663382060915773e-07,
+      "loss": 0.7021,
+      "step": 9316
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.06150530590724e-07,
+      "loss": 0.3602,
+      "step": 9317
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.056683324104324e-07,
+      "loss": 0.6862,
+      "step": 9318
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.051872261214959e-07,
+      "loss": 0.6613,
+      "step": 9319
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047072117770076e-07,
+      "loss": 0.723,
+      "step": 9320
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0422828942993734e-07,
+      "loss": 0.4593,
+      "step": 9321
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0375045913314063e-07,
+      "loss": 0.4162,
+      "step": 9322
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.032737209393464e-07,
+      "loss": 0.5986,
+      "step": 9323
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0279807490116367e-07,
+      "loss": 0.9158,
+      "step": 9324
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.02323521071086e-07,
+      "loss": 0.6366,
+      "step": 9325
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0185005950147931e-07,
+      "loss": 0.7326,
+      "step": 9326
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0137769024459398e-07,
+      "loss": 0.6234,
+      "step": 9327
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0090641335255946e-07,
+      "loss": 0.6172,
+      "step": 9328
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0043622887738413e-07,
+      "loss": 0.7559,
+      "step": 9329
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.996713687095316e-08,
+      "loss": 0.7027,
+      "step": 9330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.949913738503291e-08,
+      "loss": 0.6618,
+      "step": 9331
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.903223047127208e-08,
+      "loss": 0.7648,
+      "step": 9332
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.85664161811939e-08,
+      "loss": 0.7445,
+      "step": 9333
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.810169456620611e-08,
+      "loss": 0.4505,
+      "step": 9334
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.763806567759105e-08,
+      "loss": 0.5434,
+      "step": 9335
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.717552956651333e-08,
+      "loss": 0.8379,
+      "step": 9336
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.671408628401546e-08,
+      "loss": 0.9121,
+      "step": 9337
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.625373588102116e-08,
+      "loss": 0.9767,
+      "step": 9338
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.579447840833201e-08,
+      "loss": 0.5564,
+      "step": 9339
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.533631391662967e-08,
+      "loss": 0.585,
+      "step": 9340
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.487924245647373e-08,
+      "loss": 0.5014,
+      "step": 9341
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.442326407830715e-08,
+      "loss": 0.6421,
+      "step": 9342
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.396837883244747e-08,
+      "loss": 0.6276,
+      "step": 9343
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.351458676909453e-08,
+      "loss": 0.6587,
+      "step": 9344
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.306188793832604e-08,
+      "loss": 0.4726,
+      "step": 9345
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.261028239009984e-08,
+      "loss": 0.4144,
+      "step": 9346
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.21597701742527e-08,
+      "loss": 0.6674,
+      "step": 9347
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.171035134050155e-08,
+      "loss": 0.4036,
+      "step": 9348
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.126202593844224e-08,
+      "loss": 0.5894,
+      "step": 9349
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.081479401754967e-08,
+      "loss": 0.5797,
+      "step": 9350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.036865562717767e-08,
+      "loss": 0.6248,
+      "step": 9351
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.99236108165602e-08,
+      "loss": 0.8801,
+      "step": 9352
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.947965963481132e-08,
+      "loss": 0.4275,
+      "step": 9353
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.90368021309207e-08,
+      "loss": 0.7786,
+      "step": 9354
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.85950383537637e-08,
+      "loss": 0.5169,
+      "step": 9355
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.815436835208913e-08,
+      "loss": 0.7098,
+      "step": 9356
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.771479217452694e-08,
+      "loss": 0.7083,
+      "step": 9357
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.727630986958724e-08,
+      "loss": 0.7783,
+      "step": 9358
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.683892148565909e-08,
+      "loss": 0.3518,
+      "step": 9359
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.640262707100944e-08,
+      "loss": 0.4342,
+      "step": 9360
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.596742667378644e-08,
+      "loss": 0.5761,
+      "step": 9361
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.553332034201723e-08,
+      "loss": 0.5508,
+      "step": 9362
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.510030812360792e-08,
+      "loss": 0.5527,
+      "step": 9363
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466839006634364e-08,
+      "loss": 0.5758,
+      "step": 9364
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.423756621788626e-08,
+      "loss": 0.7381,
+      "step": 9365
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.380783662578218e-08,
+      "loss": 0.5844,
+      "step": 9366
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.337920133745348e-08,
+      "loss": 0.4023,
+      "step": 9367
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.295166040020341e-08,
+      "loss": 0.4732,
+      "step": 9368
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.252521386121093e-08,
+      "loss": 0.6757,
+      "step": 9369
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.209986176753947e-08,
+      "loss": 0.7414,
+      "step": 9370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.167560416612597e-08,
+      "loss": 0.9592,
+      "step": 9371
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.125244110379182e-08,
+      "loss": 0.5861,
+      "step": 9372
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.083037262723415e-08,
+      "loss": 0.424,
+      "step": 9373
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040939878303122e-08,
+      "loss": 0.6541,
+      "step": 9374
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.998951961763813e-08,
+      "loss": 0.4624,
+      "step": 9375
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.957073517739223e-08,
+      "loss": 0.5564,
+      "step": 9376
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.915304550850878e-08,
+      "loss": 0.5299,
+      "step": 9377
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.873645065708091e-08,
+      "loss": 0.6715,
+      "step": 9378
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.832095066908185e-08,
+      "loss": 0.61,
+      "step": 9379
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.790654559036603e-08,
+      "loss": 0.5025,
+      "step": 9380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.749323546666355e-08,
+      "loss": 0.5213,
+      "step": 9381
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.708102034358566e-08,
+      "loss": 0.7413,
+      "step": 9382
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.666990026662268e-08,
+      "loss": 0.5796,
+      "step": 9383
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.625987528114387e-08,
+      "loss": 0.5649,
+      "step": 9384
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.585094543239857e-08,
+      "loss": 0.6607,
+      "step": 9385
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.544311076551181e-08,
+      "loss": 0.893,
+      "step": 9386
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.503637132549313e-08,
+      "loss": 0.653,
+      "step": 9387
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.463072715722552e-08,
+      "loss": 0.5005,
+      "step": 9388
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.422617830547652e-08,
+      "loss": 0.8823,
+      "step": 9389
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.382272481488928e-08,
+      "loss": 0.6324,
+      "step": 9390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.342036672998708e-08,
+      "loss": 0.4698,
+      "step": 9391
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.30191040951722e-08,
+      "loss": 0.7455,
+      "step": 9392
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.261893695472477e-08,
+      "loss": 0.6375,
+      "step": 9393
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.221986535280723e-08,
+      "loss": 0.5301,
+      "step": 9394
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.182188933345768e-08,
+      "loss": 0.6935,
+      "step": 9395
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.142500894059656e-08,
+      "loss": 0.688,
+      "step": 9396
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.102922421802105e-08,
+      "loss": 0.3966,
+      "step": 9397
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.063453520940844e-08,
+      "loss": 0.5081,
+      "step": 9398
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.024094195831277e-08,
+      "loss": 0.5057,
+      "step": 9399
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.984844450817042e-08,
+      "loss": 0.4559,
+      "step": 9400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.945704290229672e-08,
+      "loss": 0.4709,
+      "step": 9401
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.906673718388268e-08,
+      "loss": 0.5959,
+      "step": 9402
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.867752739600276e-08,
+      "loss": 0.7071,
+      "step": 9403
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.82894135816059e-08,
+      "loss": 0.7592,
+      "step": 9404
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.790239578352453e-08,
+      "loss": 0.7044,
+      "step": 9405
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.751647404446782e-08,
+      "loss": 0.7684,
+      "step": 9406
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.71316484070228e-08,
+      "loss": 0.5405,
+      "step": 9407
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.674791891365884e-08,
+      "loss": 0.5924,
+      "step": 9408
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.636528560672206e-08,
+      "loss": 0.7586,
+      "step": 9409
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.598374852843648e-08,
+      "loss": 0.5735,
+      "step": 9410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.560330772090839e-08,
+      "loss": 0.4629,
+      "step": 9411
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.52239632261209e-08,
+      "loss": 0.7709,
+      "step": 9412
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.484571508593718e-08,
+      "loss": 0.7196,
+      "step": 9413
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.446856334209716e-08,
+      "loss": 0.5415,
+      "step": 9414
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.409250803622313e-08,
+      "loss": 0.8189,
+      "step": 9415
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.371754920981522e-08,
+      "loss": 0.6083,
+      "step": 9416
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.33436869042503e-08,
+      "loss": 0.79,
+      "step": 9417
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.297092116078874e-08,
+      "loss": 0.5488,
+      "step": 9418
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.259925202056427e-08,
+      "loss": 0.4546,
+      "step": 9419
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.222867952459299e-08,
+      "loss": 0.5257,
+      "step": 9420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.185920371377107e-08,
+      "loss": 0.7611,
+      "step": 9421
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.149082462887146e-08,
+      "loss": 0.6185,
+      "step": 9422
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.112354231054607e-08,
+      "loss": 0.6212,
+      "step": 9423
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.075735679932693e-08,
+      "loss": 0.5227,
+      "step": 9424
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.039226813562504e-08,
+      "loss": 0.4521,
+      "step": 9425
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.002827635972819e-08,
+      "loss": 0.7776,
+      "step": 9426
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.966538151180645e-08,
+      "loss": 0.7693,
+      "step": 9427
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.930358363190669e-08,
+      "loss": 0.4494,
+      "step": 9428
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.894288275995585e-08,
+      "loss": 0.5044,
+      "step": 9429
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.858327893575766e-08,
+      "loss": 0.6041,
+      "step": 9430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8224772198997024e-08,
+      "loss": 0.5889,
+      "step": 9431
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.786736258923786e-08,
+      "loss": 0.5962,
+      "step": 9432
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.751105014592084e-08,
+      "loss": 0.5823,
+      "step": 9433
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7155834908366734e-08,
+      "loss": 0.5451,
+      "step": 9434
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6801716915777516e-08,
+      "loss": 0.4407,
+      "step": 9435
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.64486962072297e-08,
+      "loss": 0.4003,
+      "step": 9436
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6096772821682134e-08,
+      "loss": 0.5219,
+      "step": 9437
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.574594679797041e-08,
+      "loss": 0.6169,
+      "step": 9438
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5396218174812445e-08,
+      "loss": 0.7811,
+      "step": 9439
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5047586990799596e-08,
+      "loss": 0.5641,
+      "step": 9440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.470005328440664e-08,
+      "loss": 0.6909,
+      "step": 9441
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.435361709398623e-08,
+      "loss": 0.6655,
+      "step": 9442
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.400827845776779e-08,
+      "loss": 0.7349,
+      "step": 9443
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.3664037413861944e-08,
+      "loss": 0.6158,
+      "step": 9444
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.3320894000259415e-08,
+      "loss": 0.6034,
+      "step": 9445
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.297884825482547e-08,
+      "loss": 1.301,
+      "step": 9446
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.263790021530768e-08,
+      "loss": 0.445,
+      "step": 9447
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.229804991933041e-08,
+      "loss": 0.7566,
+      "step": 9448
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.19592974043992e-08,
+      "loss": 0.7748,
+      "step": 9449
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.162164270789638e-08,
+      "loss": 0.7048,
+      "step": 9450
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1285085867084364e-08,
+      "loss": 0.7932,
+      "step": 9451
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0949626919103437e-08,
+      "loss": 0.5191,
+      "step": 9452
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.06152659009751e-08,
+      "loss": 0.7814,
+      "step": 9453
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.028200284959428e-08,
+      "loss": 0.4547,
+      "step": 9454
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9949837801741564e-08,
+      "loss": 0.474,
+      "step": 9455
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.961877079407096e-08,
+      "loss": 0.5982,
+      "step": 9456
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.928880186311991e-08,
+      "loss": 0.7048,
+      "step": 9457
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.895993104529928e-08,
+      "loss": 0.4468,
+      "step": 9458
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8632158376904494e-08,
+      "loss": 0.7821,
+      "step": 9459
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.830548389410439e-08,
+      "loss": 0.6317,
+      "step": 9460
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7979907632951236e-08,
+      "loss": 0.5714,
+      "step": 9461
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7655429629372975e-08,
+      "loss": 0.36,
+      "step": 9462
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.733204991917761e-08,
+      "loss": 0.8901,
+      "step": 9463
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.700976853805217e-08,
+      "loss": 0.818,
+      "step": 9464
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.668858552156152e-08,
+      "loss": 0.5959,
+      "step": 9465
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.636850090515066e-08,
+      "loss": 0.5544,
+      "step": 9466
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.604951472414243e-08,
+      "loss": 0.6152,
+      "step": 9467
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.573162701373757e-08,
+      "loss": 0.4722,
+      "step": 9468
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.541483780901801e-08,
+      "loss": 0.6873,
+      "step": 9469
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5099147144942454e-08,
+      "loss": 0.6027,
+      "step": 9470
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.47845550563486e-08,
+      "loss": 0.5786,
+      "step": 9471
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.447106157795422e-08,
+      "loss": 0.662,
+      "step": 9472
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.415866674435387e-08,
+      "loss": 0.7064,
+      "step": 9473
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.384737059002442e-08,
+      "loss": 0.573,
+      "step": 9474
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.353717314931616e-08,
+      "loss": 0.5747,
+      "step": 9475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.322807445646171e-08,
+      "loss": 0.5986,
+      "step": 9476
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.292007454557268e-08,
+      "loss": 0.4034,
+      "step": 9477
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.261317345063742e-08,
+      "loss": 0.7286,
+      "step": 9478
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.230737120552442e-08,
+      "loss": 0.5162,
+      "step": 9479
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.200266784398221e-08,
+      "loss": 0.5545,
+      "step": 9480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.169906339963392e-08,
+      "loss": 0.5026,
+      "step": 9481
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1396557905984956e-08,
+      "loss": 0.6569,
+      "step": 9482
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.109515139641862e-08,
+      "loss": 0.6088,
+      "step": 9483
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.079484390419608e-08,
+      "loss": 0.6064,
+      "step": 9484
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.04956354624586e-08,
+      "loss": 0.5671,
+      "step": 9485
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.019752610422423e-08,
+      "loss": 0.4483,
+      "step": 9486
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9900515862392184e-08,
+      "loss": 0.491,
+      "step": 9487
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.960460476973849e-08,
+      "loss": 0.4512,
+      "step": 9488
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9309792858918116e-08,
+      "loss": 0.6378,
+      "step": 9489
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.901608016246616e-08,
+      "loss": 0.667,
+      "step": 9490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8723466712794476e-08,
+      "loss": 0.4649,
+      "step": 9491
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.843195254219501e-08,
+      "loss": 0.6331,
+      "step": 9492
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.814153768283646e-08,
+      "loss": 0.6803,
+      "step": 9493
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.785222216676987e-08,
+      "loss": 0.4425,
+      "step": 9494
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7564006025921915e-08,
+      "loss": 0.7575,
+      "step": 9495
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.727688929209827e-08,
+      "loss": 0.4441,
+      "step": 9496
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.699087199698359e-08,
+      "loss": 0.533,
+      "step": 9497
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.670595417214151e-08,
+      "loss": 0.6718,
+      "step": 9498
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.642213584901466e-08,
+      "loss": 0.5458,
+      "step": 9499
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6139417058923543e-08,
+      "loss": 0.6662,
+      "step": 9500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.585779783306875e-08,
+      "loss": 0.7044,
+      "step": 9501
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557727820252654e-08,
+      "loss": 0.6033,
+      "step": 9502
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5297858198255483e-08,
+      "loss": 0.4427,
+      "step": 9503
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5019537851089805e-08,
+      "loss": 0.4471,
+      "step": 9504
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4742317191743814e-08,
+      "loss": 0.505,
+      "step": 9505
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4466196250810825e-08,
+      "loss": 0.5408,
+      "step": 9506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.4191175058762015e-08,
+      "loss": 0.6616,
+      "step": 9507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.391725364594645e-08,
+      "loss": 0.4195,
+      "step": 9508
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3644432042594374e-08,
+      "loss": 0.6762,
+      "step": 9509
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3372710278811725e-08,
+      "loss": 0.5507,
+      "step": 9510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.310208838458562e-08,
+      "loss": 0.6545,
+      "step": 9511
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.283256638977994e-08,
+      "loss": 0.6895,
+      "step": 9512
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.256414432413868e-08,
+      "loss": 0.5285,
+      "step": 9513
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.229682221728258e-08,
+      "loss": 0.583,
+      "step": 9514
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2030600098711396e-08,
+      "loss": 0.6872,
+      "step": 9515
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.176547799780494e-08,
+      "loss": 0.5955,
+      "step": 9516
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1501455943822034e-08,
+      "loss": 0.5131,
+      "step": 9517
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.123853396589715e-08,
+      "loss": 0.6693,
+      "step": 9518
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.097671209304598e-08,
+      "loss": 0.6185,
+      "step": 9519
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.071599035416206e-08,
+      "loss": 0.7052,
+      "step": 9520
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.045636877801683e-08,
+      "loss": 0.4934,
+      "step": 9521
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0197847393260706e-08,
+      "loss": 0.5114,
+      "step": 9522
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.994042622842419e-08,
+      "loss": 0.5225,
+      "step": 9523
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9684105311913457e-08,
+      "loss": 0.4595,
+      "step": 9524
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9428884672015877e-08,
+      "loss": 0.5577,
+      "step": 9525
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9174764336895588e-08,
+      "loss": 0.7031,
+      "step": 9526
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8921744334597933e-08,
+      "loss": 0.495,
+      "step": 9527
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8669824693042802e-08,
+      "loss": 0.5412,
+      "step": 9528
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.841900544003129e-08,
+      "loss": 0.5869,
+      "step": 9529
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8169286603243472e-08,
+      "loss": 0.7639,
+      "step": 9530
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7920668210237312e-08,
+      "loss": 0.5689,
+      "step": 9531
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7673150288447525e-08,
+      "loss": 0.676,
+      "step": 9532
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7426732865190043e-08,
+      "loss": 0.548,
+      "step": 9533
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7181415967659773e-08,
+      "loss": 0.6053,
+      "step": 9534
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.693719962292618e-08,
+      "loss": 0.5464,
+      "step": 9535
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6694083857941035e-08,
+      "loss": 0.6918,
+      "step": 9536
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6452068699531764e-08,
+      "loss": 0.6558,
+      "step": 9537
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6211154174409227e-08,
+      "loss": 0.3524,
+      "step": 9538
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5971340309157712e-08,
+      "loss": 0.5427,
+      "step": 9539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5732627130241606e-08,
+      "loss": 0.6519,
+      "step": 9540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.54950146640065e-08,
+      "loss": 0.6015,
+      "step": 9541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5258502936670316e-08,
+      "loss": 0.5125,
+      "step": 9542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5023091974335502e-08,
+      "loss": 0.6905,
+      "step": 9543
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4788781802981276e-08,
+      "loss": 0.7187,
+      "step": 9544
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4555572448464738e-08,
+      "loss": 0.6347,
+      "step": 9545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4323463936521962e-08,
+      "loss": 0.7943,
+      "step": 9546
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.40924562927658e-08,
+      "loss": 0.5313,
+      "step": 9547
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3862549542691404e-08,
+      "loss": 0.5334,
+      "step": 9548
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.363374371166849e-08,
+      "loss": 0.6049,
+      "step": 9549
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3406038824946854e-08,
+      "loss": 0.6681,
+      "step": 9550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3179434907656394e-08,
+      "loss": 0.5692,
+      "step": 9551
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.295393198480378e-08,
+      "loss": 0.5547,
+      "step": 9552
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2729530081273544e-08,
+      "loss": 0.3999,
+      "step": 9553
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2506229221829214e-08,
+      "loss": 0.5547,
+      "step": 9554
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2284029431114408e-08,
+      "loss": 0.5122,
+      "step": 9555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2062930733650624e-08,
+      "loss": 0.3778,
+      "step": 9556
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.184293315383501e-08,
+      "loss": 0.6425,
+      "step": 9557
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.162403671594704e-08,
+      "loss": 0.4859,
+      "step": 9558
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1406241444142938e-08,
+      "loss": 0.4597,
+      "step": 9559
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.118954736245682e-08,
+      "loss": 0.5749,
+      "step": 9560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0973954494802884e-08,
+      "loss": 0.5366,
+      "step": 9561
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0759462864973213e-08,
+      "loss": 0.4868,
+      "step": 9562
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.054607249663665e-08,
+      "loss": 0.4256,
+      "step": 9563
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0333783413343243e-08,
+      "loss": 0.6257,
+      "step": 9564
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0122595638518705e-08,
+      "loss": 0.6694,
+      "step": 9565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9912509195471054e-08,
+      "loss": 0.5921,
+      "step": 9566
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9703524107382853e-08,
+      "loss": 0.6499,
+      "step": 9567
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9495640397316774e-08,
+      "loss": 0.7062,
+      "step": 9568
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288858088214457e-08,
+      "loss": 0.59,
+      "step": 9569
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.908317720289432e-08,
+      "loss": 0.7343,
+      "step": 9570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8878597764054874e-08,
+      "loss": 0.4209,
+      "step": 9571
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.86751197942725e-08,
+      "loss": 0.7434,
+      "step": 9572
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8472743316003684e-08,
+      "loss": 0.7833,
+      "step": 9573
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8271468351579447e-08,
+      "loss": 0.4694,
+      "step": 9574
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8071294923212025e-08,
+      "loss": 0.6735,
+      "step": 9575
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.787222305299152e-08,
+      "loss": 0.8115,
+      "step": 9576
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7674252762888146e-08,
+      "loss": 1.0411,
+      "step": 9577
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.747738407474775e-08,
+      "loss": 0.7223,
+      "step": 9578
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7281617010295182e-08,
+      "loss": 0.6534,
+      "step": 9579
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.708695159113538e-08,
+      "loss": 0.5349,
+      "step": 9580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6893387838750052e-08,
+      "loss": 0.6888,
+      "step": 9581
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6700925774500998e-08,
+      "loss": 0.7021,
+      "step": 9582
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6509565419626783e-08,
+      "loss": 0.5014,
+      "step": 9583
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.631930679524385e-08,
+      "loss": 0.4835,
+      "step": 9584
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6130149922349847e-08,
+      "loss": 0.4928,
+      "step": 9585
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5942094821819187e-08,
+      "loss": 0.6047,
+      "step": 9586
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.575514151440416e-08,
+      "loss": 0.616,
+      "step": 9587
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.556929002073604e-08,
+      "loss": 0.8245,
+      "step": 9588
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5384540361323974e-08,
+      "loss": 0.4097,
+      "step": 9589
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5200892556557213e-08,
+      "loss": 0.5861,
+      "step": 9590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5018346626701765e-08,
+      "loss": 0.6361,
+      "step": 9591
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4836902591902624e-08,
+      "loss": 0.7161,
+      "step": 9592
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4656560472182668e-08,
+      "loss": 0.4701,
+      "step": 9593
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4477320287444863e-08,
+      "loss": 0.5292,
+      "step": 9594
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.429918205746672e-08,
+      "loss": 0.4335,
+      "step": 9595
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4122145801910292e-08,
+      "loss": 0.5818,
+      "step": 9596
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3946211540308841e-08,
+      "loss": 0.6339,
+      "step": 9597
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3771379292080167e-08,
+      "loss": 0.6177,
+      "step": 9598
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3597649076516616e-08,
+      "loss": 0.6514,
+      "step": 9599
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3425020912790632e-08,
+      "loss": 0.6895,
+      "step": 9600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3253494819953638e-08,
+      "loss": 0.4207,
+      "step": 9601
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.308307081693272e-08,
+      "loss": 0.3802,
+      "step": 9602
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2913748922536162e-08,
+      "loss": 0.5845,
+      "step": 9603
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2745529155449021e-08,
+      "loss": 0.7305,
+      "step": 9604
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2578411534235335e-08,
+      "loss": 0.5828,
+      "step": 9605
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.241239607733702e-08,
+      "loss": 0.589,
+      "step": 9606
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2247482803076083e-08,
+      "loss": 0.3163,
+      "step": 9607
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2083671729650192e-08,
+      "loss": 0.5992,
+      "step": 9608
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1920962875137109e-08,
+      "loss": 0.5129,
+      "step": 9609
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1759356257492471e-08,
+      "loss": 0.7794,
+      "step": 9610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.15988518945509e-08,
+      "loss": 0.7024,
+      "step": 9611
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1439449804024894e-08,
+      "loss": 0.4274,
+      "step": 9612
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1281150003505936e-08,
+      "loss": 0.5899,
+      "step": 9613
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1123952510461167e-08,
+      "loss": 0.5069,
+      "step": 9614
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0967857342240041e-08,
+      "loss": 0.5159,
+      "step": 9615
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0812864516067668e-08,
+      "loss": 0.4501,
+      "step": 9616
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0658974049048142e-08,
+      "loss": 0.5823,
+      "step": 9617
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.050618595816455e-08,
+      "loss": 0.6542,
+      "step": 9618
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0354500260277844e-08,
+      "loss": 0.569,
+      "step": 9619
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0203916972127969e-08,
+      "loss": 0.6776,
+      "step": 9620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0054436110330523e-08,
+      "loss": 0.6377,
+      "step": 9621
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.90605769138342e-09,
+      "loss": 0.6984,
+      "step": 9622
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.758781731661337e-09,
+      "loss": 0.6478,
+      "step": 9623
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.612608247414612e-09,
+      "loss": 0.5163,
+      "step": 9624
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.467537254775894e-09,
+      "loss": 0.7839,
+      "step": 9625
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.323568769754599e-09,
+      "loss": 0.7693,
+      "step": 9626
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.180702808238018e-09,
+      "loss": 0.6575,
+      "step": 9627
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.038939385992429e-09,
+      "loss": 0.87,
+      "step": 9628
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.898278518660874e-09,
+      "loss": 0.6207,
+      "step": 9629
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.758720221768713e-09,
+      "loss": 0.6774,
+      "step": 9630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.620264510714737e-09,
+      "loss": 0.5726,
+      "step": 9631
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.482911400777838e-09,
+      "loss": 0.629,
+      "step": 9632
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.346660907118109e-09,
+      "loss": 0.4535,
+      "step": 9633
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.211513044769082e-09,
+      "loss": 0.5546,
+      "step": 9634
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.077467828646602e-09,
+      "loss": 0.581,
+      "step": 9635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.94452527354217e-09,
+      "loss": 0.6898,
+      "step": 9636
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.812685394127384e-09,
+      "loss": 0.6352,
+      "step": 9637
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.681948204951717e-09,
+      "loss": 0.6455,
+      "step": 9638
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.552313720441406e-09,
+      "loss": 0.4364,
+      "step": 9639
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.423781954903897e-09,
+      "loss": 0.6487,
+      "step": 9640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.296352922522287e-09,
+      "loss": 0.6541,
+      "step": 9641
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.170026637359773e-09,
+      "loss": 0.6747,
+      "step": 9642
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.044803113356313e-09,
+      "loss": 0.4894,
+      "step": 9643
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.920682364330855e-09,
+      "loss": 0.6946,
+      "step": 9644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.79766440398244e-09,
+      "loss": 0.6824,
+      "step": 9645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.675749245884655e-09,
+      "loss": 0.5835,
+      "step": 9646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.554936903493403e-09,
+      "loss": 0.5829,
+      "step": 9647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.435227390139132e-09,
+      "loss": 0.7641,
+      "step": 9648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.316620719032385e-09,
+      "loss": 0.7301,
+      "step": 9649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.199116903264912e-09,
+      "loss": 0.7007,
+      "step": 9650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.082715955800789e-09,
+      "loss": 0.5656,
+      "step": 9651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.967417889486404e-09,
+      "loss": 0.7443,
+      "step": 9652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8532227170460254e-09,
+      "loss": 0.5135,
+      "step": 9653
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.740130451080683e-09,
+      "loss": 0.6864,
+      "step": 9654
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.628141104071505e-09,
+      "loss": 0.6536,
+      "step": 9655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.517254688377494e-09,
+      "loss": 0.6709,
+      "step": 9656
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.407471216234417e-09,
+      "loss": 0.767,
+      "step": 9657
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.2987906997581385e-09,
+      "loss": 0.6073,
+      "step": 9658
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.191213150942398e-09,
+      "loss": 0.6985,
+      "step": 9659
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.0847385816577e-09,
+      "loss": 0.6542,
+      "step": 9660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.979367003655755e-09,
+      "loss": 0.6788,
+      "step": 9661
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.875098428563929e-09,
+      "loss": 0.3626,
+      "step": 9662
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.771932867888574e-09,
+      "loss": 0.5094,
+      "step": 9663
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6698703330150255e-09,
+      "loss": 0.475,
+      "step": 9664
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.568910835205387e-09,
+      "loss": 0.5588,
+      "step": 9665
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.469054385602967e-09,
+      "loss": 0.4342,
+      "step": 9666
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.370300995226728e-09,
+      "loss": 0.6101,
+      "step": 9667
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.272650674973511e-09,
+      "loss": 0.5021,
+      "step": 9668
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.176103435620249e-09,
+      "loss": 0.7264,
+      "step": 9669
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.080659287821753e-09,
+      "loss": 0.5306,
+      "step": 9670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.986318242110709e-09,
+      "loss": 0.5258,
+      "step": 9671
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.893080308898789e-09,
+      "loss": 0.6114,
+      "step": 9672
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.800945498473319e-09,
+      "loss": 0.5137,
+      "step": 9673
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.709913821003941e-09,
+      "loss": 0.7039,
+      "step": 9674
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.619985286534844e-09,
+      "loss": 0.3896,
+      "step": 9675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5311599049914212e-09,
+      "loss": 0.3471,
+      "step": 9676
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.443437686174722e-09,
+      "loss": 0.7023,
+      "step": 9677
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.356818639767001e-09,
+      "loss": 0.6593,
+      "step": 9678
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.271302775325058e-09,
+      "loss": 0.5909,
+      "step": 9679
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1868901022880093e-09,
+      "loss": 0.6211,
+      "step": 9680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1035806299706263e-09,
+      "loss": 0.4932,
+      "step": 9681
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.021374367565555e-09,
+      "loss": 0.4549,
+      "step": 9682
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9402713241466485e-09,
+      "loss": 0.711,
+      "step": 9683
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.860271508661194e-09,
+      "loss": 0.5511,
+      "step": 9684
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7813749299399063e-09,
+      "loss": 0.5243,
+      "step": 9685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7035815966891533e-09,
+      "loss": 0.5843,
+      "step": 9686
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.62689151749318e-09,
+      "loss": 0.6023,
+      "step": 9687
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5513047008152158e-09,
+      "loss": 0.5997,
+      "step": 9688
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4768211549974775e-09,
+      "loss": 0.6298,
+      "step": 9689
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4034408882589454e-09,
+      "loss": 0.6185,
+      "step": 9690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3311639086975866e-09,
+      "loss": 0.6639,
+      "step": 9691
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2599902242892437e-09,
+      "loss": 0.5821,
+      "step": 9692
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1899198428876333e-09,
+      "loss": 0.6747,
+      "step": 9693
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.12095277222768e-09,
+      "loss": 0.5758,
+      "step": 9694
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0530890199188525e-09,
+      "loss": 0.6863,
+      "step": 9695
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9863285934496048e-09,
+      "loss": 0.2976,
+      "step": 9696
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.920671500188487e-09,
+      "loss": 0.7141,
+      "step": 9697
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8561177473808146e-09,
+      "loss": 0.6051,
+      "step": 9698
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7926673421497786e-09,
+      "loss": 0.6929,
+      "step": 9699
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7303202914986661e-09,
+      "loss": 0.9201,
+      "step": 9700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6690766023064187e-09,
+      "loss": 0.8689,
+      "step": 9701
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.608936281332074e-09,
+      "loss": 0.6542,
+      "step": 9702
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.549899335213656e-09,
+      "loss": 0.7475,
+      "step": 9703
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4919657704648428e-09,
+      "loss": 0.6428,
+      "step": 9704
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4351355934782985e-09,
+      "loss": 0.6824,
+      "step": 9705
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3794088105278936e-09,
+      "loss": 0.343,
+      "step": 9706
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.324785427760933e-09,
+      "loss": 0.4386,
+      "step": 9707
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2712654512070378e-09,
+      "loss": 0.5264,
+      "step": 9708
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2188488867714843e-09,
+      "loss": 0.6111,
+      "step": 9709
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1675357402385346e-09,
+      "loss": 0.6267,
+      "step": 9710
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1173260172725464e-09,
+      "loss": 0.7262,
+      "step": 9711
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0682197234124226e-09,
+      "loss": 0.5835,
+      "step": 9712
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0202168640782716e-09,
+      "loss": 0.5985,
+      "step": 9713
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.73317444566968e-10,
+      "loss": 0.3763,
+      "step": 9714
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.27521470054371e-10,
+      "loss": 0.3786,
+      "step": 9715
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.828289455953265e-10,
+      "loss": 0.4853,
+      "step": 9716
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.392398761203346e-10,
+      "loss": 0.6178,
+      "step": 9717
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.967542664399918e-10,
+      "loss": 0.5897,
+      "step": 9718
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.553721212438803e-10,
+      "loss": 0.3439,
+      "step": 9719
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.150934450972368e-10,
+      "loss": 0.659,
+      "step": 9720
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.759182424453948e-10,
+      "loss": 0.6993,
+      "step": 9721
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.378465176126725e-10,
+      "loss": 0.5059,
+      "step": 9722
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.008782748001541e-10,
+      "loss": 0.8145,
+      "step": 9723
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.650135180856886e-10,
+      "loss": 0.5697,
+      "step": 9724
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.302522514294417e-10,
+      "loss": 0.5663,
+      "step": 9725
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.965944786661236e-10,
+      "loss": 0.5728,
+      "step": 9726
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.640402035105407e-10,
+      "loss": 0.6304,
+      "step": 9727
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3258942955537455e-10,
+      "loss": 0.5197,
+      "step": 9728
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0224216027118233e-10,
+      "loss": 0.5982,
+      "step": 9729
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7299839900639676e-10,
+      "loss": 0.3171,
+      "step": 9730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4485814898954637e-10,
+      "loss": 0.6913,
+      "step": 9731
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1782141332481475e-10,
+      "loss": 0.661,
+      "step": 9732
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9188819499537113e-10,
+      "loss": 0.7339,
+      "step": 9733
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.670584968655909e-10,
+      "loss": 1.0513,
+      "step": 9734
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.433323216721739e-10,
+      "loss": 0.5109,
+      "step": 9735
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2070967203635663e-10,
+      "loss": 0.5536,
+      "step": 9736
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9919055045392043e-10,
+      "loss": 0.5265,
+      "step": 9737
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.787749592985222e-10,
+      "loss": 0.6666,
+      "step": 9738
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5946290082391457e-10,
+      "loss": 0.6252,
+      "step": 9739
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4125437716172585e-10,
+      "loss": 0.4791,
+      "step": 9740
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2414939032034945e-10,
+      "loss": 0.7904,
+      "step": 9741
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0814794218716451e-10,
+      "loss": 0.5264,
+      "step": 9742
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.325003452964609e-11,
+      "loss": 0.7354,
+      "step": 9743
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.945566899092427e-11,
+      "loss": 0.6239,
+      "step": 9744
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.676484709311481e-11,
+      "loss": 0.4494,
+      "step": 9745
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.517757023731918e-11,
+      "loss": 0.5015,
+      "step": 9746
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.469383970140406e-11,
+      "loss": 0.6426,
+      "step": 9747
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5313656643332085e-11,
+      "loss": 0.6238,
+      "step": 9748
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7037022096720876e-11,
+      "loss": 0.6762,
+      "step": 9749
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9863936976394215e-11,
+      "loss": 0.6421,
+      "step": 9750
+    },
+    {
+      "epoch": 3.0,
+      "step": 9750,
+      "total_flos": 1.19966822529237e+17,
+      "train_loss": 1.0529909285184664,
+      "train_runtime": 11052.0344,
+      "train_samples_per_second": 14.116,
+      "train_steps_per_second": 0.882
+    }
+  ],
+  "max_steps": 9750,
+  "num_train_epochs": 3,
+  "total_flos": 1.19966822529237e+17,
+  "trial_name": null,
+  "trial_params": null
+}