diff --git "a/multimodal-vicuna-7b-v1.5-vision-naivemc/trainer_state.json" "b/multimodal-vicuna-7b-v1.5-vision-naivemc/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/multimodal-vicuna-7b-v1.5-vision-naivemc/trainer_state.json"
@@ -0,0 +1,29281 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998462090531604,
+  "global_step": 4876,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.360544217687075e-06,
+      "loss": 1.3409,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-06,
+      "loss": 1.4285,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.081632653061224e-06,
+      "loss": 1.3418,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4421768707483e-06,
+      "loss": 1.3274,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.802721088435375e-06,
+      "loss": 1.3434,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122448e-06,
+      "loss": 1.2784,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 1.3533,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08843537414966e-05,
+      "loss": 1.3285,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.2893,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.360544217687075e-05,
+      "loss": 1.3474,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4965986394557824e-05,
+      "loss": 1.2319,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-05,
+      "loss": 1.196,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 1.1857,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 1.1224,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 1.1246,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.17687074829932e-05,
+      "loss": 1.1049,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3129251700680275e-05,
+      "loss": 1.0454,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 1.0551,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5850340136054425e-05,
+      "loss": 1.0973,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-05,
+      "loss": 1.0891,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 1.0977,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9931972789115647e-05,
+      "loss": 1.0688,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1292517006802724e-05,
+      "loss": 1.0518,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.265306122448979e-05,
+      "loss": 1.1357,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.401360544217687e-05,
+      "loss": 1.1141,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5374149659863946e-05,
+      "loss": 1.0688,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 1.0257,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.9683,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.945578231292517e-05,
+      "loss": 0.9921,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0816326530612245e-05,
+      "loss": 1.021,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217687074829932e-05,
+      "loss": 1.0337,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35374149659864e-05,
+      "loss": 0.9427,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4897959183673474e-05,
+      "loss": 1.019,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.625850340136055e-05,
+      "loss": 0.9791,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 1.0302,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 1.0105,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.034013605442177e-05,
+      "loss": 1.0364,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.170068027210885e-05,
+      "loss": 1.0055,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3061224489795926e-05,
+      "loss": 1.0439,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4421768707483e-05,
+      "loss": 1.0237,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5782312925170065e-05,
+      "loss": 0.9458,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 1.0469,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.850340136054422e-05,
+      "loss": 0.8926,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9863945578231295e-05,
+      "loss": 1.0232,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591838e-05,
+      "loss": 0.9258,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.258503401360545e-05,
+      "loss": 0.9983,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.394557823129253e-05,
+      "loss": 0.9786,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-05,
+      "loss": 0.8784,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.984,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.802721088435374e-05,
+      "loss": 0.9644,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-05,
+      "loss": 0.9625,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.074829931972789e-05,
+      "loss": 1.0038,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.210884353741498e-05,
+      "loss": 0.9812,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510205e-05,
+      "loss": 0.9399,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.482993197278913e-05,
+      "loss": 0.9731,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 0.9528,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.755102040816327e-05,
+      "loss": 1.0316,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.891156462585034e-05,
+      "loss": 1.0366,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.027210884353742e-05,
+      "loss": 0.9022,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.163265306122449e-05,
+      "loss": 0.9595,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.299319727891157e-05,
+      "loss": 0.9363,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.435374149659864e-05,
+      "loss": 1.0282,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.9666,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.70748299319728e-05,
+      "loss": 0.9871,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.843537414965987e-05,
+      "loss": 1.0303,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.979591836734695e-05,
+      "loss": 0.9472,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.115646258503402e-05,
+      "loss": 0.9752,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.25170068027211e-05,
+      "loss": 0.8979,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.387755102040817e-05,
+      "loss": 0.9567,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.8927,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.659863945578231e-05,
+      "loss": 0.9043,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.79591836734694e-05,
+      "loss": 0.9118,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.931972789115646e-05,
+      "loss": 0.9133,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010068027210884355,
+      "loss": 0.9575,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010204081632653062,
+      "loss": 0.9565,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001034013605442177,
+      "loss": 0.9321,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010476190476190477,
+      "loss": 0.9673,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010612244897959185,
+      "loss": 0.995,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010748299319727892,
+      "loss": 0.8956,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000108843537414966,
+      "loss": 0.9653,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011020408163265306,
+      "loss": 0.9897,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011156462585034013,
+      "loss": 0.9224,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011292517006802721,
+      "loss": 0.9094,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 0.9548,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011564625850340137,
+      "loss": 0.9553,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011700680272108844,
+      "loss": 0.9559,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011836734693877552,
+      "loss": 0.8901,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011972789115646259,
+      "loss": 0.9077,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012108843537414967,
+      "loss": 0.9337,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012244897959183676,
+      "loss": 0.9424,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001238095238095238,
+      "loss": 0.9794,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001251700680272109,
+      "loss": 0.8588,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012653061224489798,
+      "loss": 0.8869,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012789115646258506,
+      "loss": 0.8988,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012925170068027212,
+      "loss": 0.9483,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013061224489795917,
+      "loss": 1.0029,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013197278911564626,
+      "loss": 0.9698,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.9497,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001346938775510204,
+      "loss": 0.9623,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013605442176870748,
+      "loss": 0.9599,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013741496598639456,
+      "loss": 0.8757,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013877551020408165,
+      "loss": 0.8686,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001401360544217687,
+      "loss": 0.9174,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014149659863945578,
+      "loss": 0.9532,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 0.9561,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014421768707482995,
+      "loss": 0.8817,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000145578231292517,
+      "loss": 0.8843,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001469387755102041,
+      "loss": 0.9689,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014829931972789117,
+      "loss": 0.9602,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014965986394557826,
+      "loss": 0.916,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001510204081632653,
+      "loss": 0.9763,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015238095238095237,
+      "loss": 0.9648,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015374149659863945,
+      "loss": 0.8562,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015510204081632654,
+      "loss": 0.9424,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015646258503401362,
+      "loss": 0.8838,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015782312925170067,
+      "loss": 0.9805,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015918367346938776,
+      "loss": 0.9755,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016054421768707484,
+      "loss": 0.9048,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016190476190476192,
+      "loss": 0.8939,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016326530612244898,
+      "loss": 0.9729,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016462585034013606,
+      "loss": 0.985,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016598639455782315,
+      "loss": 0.9615,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016734693877551023,
+      "loss": 0.9038,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016870748299319729,
+      "loss": 0.9703,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017006802721088434,
+      "loss": 0.927,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 0.9106,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001727891156462585,
+      "loss": 0.9189,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001741496598639456,
+      "loss": 0.9323,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017551020408163265,
+      "loss": 0.9225,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017687074829931973,
+      "loss": 0.869,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017823129251700681,
+      "loss": 0.9347,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001795918367346939,
+      "loss": 0.9519,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018095238095238095,
+      "loss": 0.8146,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018231292517006804,
+      "loss": 0.8756,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018367346938775512,
+      "loss": 0.9341,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001850340136054422,
+      "loss": 0.9213,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018639455782312926,
+      "loss": 0.9165,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018775510204081634,
+      "loss": 0.9489,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018911564625850343,
+      "loss": 0.9011,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019047619047619048,
+      "loss": 0.9929,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019183673469387756,
+      "loss": 0.9659,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019319727891156462,
+      "loss": 0.9047,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001945578231292517,
+      "loss": 0.9645,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001959183673469388,
+      "loss": 0.963,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019727891156462587,
+      "loss": 0.9094,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019863945578231293,
+      "loss": 0.9104,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9224,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999997793361936,
+      "loss": 0.9077,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999999117344871,
+      "loss": 0.9727,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999980140263254,
+      "loss": 0.9186,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999964693810426,
+      "loss": 0.8737,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999944834097052,
+      "loss": 0.8427,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999992056113189,
+      "loss": 0.9265,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999891874925656,
+      "loss": 0.8953,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999985877549101,
+      "loss": 0.9688,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999982126284256,
+      "loss": 0.933,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999779336996856,
+      "loss": 0.9199,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999973299797241,
+      "loss": 0.8867,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999682245789666,
+      "loss": 0.9231,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999627080471023,
+      "loss": 0.9302,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999567502040832,
+      "loss": 0.9358,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999503510525376,
+      "loss": 0.8782,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999435105952907,
+      "loss": 0.9181,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999362288353613,
+      "loss": 0.8758,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999285057759622,
+      "loss": 0.8958,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999920341420503,
+      "loss": 0.86,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999117357725855,
+      "loss": 0.9523,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999902688836009,
+      "loss": 0.8713,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998932006147648,
+      "loss": 0.8895,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998832711130412,
+      "loss": 0.9176,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998729003352201,
+      "loss": 0.8923,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998620882858786,
+      "loss": 0.9355,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998508349697883,
+      "loss": 0.8433,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998391403919152,
+      "loss": 0.9041,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999827004557421,
+      "loss": 0.9022,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998144274716611,
+      "loss": 0.9616,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998014091401865,
+      "loss": 0.9276,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997879495687424,
+      "loss": 0.8688,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999774048763269,
+      "loss": 0.9099,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997597067299006,
+      "loss": 0.9828,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997449234749674,
+      "loss": 1.0208,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997296990049935,
+      "loss": 0.9091,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997140333266974,
+      "loss": 0.8272,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996979264469932,
+      "loss": 0.844,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996813783729894,
+      "loss": 0.8245,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999664389111989,
+      "loss": 0.9268,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996469586714896,
+      "loss": 0.9239,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996290870591844,
+      "loss": 0.9627,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996107742829597,
+      "loss": 0.8953,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995920203508982,
+      "loss": 0.8809,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995728252712763,
+      "loss": 0.9431,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995531890525648,
+      "loss": 0.8997,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995331117034305,
+      "loss": 0.9147,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995125932327338,
+      "loss": 0.9438,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994916336495301,
+      "loss": 0.9618,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994702329630693,
+      "loss": 0.9646,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999448391182796,
+      "loss": 0.8878,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199942610831835,
+      "loss": 0.8952,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999403384379565,
+      "loss": 0.8302,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199938021937647,
+      "loss": 0.8771,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993566133192877,
+      "loss": 0.9813,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993325662184371,
+      "loss": 0.9005,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199930807808453,
+      "loss": 0.8721,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992831489283738,
+      "loss": 0.9496,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999257778760971,
+      "loss": 0.9427,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992319675935178,
+      "loss": 0.9215,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992057154374053,
+      "loss": 0.9229,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991790223042191,
+      "loss": 0.9018,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991518882057403,
+      "loss": 0.9472,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991243131539435,
+      "loss": 0.9036,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990962971609978,
+      "loss": 0.8846,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990678402392686,
+      "loss": 0.8934,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990389424013138,
+      "loss": 0.9578,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999009603659887,
+      "loss": 0.8637,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989798240279365,
+      "loss": 0.83,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989496035186047,
+      "loss": 0.9998,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989189421452288,
+      "loss": 0.897,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988878399213405,
+      "loss": 0.9042,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998856296860666,
+      "loss": 0.8988,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988243129771257,
+      "loss": 0.8967,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987918882848359,
+      "loss": 0.8683,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987590227981057,
+      "loss": 0.8992,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987257165314398,
+      "loss": 0.9349,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986919694995378,
+      "loss": 0.9211,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998657781717292,
+      "loss": 0.8845,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986231531997914,
+      "loss": 0.9409,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985880839623181,
+      "loss": 0.8659,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998552574020349,
+      "loss": 0.9601,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985166233895562,
+      "loss": 0.8789,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998480232085805,
+      "loss": 0.9222,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984434001251567,
+      "loss": 0.9189,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984061275238652,
+      "loss": 1.0322,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983684142983808,
+      "loss": 0.9622,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983302604653468,
+      "loss": 0.9359,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998291666041602,
+      "loss": 0.9128,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998252631044179,
+      "loss": 0.8439,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982131554903055,
+      "loss": 0.8729,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998173239397402,
+      "loss": 0.8754,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981328827830855,
+      "loss": 0.8918,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980920856651663,
+      "loss": 0.9058,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980508480616491,
+      "loss": 0.9373,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980091699907333,
+      "loss": 0.9598,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997967051470813,
+      "loss": 0.8909,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979244925204755,
+      "loss": 0.8995,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978814931585038,
+      "loss": 0.8423,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978380534038743,
+      "loss": 0.9213,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977941732757587,
+      "loss": 0.9379,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997749852793522,
+      "loss": 0.9772,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977050919767244,
+      "loss": 0.8528,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976598908451196,
+      "loss": 0.902,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997614249418657,
+      "loss": 0.8955,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975681677174785,
+      "loss": 0.8451,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975216457619217,
+      "loss": 0.8651,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974746835725178,
+      "loss": 0.9673,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974272811699927,
+      "loss": 0.8868,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973794385752666,
+      "loss": 1.0248,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997331155809453,
+      "loss": 0.8671,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997282432893861,
+      "loss": 0.894,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972332698499936,
+      "loss": 0.9236,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971836666995472,
+      "loss": 0.8455,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971336234644133,
+      "loss": 0.935,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970831401666775,
+      "loss": 0.8753,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970322168286192,
+      "loss": 0.9226,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969808534727127,
+      "loss": 0.9835,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969290501216255,
+      "loss": 0.8897,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199687680679822,
+      "loss": 0.9769,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968241235255533,
+      "loss": 0.8896,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996771000326875,
+      "loss": 0.9441,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996717437225631,
+      "loss": 0.8781,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966634342454586,
+      "loss": 0.8549,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996608991410192,
+      "loss": 0.8729,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965541087438578,
+      "loss": 0.9194,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964987862706773,
+      "loss": 0.908,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964430240150666,
+      "loss": 0.9173,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963868220016341,
+      "loss": 0.8455,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963301802551837,
+      "loss": 0.8868,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996273098800713,
+      "loss": 0.8794,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962155776634134,
+      "loss": 0.832,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961576168686708,
+      "loss": 0.94,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996099216442065,
+      "loss": 0.9208,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960403764093693,
+      "loss": 0.877,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959810967965516,
+      "loss": 0.9112,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959213776297744,
+      "loss": 0.9185,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958612189353923,
+      "loss": 0.9135,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958006207399553,
+      "loss": 0.9313,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957395830702071,
+      "loss": 0.8487,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995678105953086,
+      "loss": 0.928,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956161894157224,
+      "loss": 0.9242,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955538334854426,
+      "loss": 0.899,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954910381897654,
+      "loss": 0.8592,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954278035564044,
+      "loss": 0.916,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953641296132672,
+      "loss": 0.9158,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953000163884543,
+      "loss": 0.9685,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952354639102607,
+      "loss": 0.8416,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951704722071756,
+      "loss": 0.8479,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951050413078808,
+      "loss": 0.9584,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950391712412539,
+      "loss": 0.9335,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949728620363642,
+      "loss": 0.925,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949061137224762,
+      "loss": 0.8894,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994838926329048,
+      "loss": 0.96,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947712998857306,
+      "loss": 0.8862,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947032344223698,
+      "loss": 0.9717,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994634729969005,
+      "loss": 0.8867,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994565786555869,
+      "loss": 0.8897,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994496404213388,
+      "loss": 0.9626,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944265829721828,
+      "loss": 0.8997,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943563228630673,
+      "loss": 0.8681,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942856239170493,
+      "loss": 0.9465,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942144861653297,
+      "loss": 0.9145,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941429096393044,
+      "loss": 0.9764,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940708943705617,
+      "loss": 0.8685,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939984403908838,
+      "loss": 0.8649,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939255477322468,
+      "loss": 0.9086,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199385221642682,
+      "loss": 0.9003,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937784465069667,
+      "loss": 0.896,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993704238005244,
+      "loss": 0.9343,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993629590954401,
+      "loss": 0.8583,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935545053873828,
+      "loss": 0.9044,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934789813373261,
+      "loss": 0.8724,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934030188375622,
+      "loss": 0.9259,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933266179216144,
+      "loss": 0.8853,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932497786232022,
+      "loss": 0.9718,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993172500976235,
+      "loss": 0.9178,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930947850148193,
+      "loss": 0.8718,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993016630773252,
+      "loss": 0.9185,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929380382860255,
+      "loss": 0.9344,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928590075878245,
+      "loss": 0.8357,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927795387135274,
+      "loss": 0.9051,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992699631698206,
+      "loss": 0.7627,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926192865771258,
+      "loss": 0.8766,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992538503385745,
+      "loss": 0.9254,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924572821597157,
+      "loss": 0.8768,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923756229348828,
+      "loss": 0.7977,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922935257472849,
+      "loss": 0.905,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922109906331535,
+      "loss": 0.8793,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921280176289142,
+      "loss": 0.8639,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992044606771185,
+      "loss": 0.9299,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919607580967772,
+      "loss": 0.8507,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918764716426958,
+      "loss": 0.8973,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917917474461384,
+      "loss": 0.9187,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917065855444967,
+      "loss": 0.8918,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916209859753545,
+      "loss": 0.8915,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915349487764893,
+      "loss": 0.8696,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991448473985872,
+      "loss": 0.8697,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991361561641666,
+      "loss": 0.908,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912742117822284,
+      "loss": 0.8456,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911864244461088,
+      "loss": 0.8279,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910981996720505,
+      "loss": 0.8458,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910095374989892,
+      "loss": 0.8802,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909204379660542,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908309011125675,
+      "loss": 0.9248,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990740926978044,
+      "loss": 0.9468,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990650515602192,
+      "loss": 0.9131,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990559667024913,
+      "loss": 0.9477,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904683812862998,
+      "loss": 0.8936,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903766584266404,
+      "loss": 0.9168,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990284498486414,
+      "loss": 0.8602,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901919015062936,
+      "loss": 0.93,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900988675271452,
+      "loss": 0.8788,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900053965900264,
+      "loss": 0.8662,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899114887361887,
+      "loss": 0.8909,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898171440070768,
+      "loss": 0.916,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989722362444327,
+      "loss": 0.9047,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896271440897695,
+      "loss": 0.916,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989531488985427,
+      "loss": 0.928,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989435397173514,
+      "loss": 0.8779,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893388686964385,
+      "loss": 0.9438,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892419035968018,
+      "loss": 0.9007,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989144501917397,
+      "loss": 0.9302,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198904666370121,
+      "loss": 0.8907,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198894838899142,
+      "loss": 0.8301,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888496778313973,
+      "loss": 0.8314,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988750530264707,
+      "loss": 0.8061,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886509463351047,
+      "loss": 0.918,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885509260865404,
+      "loss": 0.9001,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884504695631556,
+      "loss": 0.9186,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883495768092843,
+      "loss": 0.9086,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882482478694532,
+      "loss": 0.9387,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881464827883815,
+      "loss": 0.9607,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880442816109812,
+      "loss": 0.8858,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879416443823567,
+      "loss": 0.8516,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878385711478042,
+      "loss": 0.8559,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987735061952813,
+      "loss": 0.8666,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987631116843064,
+      "loss": 0.9025,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987526735864432,
+      "loss": 0.873,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987421919062983,
+      "loss": 0.8837,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873166664849748,
+      "loss": 0.8345,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872109781768587,
+      "loss": 0.8643,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871048541852786,
+      "loss": 0.8348,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869982945570688,
+      "loss": 0.9338,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868912993392576,
+      "loss": 0.8993,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867838685790648,
+      "loss": 0.8263,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866760023239027,
+      "loss": 0.8726,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865677006213753,
+      "loss": 0.9027,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864589635192794,
+      "loss": 0.8893,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986349791065604,
+      "loss": 0.8856,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862401833085292,
+      "loss": 0.8893,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861301402964286,
+      "loss": 0.901,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986019662077867,
+      "loss": 0.8082,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985908748701601,
+      "loss": 0.8446,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857974002165807,
+      "loss": 0.8058,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856856166719465,
+      "loss": 0.8096,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855733981170314,
+      "loss": 0.7847,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854607446013616,
+      "loss": 0.8533,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985347656174653,
+      "loss": 0.93,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852341328868155,
+      "loss": 0.8513,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851201747879497,
+      "loss": 0.8328,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850057819283486,
+      "loss": 0.8748,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848909543584967,
+      "loss": 0.8992,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847756921290705,
+      "loss": 0.9039,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846599952909387,
+      "loss": 0.9079,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845438638951617,
+      "loss": 0.8336,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844272979929913,
+      "loss": 0.9234,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984310297635871,
+      "loss": 0.9084,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841928628754364,
+      "loss": 0.8611,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840749937635147,
+      "loss": 0.9316,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839566903521248,
+      "loss": 0.8537,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838379526934775,
+      "loss": 0.8618,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837187808399747,
+      "loss": 0.9079,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198359917484421,
+      "loss": 0.8857,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834791347589698,
+      "loss": 0.9077,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833586606372302,
+      "loss": 0.8634,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198323775253216,
+      "loss": 0.9493,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831164104971196,
+      "loss": 0.8524,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829946345856602,
+      "loss": 0.8535,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982872424851525,
+      "loss": 0.88,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827497813486484,
+      "loss": 0.8612,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826267041311564,
+      "loss": 0.8644,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825031932533667,
+      "loss": 0.8909,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823792487697877,
+      "loss": 0.8582,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822548707351198,
+      "loss": 0.8918,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821300592042543,
+      "loss": 0.9387,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820048142322737,
+      "loss": 0.8831,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818791358744523,
+      "loss": 0.8566,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817530241862558,
+      "loss": 0.8687,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816264792233402,
+      "loss": 0.9146,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814995010415535,
+      "loss": 0.9384,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981372089696935,
+      "loss": 0.926,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812442452457144,
+      "loss": 0.8784,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981115967744313,
+      "loss": 0.8244,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809872572493434,
+      "loss": 0.8571,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808581138176093,
+      "loss": 0.8865,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807285375061047,
+      "loss": 0.8542,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805985283720157,
+      "loss": 0.915,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804680864727185,
+      "loss": 0.885,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803372118657811,
+      "loss": 0.9308,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980205904608962,
+      "loss": 0.9385,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800741647602106,
+      "loss": 0.866,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799419923776675,
+      "loss": 0.8563,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798093875196635,
+      "loss": 0.904,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796763502447214,
+      "loss": 0.8272,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795428806115538,
+      "loss": 0.9288,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794089786790648,
+      "loss": 0.9729,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979274644506349,
+      "loss": 0.8646,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979139878152692,
+      "loss": 0.8784,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790046796775695,
+      "loss": 0.8458,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788690491406483,
+      "loss": 0.8662,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019787329866017861,
+      "loss": 0.9541,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785964921210312,
+      "loss": 0.8574,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978459565758622,
+      "loss": 0.8733,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783222075749881,
+      "loss": 0.8888,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781844176307496,
+      "loss": 0.8755,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780461959867168,
+      "loss": 0.8095,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019779075427038906,
+      "loss": 0.8211,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977768457843463,
+      "loss": 0.8838,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776289414668154,
+      "loss": 0.8372,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977488993635521,
+      "loss": 0.8824,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019773486144113417,
+      "loss": 0.9279,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772078038562314,
+      "loss": 0.8623,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770665620323335,
+      "loss": 0.8367,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769248890019818,
+      "loss": 0.8505,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767827848277006,
+      "loss": 0.9239,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766402495722046,
+      "loss": 0.9502,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764972832983982,
+      "loss": 0.7702,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763538860693763,
+      "loss": 0.8799,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762100579484248,
+      "loss": 0.872,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976065798999018,
+      "loss": 0.9189,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975921109284822,
+      "loss": 0.835,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757759888696925,
+      "loss": 0.8954,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756304378176748,
+      "loss": 0.9114,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754844561930047,
+      "loss": 0.9319,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753380440601078,
+      "loss": 0.9567,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751912014836,
+      "loss": 0.8181,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750439285282868,
+      "loss": 0.7988,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748962252591636,
+      "loss": 0.8406,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747480917414167,
+      "loss": 0.8612,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745995280404208,
+      "loss": 0.8519,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744505342217416,
+      "loss": 0.7557,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743011103511338,
+      "loss": 0.9048,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741512564945427,
+      "loss": 0.7978,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974000972718103,
+      "loss": 0.8912,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973850259088138,
+      "loss": 0.9115,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736991156711633,
+      "loss": 0.8572,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735475425338816,
+      "loss": 0.8609,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973395539743187,
+      "loss": 0.9064,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973243107366162,
+      "loss": 0.85,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019730902454700796,
+      "loss": 0.7999,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729369541224013,
+      "loss": 0.8633,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727832333907798,
+      "loss": 0.8719,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726290833430555,
+      "loss": 0.871,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019724745040472597,
+      "loss": 0.8976,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972319495571612,
+      "loss": 0.867,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972164057984522,
+      "loss": 0.8215,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972008191354589,
+      "loss": 0.8562,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019718518957506008,
+      "loss": 0.8344,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971695171241535,
+      "loss": 0.953,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715380178965587,
+      "loss": 0.9272,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713804357850278,
+      "loss": 0.861,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971222424976488,
+      "loss": 0.8257,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971063985540673,
+      "loss": 0.7954,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019709051175475077,
+      "loss": 0.8255,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707458210671043,
+      "loss": 0.9301,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019705860961697643,
+      "loss": 0.845,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704259429259794,
+      "loss": 0.8419,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970265361406429,
+      "loss": 0.8483,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701043516819828,
+      "loss": 0.8283,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699429138236987,
+      "loss": 0.8904,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019697810479028232,
+      "loss": 0.8596,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696187539907926,
+      "loss": 0.9018,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969456032159232,
+      "loss": 0.8926,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019692928824799547,
+      "loss": 0.8925,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691293050249628,
+      "loss": 0.8948,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968965299866448,
+      "loss": 0.8735,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688008670767904,
+      "loss": 0.7622,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686360067285583,
+      "loss": 0.8471,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684707188945093,
+      "loss": 0.9103,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019683050036475898,
+      "loss": 0.8582,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968138861060934,
+      "loss": 0.8467,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679722912078655,
+      "loss": 0.8472,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678052941618962,
+      "loss": 0.8603,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019676378699967265,
+      "loss": 0.8749,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019674700187862453,
+      "loss": 0.9944,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673017406045298,
+      "loss": 0.8489,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967133035525846,
+      "loss": 0.8815,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966963903624648,
+      "loss": 0.9149,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667943449755784,
+      "loss": 0.7968,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019666243596534679,
+      "loss": 0.8622,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019664539477333364,
+      "loss": 0.8984,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966283109290391,
+      "loss": 0.8693,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019661118444000268,
+      "loss": 0.8597,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659401531378284,
+      "loss": 0.8439,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965768035579568,
+      "loss": 0.8502,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019655954918012055,
+      "loss": 0.8536,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654225218788897,
+      "loss": 0.8504,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652491258889562,
+      "loss": 0.8759,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196507530390793,
+      "loss": 0.7816,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964901056012524,
+      "loss": 0.9246,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647263822796372,
+      "loss": 0.8972,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964551282786359,
+      "loss": 0.9105,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019643757576099654,
+      "loss": 0.9012,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641998068279208,
+      "loss": 0.8723,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019640234305178767,
+      "loss": 0.9501,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963846628757673,
+      "loss": 0.8998,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636694016253372,
+      "loss": 0.828,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634917491990846,
+      "loss": 0.8206,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963313671557318,
+      "loss": 0.8274,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963135168778628,
+      "loss": 0.8619,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019629562409417928,
+      "loss": 0.8616,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019627768881257782,
+      "loss": 0.8399,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019625971104097377,
+      "loss": 0.8892,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001962416907873012,
+      "loss": 0.8387,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622362805951293,
+      "loss": 0.8562,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019620552286558058,
+      "loss": 0.856,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019618737521349445,
+      "loss": 0.8448,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961691851112636,
+      "loss": 0.9225,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615095256691586,
+      "loss": 0.8574,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961326775884977,
+      "loss": 0.8653,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019611436018407438,
+      "loss": 0.9077,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960960003617299,
+      "loss": 0.8479,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607759812956697,
+      "loss": 0.869,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019605915349570695,
+      "loss": 0.823,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604066646829,
+      "loss": 0.9056,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602213705547498,
+      "loss": 0.8753,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600356526543937,
+      "loss": 0.9493,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019598495110637946,
+      "loss": 0.8234,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959662945865102,
+      "loss": 0.9686,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019594759571406518,
+      "loss": 0.902,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592885449729676,
+      "loss": 0.79,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019591007094447597,
+      "loss": 0.8312,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958912450638925,
+      "loss": 0.9203,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587237686385472,
+      "loss": 0.8653,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019585346635268968,
+      "loss": 0.8786,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583451353874314,
+      "loss": 0.833,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581551843037947,
+      "loss": 0.8857,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579648103598175,
+      "loss": 0.7839,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957774013639517,
+      "loss": 0.8771,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019575827942270975,
+      "loss": 0.8601,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019573911522069486,
+      "loss": 0.902,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571990876636477,
+      "loss": 0.9021,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019570066006819583,
+      "loss": 0.9116,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195681369134683,
+      "loss": 0.8388,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566203597433992,
+      "loss": 0.8856,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019564266059569882,
+      "loss": 0.8992,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019562324300731064,
+      "loss": 0.8957,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019560378321774481,
+      "loss": 0.8718,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019558428123558956,
+      "loss": 0.8897,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955647370694516,
+      "loss": 0.8618,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554515072795633,
+      "loss": 0.8988,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019552552221974775,
+      "loss": 0.8929,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019550585155348848,
+      "loss": 0.8965,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019548613873785967,
+      "loss": 0.8513,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019546638378156118,
+      "loss": 0.8625,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544658669331136,
+      "loss": 0.9,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019542674748184732,
+      "loss": 0.8963,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019540686615592454,
+      "loss": 0.8459,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538694272431726,
+      "loss": 0.8709,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953669771958182,
+      "loss": 0.8177,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019534696957923875,
+      "loss": 0.8375,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019532691988340876,
+      "loss": 0.9099,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019530682811717676,
+      "loss": 0.8522,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019528669428940982,
+      "loss": 0.8611,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952665184089935,
+      "loss": 0.8481,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019524630048483202,
+      "loss": 0.9203,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522604052584806,
+      "loss": 0.8503,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520573854098295,
+      "loss": 0.8691,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518539453919646,
+      "loss": 0.9319,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516500852946703,
+      "loss": 0.8193,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514458052079153,
+      "loss": 0.901,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512411052218537,
+      "loss": 0.8809,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019510359854268259,
+      "loss": 0.9004,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019508304459133567,
+      "loss": 0.8877,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950624486772156,
+      "loss": 0.8389,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504181080941198,
+      "loss": 0.9771,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502113099703283,
+      "loss": 0.8752,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500040924920475,
+      "loss": 0.8295,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949796455750728,
+      "loss": 0.9097,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495883998380057,
+      "loss": 0.8705,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493799248457018,
+      "loss": 0.8732,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491710308658212,
+      "loss": 0.946,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489617179905554,
+      "loss": 0.8572,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487519863122795,
+      "loss": 0.8301,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948541835923554,
+      "loss": 0.8776,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483312669171243,
+      "loss": 0.8927,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481202793859202,
+      "loss": 0.8383,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479088734230556,
+      "loss": 0.8664,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947697049121831,
+      "loss": 0.8646,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474848065757297,
+      "loss": 0.8516,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194727214587842,
+      "loss": 0.8402,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019470590671237552,
+      "loss": 0.8835,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019468455704057727,
+      "loss": 0.9348,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946631655818695,
+      "loss": 0.817,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019464173234569278,
+      "loss": 0.873,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946202573415062,
+      "loss": 0.8192,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459874057878732,
+      "loss": 0.839,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019457718206703208,
+      "loss": 0.8704,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945555818157548,
+      "loss": 0.8921,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019453393983448827,
+      "loss": 0.9025,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019451225613278372,
+      "loss": 0.8899,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449053072021077,
+      "loss": 0.8478,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446876360635745,
+      "loss": 0.8587,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444695480083012,
+      "loss": 0.7565,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442510431325372,
+      "loss": 0.8545,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019440321215327138,
+      "loss": 0.8825,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019438127833054474,
+      "loss": 0.8946,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435930285475382,
+      "loss": 0.8093,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194337285735597,
+      "loss": 0.8442,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019431522698279101,
+      "loss": 0.8709,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194293126606071,
+      "loss": 0.8205,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019427098461519056,
+      "loss": 0.8355,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019424880101992145,
+      "loss": 0.8608,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019422657583005392,
+      "loss": 0.8364,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942043090553966,
+      "loss": 0.8898,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418200070577647,
+      "loss": 0.787,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415965079103872,
+      "loss": 0.8496,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413725932104705,
+      "loss": 0.9056,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941148263056834,
+      "loss": 0.884,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409235175484808,
+      "loss": 0.8586,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019406983567845978,
+      "loss": 0.8386,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019404727808645543,
+      "loss": 0.8298,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940246789887903,
+      "loss": 0.8491,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000194002038395438,
+      "loss": 0.903,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939793563163905,
+      "loss": 0.8618,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193956632761658,
+      "loss": 0.8856,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019393386774126898,
+      "loss": 0.8094,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019391106126527034,
+      "loss": 0.8729,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938882133437272,
+      "loss": 0.7968,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386532398672294,
+      "loss": 0.8389,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938423932043593,
+      "loss": 0.8594,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019381942100675625,
+      "loss": 0.8528,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019379640740405208,
+      "loss": 0.7933,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937733524064033,
+      "loss": 0.8557,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937502560239847,
+      "loss": 0.8416,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937271182669894,
+      "loss": 0.8352,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937039391456287,
+      "loss": 0.8928,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936807186701322,
+      "loss": 0.9191,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019365745685074774,
+      "loss": 0.8483,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936341536977414,
+      "loss": 0.8735,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361080922139748,
+      "loss": 0.8639,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019358742343201856,
+      "loss": 0.8607,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019356399633992547,
+      "loss": 0.8782,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019354052795545718,
+      "loss": 0.853,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019351701828897096,
+      "loss": 0.827,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019349346735084226,
+      "loss": 0.8877,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019346987515146477,
+      "loss": 0.8461,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344624170125035,
+      "loss": 0.86,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934225670106292,
+      "loss": 0.832,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019339885109004946,
+      "loss": 0.8159,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337509394997776,
+      "loss": 0.821,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335129560089868,
+      "loss": 0.8227,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019332745605331515,
+      "loss": 0.8387,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933035753177482,
+      "loss": 0.8668,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019327965340473703,
+      "loss": 0.8875,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019325569032483907,
+      "loss": 0.8477,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019323168608862993,
+      "loss": 0.8052,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320764070670328,
+      "loss": 0.8896,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318355418967102,
+      "loss": 0.8903,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001931594265481632,
+      "loss": 0.8487,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313525779282803,
+      "loss": 0.8588,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311104793433187,
+      "loss": 0.8723,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930867969833591,
+      "loss": 0.8484,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306250495061247,
+      "loss": 0.8757,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930381718468126,
+      "loss": 0.8558,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019301379768269845,
+      "loss": 0.9,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019298938246902698,
+      "loss": 0.86,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929649262165733,
+      "loss": 0.8897,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294042893613063,
+      "loss": 0.8799,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019291589063851028,
+      "loss": 0.8372,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928913113345417,
+      "loss": 0.8325,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928666910350724,
+      "loss": 0.8574,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192842029750968,
+      "loss": 0.8757,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019281732749311224,
+      "loss": 0.8048,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019279258427240686,
+      "loss": 0.8544,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019276780009977172,
+      "loss": 0.8803,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274297498614483,
+      "loss": 0.9161,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271810894248216,
+      "loss": 0.8424,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019269320197975773,
+      "loss": 0.8946,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019266825410896372,
+      "loss": 0.8912,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019264326534111032,
+      "loss": 0.8175,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926182356872258,
+      "loss": 0.8255,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019259316515835633,
+      "loss": 0.8019,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925680537655663,
+      "loss": 0.8683,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019254290151993807,
+      "loss": 0.8857,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192517708432572,
+      "loss": 0.922,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019249247451458645,
+      "loss": 0.8738,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019246719977711793,
+      "loss": 0.8735,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019244188423132082,
+      "loss": 0.9703,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019241652788836762,
+      "loss": 0.931,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019239113075944871,
+      "loss": 0.8369,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236569285577262,
+      "loss": 0.9012,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019234021418856575,
+      "loss": 0.8552,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019231469476907256,
+      "loss": 0.8399,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019228913460855548,
+      "loss": 0.887,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226353371829487,
+      "loss": 0.915,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922378921095892,
+      "loss": 0.8276,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922122097937547,
+      "loss": 0.7806,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921864867821258,
+      "loss": 0.8781,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921607230860547,
+      "loss": 0.9057,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019213491871691166,
+      "loss": 0.859,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019210907368608484,
+      "loss": 0.8559,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920831880049804,
+      "loss": 0.8417,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019205726168502237,
+      "loss": 0.9236,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920312947376528,
+      "loss": 0.8428,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019200528717433157,
+      "loss": 0.7814,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019197923900653657,
+      "loss": 0.9697,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019195315024576355,
+      "loss": 0.9129,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019192702090352622,
+      "loss": 0.7537,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019190085099135616,
+      "loss": 0.7915,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019187464052080287,
+      "loss": 0.7976,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918483895034338,
+      "loss": 0.8482,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019182209795083418,
+      "loss": 0.8741,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019179576587460728,
+      "loss": 0.8872,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019176939328637414,
+      "loss": 0.8759,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019174298019777365,
+      "loss": 0.8494,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019171652662046274,
+      "loss": 0.9178,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191690032566116,
+      "loss": 0.9,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916634980464261,
+      "loss": 0.8536,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019163692307310335,
+      "loss": 0.8375,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019161030765787608,
+      "loss": 0.8424,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019158365181249037,
+      "loss": 0.9597,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155695554871023,
+      "loss": 0.8724,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915302188783174,
+      "loss": 0.9442,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915034418131116,
+      "loss": 0.8889,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914766243649102,
+      "loss": 0.8881,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914497665455485,
+      "loss": 0.9316,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914228683668796,
+      "loss": 0.8542,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019139592984077447,
+      "loss": 0.8381,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913689509791217,
+      "loss": 0.9194,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134193179382792,
+      "loss": 0.8349,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913148722968174,
+      "loss": 0.9049,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019128777250003226,
+      "loss": 0.8738,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019126063241543234,
+      "loss": 0.911,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019123345205499534,
+      "loss": 0.8652,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120623143071674,
+      "loss": 0.8598,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019117897055460968,
+      "loss": 0.803,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911516694387052,
+      "loss": 0.8606,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019112432809505203,
+      "loss": 0.8149,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019109694653571664,
+      "loss": 0.9154,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019106952477278323,
+      "loss": 0.843,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019104206281835386,
+      "loss": 0.8307,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910145606845482,
+      "loss": 0.8411,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019098701838350372,
+      "loss": 0.8927,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019095943592737556,
+      "loss": 0.8445,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909318133283367,
+      "loss": 0.9083,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019090415059857768,
+      "loss": 0.878,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019087644775030684,
+      "loss": 0.8602,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019084870479575024,
+      "loss": 0.8895,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019082092174715158,
+      "loss": 0.8281,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907930986167723,
+      "loss": 0.8945,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019076523541689154,
+      "loss": 0.8596,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073733215980605,
+      "loss": 0.8896,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070938885783036,
+      "loss": 0.9456,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906814055232966,
+      "loss": 0.8538,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019065338216855458,
+      "loss": 0.8649,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019062531880597176,
+      "loss": 0.7554,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019059721544793332,
+      "loss": 0.8401,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019056907210684203,
+      "loss": 0.8391,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019054088879511832,
+      "loss": 0.8895,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019051266552520027,
+      "loss": 0.9037,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019048440230954358,
+      "loss": 0.8819,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001904560991606216,
+      "loss": 0.8531,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019042775609092526,
+      "loss": 0.8718,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903993731129632,
+      "loss": 0.8716,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019037095023926155,
+      "loss": 0.7664,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019034248748236413,
+      "loss": 0.8044,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019031398485483235,
+      "loss": 0.8505,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019028544236924518,
+      "loss": 0.8737,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019025686003819927,
+      "loss": 0.8831,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022823787430873,
+      "loss": 0.867,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019019957589020534,
+      "loss": 0.9141,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901708740985384,
+      "loss": 0.8299,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019014213251197486,
+      "loss": 0.8602,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019011335114319907,
+      "loss": 0.8962,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019008453000491317,
+      "loss": 0.8509,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900556691098366,
+      "loss": 0.8914,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019002676847070657,
+      "loss": 0.8862,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899978281002777,
+      "loss": 0.8398,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996884801132215,
+      "loss": 0.815,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899398282166297,
+      "loss": 0.885,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018991076872900745,
+      "loss": 0.8484,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018988166956128031,
+      "loss": 0.8989,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018985253072629043,
+      "loss": 0.8154,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018982335223689765,
+      "loss": 0.8079,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979413410597924,
+      "loss": 0.8713,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976487634642992,
+      "loss": 0.8255,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189735578971162,
+      "loss": 0.8638,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018970624199310518,
+      "loss": 0.9014,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896768654252067,
+      "loss": 0.8737,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018964744928043128,
+      "loss": 0.8367,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018961799357176102,
+      "loss": 0.8438,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018958849831219557,
+      "loss": 0.8521,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189558963514752,
+      "loss": 0.8328,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018952938919246482,
+      "loss": 0.8198,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018949977535838599,
+      "loss": 0.8944,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894701220255849,
+      "loss": 0.902,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018944042920714845,
+      "loss": 0.898,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018941069691618087,
+      "loss": 0.8478,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893809251658038,
+      "loss": 0.7955,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893511139691564,
+      "loss": 0.8541,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018932126333939513,
+      "loss": 0.8744,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929137328969387,
+      "loss": 0.8466,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018926144383324403,
+      "loss": 0.7866,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018923147498325416,
+      "loss": 0.8742,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018920146675295048,
+      "loss": 0.7912,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018917141915557637,
+      "loss": 0.8726,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018914133220439267,
+      "loss": 0.882,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018911120591267761,
+      "loss": 0.9136,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018908104029372673,
+      "loss": 0.8409,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018905083536085296,
+      "loss": 0.8633,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018902059112738663,
+      "loss": 0.8149,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899030760667523,
+      "loss": 0.9005,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889599848120838,
+      "loss": 0.8791,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889296227569946,
+      "loss": 0.8955,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888992214548073,
+      "loss": 0.7689,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018886878091893872,
+      "loss": 0.7679,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888383011628232,
+      "loss": 0.8521,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018880778219991227,
+      "loss": 0.9254,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877722404367478,
+      "loss": 0.8308,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018874662670759692,
+      "loss": 0.8582,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018871599020518212,
+      "loss": 0.8428,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886853145499511,
+      "loss": 0.8496,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886545997554419,
+      "loss": 0.8648,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018862384583520983,
+      "loss": 0.8345,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018859305280282735,
+      "loss": 0.8394,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018856222067188436,
+      "loss": 0.8773,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885313494559879,
+      "loss": 0.9097,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885004391687623,
+      "loss": 0.8651,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018846948982384915,
+      "loss": 0.8005,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018843850143490721,
+      "loss": 0.8006,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884074740156125,
+      "loss": 0.8458,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883764075796583,
+      "loss": 0.7864,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834530214075512,
+      "loss": 0.8663,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018831415771263058,
+      "loss": 0.8642,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018828297430902963,
+      "loss": 0.8376,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882517519437143,
+      "loss": 0.7803,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018822049063046397,
+      "loss": 0.8176,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018818919038307507,
+      "loss": 0.8409,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018815785121536132,
+      "loss": 0.8623,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018812647314115345,
+      "loss": 0.8174,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018809505617429955,
+      "loss": 0.8926,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880636003286648,
+      "loss": 0.8815,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880321056181315,
+      "loss": 0.8459,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018800057205659913,
+      "loss": 0.8487,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796899965798435,
+      "loss": 0.8424,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018793738843622096,
+      "loss": 0.8258,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018790573840525977,
+      "loss": 0.8056,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878740495790689,
+      "loss": 0.8417,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018784232197163349,
+      "loss": 0.8492,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018781055559695574,
+      "loss": 0.8237,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777875046905508,
+      "loss": 0.8652,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774690660196803,
+      "loss": 0.8596,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018771502400974813,
+      "loss": 0.8565,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000187683102706466,
+      "loss": 0.8017,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876511427062095,
+      "loss": 0.7874,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018761914402308337,
+      "loss": 0.842,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018758710667120954,
+      "loss": 0.8132,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000187555030664727,
+      "loss": 0.7935,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752291601779173,
+      "loss": 0.8741,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018749076274457687,
+      "loss": 0.8584,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018745857085927247,
+      "loss": 0.8169,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018742634037608577,
+      "loss": 0.8442,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018739407130924097,
+      "loss": 0.9019,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736176367297925,
+      "loss": 0.8481,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873294174815589,
+      "loss": 0.8336,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018729703274925516,
+      "loss": 0.852,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018726460949036034,
+      "loss": 0.8889,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872321477191837,
+      "loss": 0.8282,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871996474500515,
+      "loss": 0.8116,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018716710869730706,
+      "loss": 0.9097,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018713453147531058,
+      "loss": 0.8976,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018710191579843926,
+      "loss": 0.8913,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870692616810874,
+      "loss": 0.887,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018703656913766608,
+      "loss": 0.853,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018700383818260344,
+      "loss": 0.8476,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018697106883034458,
+      "loss": 0.8057,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869382610953515,
+      "loss": 0.8326,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018690541499210311,
+      "loss": 0.8682,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018687253053509538,
+      "loss": 0.7506,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018683960773884114,
+      "loss": 0.792,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018680664661787006,
+      "loss": 0.8474,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018677364718672882,
+      "loss": 0.8881,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186740609459981,
+      "loss": 0.8355,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018670753345220704,
+      "loss": 0.8629,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018667441917800427,
+      "loss": 0.9341,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186641266651987,
+      "loss": 0.8701,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660807588878633,
+      "loss": 0.9048,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865748469030502,
+      "loss": 0.8182,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865415797094436,
+      "loss": 0.8306,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018650827432264812,
+      "loss": 0.9141,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018647493075736244,
+      "loss": 0.8245,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186441549028302,
+      "loss": 0.8419,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186408129150199,
+      "loss": 0.8687,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018637467113780261,
+      "loss": 0.9086,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634117500587882,
+      "loss": 0.8435,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018630764076921032,
+      "loss": 0.751,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018627406844259672,
+      "loss": 0.7803,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018624045804085444,
+      "loss": 0.7961,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018620680957881665,
+      "loss": 0.8619,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018617312307133332,
+      "loss": 0.8021,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613939853327129,
+      "loss": 0.8665,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018610563597951414,
+      "loss": 0.8849,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018607183542496215,
+      "loss": 0.8355,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018603799688453245,
+      "loss": 0.8151,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186004120373159,
+      "loss": 0.9679,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018597020590579235,
+      "loss": 0.8449,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018593625349739993,
+      "loss": 0.8701,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859022631629659,
+      "loss": 0.8953,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858682349174911,
+      "loss": 0.8481,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858341687759931,
+      "loss": 0.7509,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018580006475350628,
+      "loss": 0.8372,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857659228650817,
+      "loss": 0.7757,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018573174312578708,
+      "loss": 0.8691,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018569752555070687,
+      "loss": 0.8237,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018566327015494228,
+      "loss": 0.8665,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018562897695361115,
+      "loss": 0.8158,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000185594645961848,
+      "loss": 0.8619,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018556027719480404,
+      "loss": 0.8447,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018552587066764716,
+      "loss": 0.8794,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018549142639556196,
+      "loss": 0.8696,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018545694439374958,
+      "loss": 0.9261,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018542242467742788,
+      "loss": 0.8357,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853878672618314,
+      "loss": 0.8622,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853532721622113,
+      "loss": 0.91,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853186393938353,
+      "loss": 0.8607,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852839689719878,
+      "loss": 0.9448,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852492609119699,
+      "loss": 0.8074,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018521451522909908,
+      "loss": 0.8646,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851797319387097,
+      "loss": 0.8688,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851449110561525,
+      "loss": 0.8271,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018511005259679494,
+      "loss": 0.86,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018507515657602107,
+      "loss": 0.806,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018504022300923137,
+      "loss": 0.8488,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018500525191184302,
+      "loss": 0.861,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018497024329928974,
+      "loss": 0.8867,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849351971870218,
+      "loss": 0.8377,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184900113590506,
+      "loss": 0.9318,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018486499252522574,
+      "loss": 0.8605,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018482983400668087,
+      "loss": 0.7957,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018479463805038787,
+      "loss": 0.8753,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847594046718796,
+      "loss": 0.8848,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018472413388670558,
+      "loss": 0.8622,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018468882571043182,
+      "loss": 0.8561,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018465348015864072,
+      "loss": 0.8442,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018461809724693132,
+      "loss": 0.841,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018458267699091895,
+      "loss": 0.8597,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018454721940623569,
+      "loss": 0.8178,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451172450852988,
+      "loss": 0.7709,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447619231346644,
+      "loss": 0.8321,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018444062283672663,
+      "loss": 0.8708,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018440501609400832,
+      "loss": 0.8981,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018436937210102573,
+      "loss": 0.8354,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018433369087350948,
+      "loss": 0.8364,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018429797242720678,
+      "loss": 0.9265,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018426221677788108,
+      "loss": 0.877,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842264239413124,
+      "loss": 0.8258,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018419059393329709,
+      "loss": 0.9086,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018415472676964787,
+      "loss": 0.8129,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411882246619396,
+      "loss": 0.8535,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840828810387809,
+      "loss": 0.9288,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018404690250327068,
+      "loss": 0.8184,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840108868755416,
+      "loss": 0.8046,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018397483417148828,
+      "loss": 0.8651,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018393874440702186,
+      "loss": 0.8575,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018390261759806972,
+      "loss": 0.8702,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018386645376057561,
+      "loss": 0.9166,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018383025291049968,
+      "loss": 0.7445,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379401506381826,
+      "loss": 0.8323,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837577402365242,
+      "loss": 0.8983,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837214284446266,
+      "loss": 0.8368,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018368507970415078,
+      "loss": 0.817,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018364869403113846,
+      "loss": 0.8901,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836122714416477,
+      "loss": 0.8362,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018357581195175274,
+      "loss": 0.7963,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835393155775442,
+      "loss": 0.761,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018350278233512888,
+      "loss": 0.9001,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018346621224062993,
+      "loss": 0.8088,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018342960531018678,
+      "loss": 0.8107,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018339296155995505,
+      "loss": 0.8738,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018335628100610665,
+      "loss": 0.8515,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833195636648297,
+      "loss": 0.8145,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832828095523286,
+      "loss": 0.8106,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018324601868482395,
+      "loss": 0.8038,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018320919107855254,
+      "loss": 0.8801,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018317232674976746,
+      "loss": 0.9309,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018313542571473792,
+      "loss": 0.7789,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830984879897494,
+      "loss": 0.8831,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018306151359110353,
+      "loss": 0.8226,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830245025351181,
+      "loss": 0.7975,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018298745483812714,
+      "loss": 0.8354,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018295037051648083,
+      "loss": 0.8312,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018291324958654548,
+      "loss": 0.8585,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828760920647036,
+      "loss": 0.8213,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828388979673538,
+      "loss": 0.8121,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018280166731091092,
+      "loss": 0.8596,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827644001118058,
+      "loss": 0.8331,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018272709638648556,
+      "loss": 0.8409,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018268975615141333,
+      "loss": 0.884,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018265237942306835,
+      "loss": 0.8699,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018261496621794606,
+      "loss": 0.868,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018257751655255793,
+      "loss": 0.7948,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018254003044343152,
+      "loss": 0.8292,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825025079071105,
+      "loss": 0.8149,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018246494896015456,
+      "loss": 0.8835,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824273536191395,
+      "loss": 0.8201,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018238972190065726,
+      "loss": 0.8879,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823520538213157,
+      "loss": 0.8629,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018231434939773876,
+      "loss": 0.8614,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822766086465665,
+      "loss": 0.8207,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018223883158445493,
+      "loss": 0.8209,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018220101822807608,
+      "loss": 0.8633,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821631685941181,
+      "loss": 0.8948,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000182125282699285,
+      "loss": 0.8762,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018208736056029694,
+      "loss": 0.924,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018204940219388997,
+      "loss": 0.8857,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018201140761681613,
+      "loss": 0.8149,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018197337684584358,
+      "loss": 0.8653,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018193530989775623,
+      "loss": 0.7899,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018189720678935415,
+      "loss": 0.8712,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018185906753745329,
+      "loss": 0.8738,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018182089215888552,
+      "loss": 0.7356,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018178268067049874,
+      "loss": 0.832,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018174443308915667,
+      "loss": 0.9276,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018170614943173905,
+      "loss": 0.8767,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166782971514156,
+      "loss": 0.8569,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816294739562757,
+      "loss": 0.8958,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018159108217206892,
+      "loss": 0.8311,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815526543794646,
+      "loss": 0.7892,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151419059542198,
+      "loss": 0.9066,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814756908369162,
+      "loss": 0.8357,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814371551209382,
+      "loss": 0.8045,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018139858346449494,
+      "loss": 0.8213,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018135997588460915,
+      "loss": 0.8385,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813213323983194,
+      "loss": 0.8188,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812826530226801,
+      "loss": 0.8398,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018124393777476153,
+      "loss": 0.8509,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018120518667164988,
+      "loss": 0.877,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181166399730447,
+      "loss": 0.8876,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018112757696827063,
+      "loss": 0.8448,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810887184022544,
+      "loss": 0.8232,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018104982404954761,
+      "loss": 0.8547,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810108939273154,
+      "loss": 0.8569,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018097192805273876,
+      "loss": 0.8843,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018093292644301437,
+      "loss": 0.7874,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018089388911535475,
+      "loss": 0.8452,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085481608698812,
+      "loss": 0.842,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081570737515847,
+      "loss": 0.8432,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807765629971256,
+      "loss": 0.8577,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018073738297016498,
+      "loss": 0.7978,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069816731156784,
+      "loss": 0.8027,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065891603864114,
+      "loss": 0.8513,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018061962916870756,
+      "loss": 0.7884,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018058030671910547,
+      "loss": 0.8239,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018054094870718893,
+      "loss": 0.8203,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018050155515032774,
+      "loss": 0.8444,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018046212606590735,
+      "loss": 0.9694,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018042266147132893,
+      "loss": 0.8834,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803831613840093,
+      "loss": 0.8565,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018034362582138087,
+      "loss": 0.8535,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018030405480089185,
+      "loss": 0.7768,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000180264448340006,
+      "loss": 0.8544,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018022480645620278,
+      "loss": 0.7435,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018018512916697716,
+      "loss": 0.8249,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018014541648983992,
+      "loss": 0.8226,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801056684423173,
+      "loss": 0.8592,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018006588504195122,
+      "loss": 0.8527,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800260663062992,
+      "loss": 0.8506,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017998621225293434,
+      "loss": 0.8143,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017994632289944533,
+      "loss": 0.9019,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799063982634365,
+      "loss": 0.8532,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798664383625276,
+      "loss": 0.8206,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017982644321435412,
+      "loss": 0.8675,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000179786412836567,
+      "loss": 0.825,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797463472468327,
+      "loss": 0.829,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017970624646283337,
+      "loss": 0.8955,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017966611050226647,
+      "loss": 0.8246,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796259393828452,
+      "loss": 0.8182,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017958573312229818,
+      "loss": 0.8362,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795454917383695,
+      "loss": 0.852,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017950521524881885,
+      "loss": 0.8457,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017946490367142128,
+      "loss": 0.7705,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017942455702396747,
+      "loss": 0.7928,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017938417532426345,
+      "loss": 0.8431,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017934375859013086,
+      "loss": 0.8533,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017930330683940663,
+      "loss": 0.8733,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017926282008994332,
+      "loss": 0.848,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792222983596088,
+      "loss": 0.8987,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017918174166628643,
+      "loss": 0.9035,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017914115002787498,
+      "loss": 0.8407,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017910052346228872,
+      "loss": 0.8461,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017905986198745724,
+      "loss": 0.8267,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017901916562132553,
+      "loss": 0.8782,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017897843438185413,
+      "loss": 0.8071,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893766828701877,
+      "loss": 0.8704,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017889686735481067,
+      "loss": 0.8438,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788560316032364,
+      "loss": 0.8217,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017881516105031794,
+      "loss": 0.8067,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001787742557140926,
+      "loss": 0.8279,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017873331561261296,
+      "loss": 0.8981,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786923407639471,
+      "loss": 0.8625,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786513311861783,
+      "loss": 0.7549,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017861028689740525,
+      "loss": 0.8757,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785692079157419,
+      "loss": 0.8489,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785280942593176,
+      "loss": 0.8801,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017848694594627685,
+      "loss": 0.834,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017844576299477958,
+      "loss": 0.854,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178404545423001,
+      "loss": 0.7974,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017836329324913148,
+      "loss": 0.8982,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017832200649137682,
+      "loss": 0.7966,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178280685167958,
+      "loss": 0.8554,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782393292971112,
+      "loss": 0.7755,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017819793889708798,
+      "loss": 0.9409,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017815651398615502,
+      "loss": 0.8795,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781150545825943,
+      "loss": 0.85,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178073560704703,
+      "loss": 0.8589,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780320323707935,
+      "loss": 0.8691,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017799046959919337,
+      "loss": 0.8312,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017794887240824548,
+      "loss": 0.9182,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017790724081630777,
+      "loss": 0.8514,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778655748417534,
+      "loss": 0.8362,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017782387450297077,
+      "loss": 0.968,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017778213981836338,
+      "loss": 0.8287,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017774037080634983,
+      "loss": 0.8866,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017769856748536404,
+      "loss": 0.8769,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017765672987385487,
+      "loss": 0.8579,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776148579902865,
+      "loss": 0.8431,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017757295185313806,
+      "loss": 0.8543,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177531011480904,
+      "loss": 0.9229,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017748903689209365,
+      "loss": 0.8602,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774470281052316,
+      "loss": 0.8538,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017740498513885752,
+      "loss": 0.8965,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017736290801152608,
+      "loss": 0.7871,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017732079674180712,
+      "loss": 0.8813,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772786513482855,
+      "loss": 0.8301,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723647184956113,
+      "loss": 0.9451,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017719425826424895,
+      "loss": 0.8643,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017715201061097906,
+      "loss": 0.8521,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017710972890839645,
+      "loss": 0.8477,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706741317516126,
+      "loss": 0.8557,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017702506342994856,
+      "loss": 0.8728,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017698267969144845,
+      "loss": 0.842,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017694026197836605,
+      "loss": 0.8336,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017689781030942153,
+      "loss": 0.7038,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017685532470334986,
+      "loss": 0.8984,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017681280517890121,
+      "loss": 0.8077,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017677025175484058,
+      "loss": 0.8033,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017672766444994803,
+      "loss": 0.8457,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017668504328301844,
+      "loss": 0.8187,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766423882728617,
+      "loss": 0.8884,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017659969943830273,
+      "loss": 0.8235,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765569767981812,
+      "loss": 0.8411,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017651422037135185,
+      "loss": 0.8774,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017647143017668427,
+      "loss": 0.7808,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764286062330629,
+      "loss": 0.8528,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017638574855938718,
+      "loss": 0.882,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017634285717457138,
+      "loss": 0.8248,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017629993209754462,
+      "loss": 0.8029,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017625697334725099,
+      "loss": 0.8201,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017621398094264928,
+      "loss": 0.8765,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761709549027133,
+      "loss": 0.8013,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017612789524643158,
+      "loss": 0.8882,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760848019928076,
+      "loss": 0.7915,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017604167516085953,
+      "loss": 0.7681,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017599851476962045,
+      "loss": 0.8162,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017595532083813825,
+      "loss": 0.9077,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017591209338547561,
+      "loss": 0.8175,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017586883243070998,
+      "loss": 0.8369,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017582553799293363,
+      "loss": 0.858,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017578221009125356,
+      "loss": 0.7997,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017573884874479164,
+      "loss": 0.831,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017569545397268433,
+      "loss": 0.8106,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175652025794083,
+      "loss": 0.867,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017560856422815373,
+      "loss": 0.8389,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017556506929407726,
+      "loss": 0.876,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017552154101104913,
+      "loss": 0.9227,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017547797939827955,
+      "loss": 0.8376,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017543438447499348,
+      "loss": 0.873,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017539075626043057,
+      "loss": 0.9146,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534709477384514,
+      "loss": 0.8754,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530340003450622,
+      "loss": 0.8409,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017525967206169748,
+      "loss": 0.8831,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017521591087471734,
+      "loss": 0.8447,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017517211649287873,
+      "loss": 0.8538,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017512828893550942,
+      "loss": 0.9461,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017508442822195168,
+      "loss": 0.8916,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017504053437156244,
+      "loss": 0.8794,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017499660740371325,
+      "loss": 0.8962,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017495264733779036,
+      "loss": 0.8123,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017490865419319448,
+      "loss": 0.8156,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017486462798934106,
+      "loss": 0.8062,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748205687456601,
+      "loss": 0.8293,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017477647648159606,
+      "loss": 0.8467,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017473235121660819,
+      "loss": 0.8684,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017468819297017008,
+      "loss": 0.8563,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017464400176177007,
+      "loss": 0.8546,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745997776109109,
+      "loss": 0.8299,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017455552053710998,
+      "loss": 0.8518,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745112305598991,
+      "loss": 0.8025,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744669076988247,
+      "loss": 0.892,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017442255197344767,
+      "loss": 0.7875,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017437816340334342,
+      "loss": 0.8798,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001743337420081018,
+      "loss": 0.9205,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017428928780732728,
+      "loss": 0.7819,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017424480082063866,
+      "loss": 0.9023,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017420028106766936,
+      "loss": 0.813,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741557285680671,
+      "loss": 0.8029,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017411114334149415,
+      "loss": 0.8373,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740665254076272,
+      "loss": 0.8214,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017402187478615736,
+      "loss": 0.8983,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739771914967902,
+      "loss": 0.8287,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017393247555924575,
+      "loss": 0.7708,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017388772699325825,
+      "loss": 0.899,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738429458185766,
+      "loss": 0.863,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017379813205496393,
+      "loss": 0.8576,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017375328572219776,
+      "loss": 0.9175,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017370840684007005,
+      "loss": 0.844,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736634954283871,
+      "loss": 0.8673,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017361855150696952,
+      "loss": 0.7297,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735735750956523,
+      "loss": 0.8774,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017352856621428482,
+      "loss": 0.8596,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017348352488273072,
+      "loss": 0.7917,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017343845112086794,
+      "loss": 0.9132,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017339334494858886,
+      "loss": 0.9066,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733482063858,
+      "loss": 0.8794,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017330303545242233,
+      "loss": 0.8422,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017325783216839096,
+      "loss": 0.8843,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017321259655365536,
+      "loss": 0.8853,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017316732862817928,
+      "loss": 0.8413,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017312202841194071,
+      "loss": 0.8976,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017307669592493187,
+      "loss": 0.8802,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017303133118715922,
+      "loss": 0.7693,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729859342186435,
+      "loss": 0.8872,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017294050503941966,
+      "loss": 0.8606,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017289504366953682,
+      "loss": 0.8777,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017284955012905834,
+      "loss": 0.8322,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001728040244380618,
+      "loss": 0.7986,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727584666166389,
+      "loss": 0.8969,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017271287668489565,
+      "loss": 0.7661,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017266725466295201,
+      "loss": 0.8675,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017262160057094237,
+      "loss": 0.8902,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017257591442901507,
+      "loss": 0.7814,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017253019625733268,
+      "loss": 0.8766,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001724844460760719,
+      "loss": 0.7924,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017243866390542355,
+      "loss": 0.7717,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017239284976559255,
+      "loss": 0.845,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017234700367679796,
+      "loss": 0.822,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017230112565927292,
+      "loss": 0.8752,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017225521573326465,
+      "loss": 0.8952,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001722092739190345,
+      "loss": 0.7415,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017216330023685786,
+      "loss": 0.816,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017211729470702414,
+      "loss": 0.7978,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720712573498369,
+      "loss": 0.8408,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017202518818561368,
+      "loss": 0.8785,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017197908723468606,
+      "loss": 0.8531,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017193295451739968,
+      "loss": 0.8423,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017188679005411417,
+      "loss": 0.8032,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718405938652032,
+      "loss": 0.878,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717943659710544,
+      "loss": 0.8795,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017174810639206944,
+      "loss": 0.8341,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017170181514866392,
+      "loss": 0.92,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017165549226126747,
+      "loss": 0.8876,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017160913775032368,
+      "loss": 0.829,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017156275163629001,
+      "loss": 0.842,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171516333939638,
+      "loss": 0.8959,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017146988468085304,
+      "loss": 0.763,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017142340388043442,
+      "loss": 0.8885,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017137689155889547,
+      "loss": 0.8945,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017133034773676332,
+      "loss": 0.8984,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128377243457907,
+      "loss": 0.8692,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017123716567289768,
+      "loss": 0.8663,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017119052747228798,
+      "loss": 0.848,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017114385785333273,
+      "loss": 0.779,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710971568366285,
+      "loss": 0.8207,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105042444278575,
+      "loss": 0.8728,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017100366069242875,
+      "loss": 0.8481,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017095686560619566,
+      "loss": 0.868,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017091003920473842,
+      "loss": 0.8145,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017086318150872287,
+      "loss": 0.8487,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017081629253882854,
+      "loss": 0.9011,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017076937231574885,
+      "loss": 0.8174,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170722420860191,
+      "loss": 0.8306,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017067543819287593,
+      "loss": 0.811,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017062842433453846,
+      "loss": 0.8408,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017058137930592703,
+      "loss": 0.8191,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017053430312780396,
+      "loss": 0.8319,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017048719582094524,
+      "loss": 0.8861,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704400574061406,
+      "loss": 0.8613,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017039288790419361,
+      "loss": 0.8376,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017034568733592138,
+      "loss": 0.8049,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017029845572215485,
+      "loss": 0.8263,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017025119308373867,
+      "loss": 0.8348,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017020389944153114,
+      "loss": 0.8123,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017015657481640425,
+      "loss": 0.8309,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017010921922924358,
+      "loss": 0.8803,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017006183270094858,
+      "loss": 0.8476,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700144152524322,
+      "loss": 0.8802,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169966966904621,
+      "loss": 0.7611,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016991948767845533,
+      "loss": 0.8434,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016987197759488906,
+      "loss": 0.8002,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016982443667488964,
+      "loss": 0.8778,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697768649394383,
+      "loss": 0.8542,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972926240952968,
+      "loss": 0.8151,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016968162910617214,
+      "loss": 0.8893,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016963396505038754,
+      "loss": 0.8859,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695862702632113,
+      "loss": 0.8866,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016953854476569252,
+      "loss": 0.89,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694907885788938,
+      "loss": 0.8664,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016944300172389118,
+      "loss": 0.8193,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693951842217744,
+      "loss": 0.8501,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016934733609364657,
+      "loss": 0.7918,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016929945736062442,
+      "loss": 0.8252,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016925154804383816,
+      "loss": 0.8143,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016920360816443153,
+      "loss": 0.8247,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016915563774356164,
+      "loss": 0.8854,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016910763680239924,
+      "loss": 0.8177,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016905960536212845,
+      "loss": 0.8201,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016901154344394685,
+      "loss": 0.8409,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689634510690655,
+      "loss": 0.8652,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689153282587089,
+      "loss": 0.9061,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016886717503411495,
+      "loss": 0.7695,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016881899141653504,
+      "loss": 0.7941,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877077742723392,
+      "loss": 0.8761,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687225330874897,
+      "loss": 0.8491,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016867425841859404,
+      "loss": 0.8581,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686259534418518,
+      "loss": 0.8587,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016857761817858134,
+      "loss": 0.8619,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685292526501143,
+      "loss": 0.7573,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016848085687779578,
+      "loss": 0.824,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016843243088298415,
+      "loss": 0.823,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016838397468705113,
+      "loss": 0.8043,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001683354883113818,
+      "loss": 0.8616,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016828697177737452,
+      "loss": 0.7974,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016823842510644096,
+      "loss": 0.8135,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016818984832000613,
+      "loss": 0.8104,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814124143950832,
+      "loss": 0.9152,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016809260448639904,
+      "loss": 0.779,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016804393748214315,
+      "loss": 0.8745,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016799524044821876,
+      "loss": 0.908,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679465134061172,
+      "loss": 0.788,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016789775637734305,
+      "loss": 0.8584,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016784896938341413,
+      "loss": 0.8489,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001678001524458615,
+      "loss": 0.8363,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016775130558622942,
+      "loss": 0.756,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016770242882607539,
+      "loss": 0.8943,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016765352218697,
+      "loss": 0.8625,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016760458569049718,
+      "loss": 0.843,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016755561935825387,
+      "loss": 0.7928,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675066232118504,
+      "loss": 0.7971,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016745759727290996,
+      "loss": 0.81,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016740854156306912,
+      "loss": 0.8128,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016735945610397757,
+      "loss": 0.878,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016731034091729802,
+      "loss": 0.8074,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672611960247064,
+      "loss": 0.8728,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016721202144789166,
+      "loss": 0.7907,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671628172085559,
+      "loss": 0.8043,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711358332841436,
+      "loss": 0.7784,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016706431982919527,
+      "loss": 0.8442,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016701502673263999,
+      "loss": 0.7568,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016696570406050286,
+      "loss": 0.8672,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016691635183455148,
+      "loss": 0.8406,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668669700765662,
+      "loss": 0.8568,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016681755880834068,
+      "loss": 0.8244,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016676811805168136,
+      "loss": 0.7889,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001667186478284079,
+      "loss": 0.834,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016666914816035281,
+      "loss": 0.8685,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016661961906936173,
+      "loss": 0.851,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016657006057729315,
+      "loss": 0.8006,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016652047270601864,
+      "loss": 0.8098,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016647085547742268,
+      "loss": 0.8871,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016642120891340273,
+      "loss": 0.8571,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663715330358692,
+      "loss": 0.8411,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663218278667454,
+      "loss": 0.856,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016627209342796764,
+      "loss": 0.8467,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016622232974148503,
+      "loss": 0.835,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661725368292597,
+      "loss": 0.8499,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016612271471326667,
+      "loss": 0.8094,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016607286341549374,
+      "loss": 0.8887,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016602298295794173,
+      "loss": 0.8401,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016597307336262425,
+      "loss": 0.8487,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016592313465156776,
+      "loss": 0.8832,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001658731668468116,
+      "loss": 0.8586,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016582316997040795,
+      "loss": 0.7796,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001657731440444218,
+      "loss": 0.9321,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016572308909093097,
+      "loss": 0.8372,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656730051320261,
+      "loss": 0.8299,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016562289218981067,
+      "loss": 0.84,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016557275028640083,
+      "loss": 0.8141,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016552257944392563,
+      "loss": 0.8787,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016547237968452682,
+      "loss": 0.897,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016542215103035898,
+      "loss": 0.8486,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016537189350358938,
+      "loss": 0.7325,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016532160712639805,
+      "loss": 0.8305,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016527129192097774,
+      "loss": 0.8222,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000165220947909534,
+      "loss": 0.8841,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016517057511428496,
+      "loss": 0.838,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016512017355746155,
+      "loss": 0.777,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650697432613074,
+      "loss": 0.8131,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016501928424807874,
+      "loss": 0.9091,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016496879654004457,
+      "loss": 0.8669,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016491828015948652,
+      "loss": 0.8941,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016486773512869882,
+      "loss": 0.8622,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648171614699884,
+      "loss": 0.8347,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647665592056748,
+      "loss": 0.8398,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016471592835809026,
+      "loss": 0.7354,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016466526894957947,
+      "loss": 0.7938,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016461458100249993,
+      "loss": 0.8792,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016456386453922157,
+      "loss": 0.745,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016451311958212697,
+      "loss": 0.8245,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016446234615361132,
+      "loss": 0.8027,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016441154427608226,
+      "loss": 0.8187,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016436071397196012,
+      "loss": 0.8323,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643098552636777,
+      "loss": 0.8236,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016425896817368036,
+      "loss": 0.8676,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016420805272442598,
+      "loss": 0.808,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016415710893838493,
+      "loss": 0.8434,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016410613683804016,
+      "loss": 0.8183,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016405513644588698,
+      "loss": 0.851,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016400410778443335,
+      "loss": 0.7961,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639530508761996,
+      "loss": 0.8707,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016390196574371854,
+      "loss": 0.8325,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638508524095355,
+      "loss": 0.8023,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016379971089620815,
+      "loss": 0.8637,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016374854122630668,
+      "loss": 0.8766,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016369734342241366,
+      "loss": 0.8287,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016364611750712408,
+      "loss": 0.7871,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016359486350304541,
+      "loss": 0.8489,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635435814327974,
+      "loss": 0.9072,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634922713190123,
+      "loss": 0.8332,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634409331843346,
+      "loss": 0.8599,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633895670514213,
+      "loss": 0.8042,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633381729429417,
+      "loss": 0.9131,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016328675088157738,
+      "loss": 0.8032,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016323530089002236,
+      "loss": 0.8541,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016318382299098297,
+      "loss": 0.8431,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016313231720717773,
+      "loss": 0.9579,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016308078356133766,
+      "loss": 0.8217,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016302922207620592,
+      "loss": 0.7887,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016297763277453808,
+      "loss": 0.8862,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016292601567910187,
+      "loss": 0.8396,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016287437081267733,
+      "loss": 0.9218,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016282269819805682,
+      "loss": 0.8117,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016277099785804485,
+      "loss": 0.8473,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016271926981545822,
+      "loss": 0.7611,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266751409312597,
+      "loss": 0.8086,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016261573071388929,
+      "loss": 0.8805,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016256391970060163,
+      "loss": 0.8246,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625120810761286,
+      "loss": 0.865,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016246021486334805,
+      "loss": 0.8549,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016240832108514998,
+      "loss": 0.782,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623563997644365,
+      "loss": 0.8801,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016230445092412196,
+      "loss": 0.881,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622524745871328,
+      "loss": 0.8401,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016220047077640764,
+      "loss": 0.9019,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016214843951489714,
+      "loss": 0.8041,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016209638082556417,
+      "loss": 0.8941,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204429473138366,
+      "loss": 0.8676,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016199218125534267,
+      "loss": 0.8434,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619400404204403,
+      "loss": 0.8627,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016188787224968772,
+      "loss": 0.9528,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618356767661082,
+      "loss": 0.862,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016178345399273707,
+      "loss": 0.8658,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617312039526216,
+      "loss": 0.8085,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016167892666882127,
+      "loss": 0.8867,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016162662216440748,
+      "loss": 0.7968,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615742904624636,
+      "loss": 0.8907,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016152193158608508,
+      "loss": 0.8012,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614695455583793,
+      "loss": 0.8186,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016141713240246574,
+      "loss": 0.8875,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016136469214147572,
+      "loss": 0.9232,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613122247985526,
+      "loss": 0.8661,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016125973039685164,
+      "loss": 0.7312,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120720895954004,
+      "loss": 0.8041,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016115466050979705,
+      "loss": 0.8531,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016110208507081372,
+      "loss": 0.8035,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016104948266579302,
+      "loss": 0.7991,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016099685331794983,
+      "loss": 0.8094,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016094419705051097,
+      "loss": 0.8389,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016089151388671508,
+      "loss": 0.902,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608388038498127,
+      "loss": 0.7672,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016078606696306625,
+      "loss": 0.815,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016073330324974994,
+      "loss": 0.835,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016068051273314987,
+      "loss": 0.8881,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016062769543656399,
+      "loss": 0.8577,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016057485138330194,
+      "loss": 0.8644,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016052198059668534,
+      "loss": 0.7636,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016046908310004748,
+      "loss": 0.842,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016041615891673354,
+      "loss": 0.908,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016036320807010036,
+      "loss": 0.8328,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016031023058351665,
+      "loss": 0.8266,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016025722648036283,
+      "loss": 0.8286,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602041957840311,
+      "loss": 0.8581,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016015113851792533,
+      "loss": 0.7969,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016009805470546112,
+      "loss": 0.7758,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016004494437006593,
+      "loss": 0.863,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015999180753517873,
+      "loss": 0.8843,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015993864422425032,
+      "loss": 0.8554,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001598854544607431,
+      "loss": 0.7666,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015983223826813124,
+      "loss": 0.8326,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597789956699004,
+      "loss": 0.801,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015972572668954818,
+      "loss": 0.7825,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001596724313505835,
+      "loss": 0.8966,
+      "step": 1549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015961910967652714,
+      "loss": 0.9143,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015956576169091143,
+      "loss": 0.8038,
+      "step": 1551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015951238741728022,
+      "loss": 0.795,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594589868791892,
+      "loss": 0.7875,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015940556010020536,
+      "loss": 0.7941,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593521071039075,
+      "loss": 0.7985,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001592986279138859,
+      "loss": 0.8035,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015924512255374233,
+      "loss": 0.8043,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015919159104709025,
+      "loss": 0.8109,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015913803341755458,
+      "loss": 0.7916,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590844496887718,
+      "loss": 0.8701,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015903083988438982,
+      "loss": 0.8397,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015897720402806818,
+      "loss": 0.88,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589235421434779,
+      "loss": 0.7956,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015886985425430137,
+      "loss": 0.8298,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015881614038423258,
+      "loss": 0.8668,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015876240055697695,
+      "loss": 0.8282,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015870863479625134,
+      "loss": 0.9103,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015865484312578405,
+      "loss": 0.8712,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015860102556931484,
+      "loss": 0.9009,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585471821505949,
+      "loss": 0.7967,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015849331289338685,
+      "loss": 0.7968,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001584394178214646,
+      "loss": 0.7936,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015838549695861357,
+      "loss": 0.8327,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015833155032863048,
+      "loss": 0.8525,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015827757795532356,
+      "loss": 0.8449,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015822357986251223,
+      "loss": 0.8513,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015816955607402742,
+      "loss": 0.7967,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015811550661371123,
+      "loss": 0.8655,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015806143150541723,
+      "loss": 0.8665,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015800733077301027,
+      "loss": 0.8017,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015795320444036645,
+      "loss": 0.8026,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015789905253137324,
+      "loss": 0.8608,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001578448750699294,
+      "loss": 0.7465,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015779067207994488,
+      "loss": 0.8456,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015773644358534102,
+      "loss": 0.8475,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576821896100503,
+      "loss": 0.8226,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015762791017801653,
+      "loss": 0.7812,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575736053131947,
+      "loss": 0.8847,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015751927503955106,
+      "loss": 0.7661,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015746491938106307,
+      "loss": 0.8994,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015741053836171935,
+      "loss": 0.8565,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015735613200551979,
+      "loss": 0.8165,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015730170033647535,
+      "loss": 0.7997,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724724337860832,
+      "loss": 0.8732,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000157192761155952,
+      "loss": 0.8195,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015713825369255088,
+      "loss": 0.8903,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015708372101246067,
+      "loss": 0.8639,
+      "step": 1597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015702916313974812,
+      "loss": 0.7237,
+      "step": 1598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001569745800984911,
+      "loss": 0.8771,
+      "step": 1599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015691997191277866,
+      "loss": 0.8173,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015686533860671085,
+      "loss": 0.8389,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568106802043989,
+      "loss": 0.8509,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015675599672996504,
+      "loss": 0.8075,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567012882075426,
+      "loss": 0.8823,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156646554661276,
+      "loss": 0.7767,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015659179611532062,
+      "loss": 0.7742,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015653701259384296,
+      "loss": 0.8466,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015648220412102048,
+      "loss": 0.8791,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015642737072104165,
+      "loss": 0.7702,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156372512418106,
+      "loss": 0.9042,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631762923642398,
+      "loss": 0.9021,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001562627212002171,
+      "loss": 0.7784,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015620778833371773,
+      "loss": 0.8425,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561528306611693,
+      "loss": 0.8433,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015609784820682615,
+      "loss": 0.8727,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015604284099495356,
+      "loss": 0.905,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001559878090498277,
+      "loss": 0.9026,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015593275239573573,
+      "loss": 0.7692,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015587767105697565,
+      "loss": 0.8401,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015582256505785634,
+      "loss": 0.7377,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015576743442269765,
+      "loss": 0.8312,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015571227917583026,
+      "loss": 0.803,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015565709934159565,
+      "loss": 0.8326,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015560189494434625,
+      "loss": 0.8383,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015554666600844526,
+      "loss": 0.7654,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015549141255826677,
+      "loss": 0.7724,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015543613461819559,
+      "loss": 0.8198,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015538083221262742,
+      "loss": 0.8018,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553255053659688,
+      "loss": 0.8834,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015527015410263695,
+      "loss": 0.799,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015521477844705986,
+      "loss": 0.8572,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015515937842367644,
+      "loss": 0.8101,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015510395405693615,
+      "loss": 0.7891,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550485053712994,
+      "loss": 0.781,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015499303239123715,
+      "loss": 0.7918,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015493753514123117,
+      "loss": 0.7738,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548820136457739,
+      "loss": 0.8601,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548264679293686,
+      "loss": 0.8081,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015477089801652904,
+      "loss": 0.8056,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015471530393177977,
+      "loss": 0.92,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015465968569965604,
+      "loss": 0.8348,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015460404334470372,
+      "loss": 0.8321,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015454837689147922,
+      "loss": 0.8681,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544926863645498,
+      "loss": 0.7957,
+      "step": 1644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544369717884931,
+      "loss": 0.8636,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015438123318789766,
+      "loss": 0.8406,
+      "step": 1646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015432547058736234,
+      "loss": 0.8312,
+      "step": 1647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015426968401149676,
+      "loss": 0.798,
+      "step": 1648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015421387348492105,
+      "loss": 0.8104,
+      "step": 1649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015415803903226594,
+      "loss": 0.8829,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541021806781728,
+      "loss": 0.7711,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015404629844729332,
+      "loss": 0.8075,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015399039236429,
+      "loss": 0.7987,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015393446245383564,
+      "loss": 0.8726,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015387850874061374,
+      "loss": 0.8693,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015382253124931818,
+      "loss": 0.867,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015376653000465334,
+      "loss": 0.8509,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015371050503133415,
+      "loss": 0.7524,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153654456354086,
+      "loss": 0.819,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015359838399764465,
+      "loss": 0.8739,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015354228798675646,
+      "loss": 0.8077,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015348616834617808,
+      "loss": 0.7872,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015343002510067672,
+      "loss": 0.902,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015337385827502993,
+      "loss": 0.7896,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015331766789402563,
+      "loss": 0.8542,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015326145398246221,
+      "loss": 0.8425,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015320521656514844,
+      "loss": 0.7693,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015314895566690345,
+      "loss": 0.7974,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015309267131255672,
+      "loss": 0.8629,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015303636352694805,
+      "loss": 0.8468,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529800323349277,
+      "loss": 0.7651,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529236777613561,
+      "loss": 0.8503,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015286729983110412,
+      "loss": 0.8058,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001528108985690529,
+      "loss": 0.7783,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015275447400009385,
+      "loss": 0.8671,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015269802614912873,
+      "loss": 0.8113,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015264155504106952,
+      "loss": 0.7865,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015258506070083846,
+      "loss": 0.8441,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015252854315336807,
+      "loss": 0.8458,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524720024236011,
+      "loss": 0.798,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015241543853649055,
+      "loss": 0.8479,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015235885151699963,
+      "loss": 0.8751,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015230224139010175,
+      "loss": 0.8152,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015224560818078052,
+      "loss": 0.77,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015218895191402968,
+      "loss": 0.8098,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521322726148533,
+      "loss": 0.8238,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015207557030826548,
+      "loss": 0.8184,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520188450192905,
+      "loss": 0.8614,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196209677296283,
+      "loss": 0.7991,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015190532559432698,
+      "loss": 0.7958,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015184853150843769,
+      "loss": 0.7988,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517917145403598,
+      "loss": 0.885,
+      "step": 1692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015173487471516808,
+      "loss": 0.8031,
+      "step": 1693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516780120579476,
+      "loss": 0.8392,
+      "step": 1694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015162112659379345,
+      "loss": 0.8536,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015156421834781067,
+      "loss": 0.9027,
+      "step": 1696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515072873451145,
+      "loss": 0.8311,
+      "step": 1697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015145033361083015,
+      "loss": 0.8378,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015139335717009286,
+      "loss": 0.823,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015133635804804793,
+      "loss": 0.9125,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015127933626985063,
+      "loss": 0.9236,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015122229186066623,
+      "loss": 0.8548,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015116522484567006,
+      "loss": 0.7527,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015110813525004733,
+      "loss": 0.7734,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510510230989932,
+      "loss": 0.916,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015099388841771297,
+      "loss": 0.8502,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015093673123142164,
+      "loss": 0.824,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015087955156534432,
+      "loss": 0.8348,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015082234944471595,
+      "loss": 0.8037,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001507651248947814,
+      "loss": 0.7697,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015070787794079543,
+      "loss": 0.7844,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506506086080227,
+      "loss": 0.8756,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015059331692173782,
+      "loss": 0.8075,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505360029072251,
+      "loss": 0.7798,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015047866658977888,
+      "loss": 0.8897,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001504213079947032,
+      "loss": 0.8412,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015036392714731203,
+      "loss": 0.7787,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015030652407292908,
+      "loss": 0.7453,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015024909879688794,
+      "loss": 0.7641,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015019165134453198,
+      "loss": 0.8889,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501341817412143,
+      "loss": 0.7635,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001500766900122979,
+      "loss": 0.8564,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001500191761831554,
+      "loss": 0.9034,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014996164027916925,
+      "loss": 0.8616,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014990408232573165,
+      "loss": 0.8028,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014984650234824452,
+      "loss": 0.7717,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497889003721195,
+      "loss": 0.7805,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497312764227779,
+      "loss": 0.873,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496736305256508,
+      "loss": 0.7784,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014961596270617888,
+      "loss": 0.8524,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014955827298981258,
+      "loss": 0.8486,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014950056140201193,
+      "loss": 0.8078,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014944282796824668,
+      "loss": 0.7966,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014938507271399618,
+      "loss": 0.7593,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493272956647494,
+      "loss": 0.8063,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014926949684600498,
+      "loss": 0.8574,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014921167628327112,
+      "loss": 0.804,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014915383400206558,
+      "loss": 0.8576,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909597002791584,
+      "loss": 0.886,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014903808438635887,
+      "loss": 0.8505,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014898017710294112,
+      "loss": 0.7813,
+      "step": 1741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014892224820321873,
+      "loss": 0.8817,
+      "step": 1742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014886429771275727,
+      "loss": 0.9291,
+      "step": 1743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014880632565713198,
+      "loss": 0.8305,
+      "step": 1744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487483320619275,
+      "loss": 0.8275,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014869031695273791,
+      "loss": 0.7402,
+      "step": 1746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014863228035516703,
+      "loss": 0.8143,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014857422229482792,
+      "loss": 0.8455,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014851614279734322,
+      "loss": 0.7809,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148458041888345,
+      "loss": 0.7473,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014839991959347484,
+      "loss": 0.7841,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014834177593838366,
+      "loss": 0.7478,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014828361094873193,
+      "loss": 0.8227,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014822542465018943,
+      "loss": 0.807,
+      "step": 1754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014816721706843532,
+      "loss": 0.8295,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001481089882291583,
+      "loss": 0.816,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014805073815805635,
+      "loss": 0.8051,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014799246688083684,
+      "loss": 0.818,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479341744232164,
+      "loss": 0.8196,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014787586081092124,
+      "loss": 0.8462,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014781752606968668,
+      "loss": 0.8364,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014775917022525747,
+      "loss": 0.8837,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014770079330338767,
+      "loss": 0.8432,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001476423953298406,
+      "loss": 0.7968,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001475839763303889,
+      "loss": 0.7169,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014752553633081453,
+      "loss": 0.8265,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014746707535690865,
+      "loss": 0.8573,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001474085934344717,
+      "loss": 0.8146,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014735009058931334,
+      "loss": 0.8092,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014729156684725254,
+      "loss": 0.7732,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472330222341174,
+      "loss": 0.8736,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014717445677574532,
+      "loss": 0.8173,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014711587049798282,
+      "loss": 0.7808,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014705726342668566,
+      "loss": 0.8235,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014699863558771876,
+      "loss": 0.8307,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014693998700695618,
+      "loss": 0.7739,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014688131771028117,
+      "loss": 0.8317,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014682262772358612,
+      "loss": 0.8487,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014676391707277254,
+      "loss": 0.8004,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014670518578375105,
+      "loss": 0.8135,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466464338824414,
+      "loss": 0.8857,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014658766139477244,
+      "loss": 0.8181,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014652886834668207,
+      "loss": 0.831,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014647005476411726,
+      "loss": 0.9115,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014641122067303412,
+      "loss": 0.9459,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014635236609939774,
+      "loss": 0.8134,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014629349106918223,
+      "loss": 0.8081,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014623459560837086,
+      "loss": 0.7996,
+      "step": 1788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014617567974295573,
+      "loss": 0.8503,
+      "step": 1789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014611674349893807,
+      "loss": 0.8399,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460577869023281,
+      "loss": 0.8345,
+      "step": 1791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014599880997914497,
+      "loss": 0.8054,
+      "step": 1792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459398127554168,
+      "loss": 0.8128,
+      "step": 1793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014588079525718075,
+      "loss": 0.799,
+      "step": 1794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458217575104828,
+      "loss": 0.7991,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014576269954137802,
+      "loss": 0.77,
+      "step": 1796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014570362137593027,
+      "loss": 0.7917,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014564452304021235,
+      "loss": 0.912,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014558540456030607,
+      "loss": 0.7866,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014552626596230198,
+      "loss": 0.8334,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014546710727229959,
+      "loss": 0.8385,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454079285164073,
+      "loss": 0.8606,
+      "step": 1802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014534872972074228,
+      "loss": 0.7952,
+      "step": 1803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014528951091143062,
+      "loss": 0.8539,
+      "step": 1804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001452302721146072,
+      "loss": 0.8398,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014517101335641573,
+      "loss": 0.7893,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014511173466300877,
+      "loss": 0.8136,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014505243606054765,
+      "loss": 0.778,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014499311757520242,
+      "loss": 0.7345,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014493377923315202,
+      "loss": 0.8215,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448744210605841,
+      "loss": 0.7739,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144815043083695,
+      "loss": 0.8467,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014475564532868992,
+      "loss": 0.8383,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014469622782178275,
+      "loss": 0.8748,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144636790589196,
+      "loss": 0.806,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014457733365716102,
+      "loss": 0.8395,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014451785705191776,
+      "loss": 0.8566,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444583607997149,
+      "loss": 0.847,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001443988449268098,
+      "loss": 0.8691,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014433930945946844,
+      "loss": 0.8734,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014427975442396547,
+      "loss": 0.8573,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014422017984658416,
+      "loss": 0.7444,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014416058575361643,
+      "loss": 0.782,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014410097217136281,
+      "loss": 0.8489,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014404133912613237,
+      "loss": 0.8002,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014398168664424288,
+      "loss": 0.8097,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014392201475202062,
+      "loss": 0.8378,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014386232347580037,
+      "loss": 0.8117,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014380261284192565,
+      "loss": 0.7335,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014374288287674828,
+      "loss": 0.8857,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014368313360662888,
+      "loss": 0.7923,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014362336505793638,
+      "loss": 0.7759,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001435635772570483,
+      "loss": 0.8117,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014350377023035068,
+      "loss": 0.9629,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014344394400423794,
+      "loss": 0.889,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433840986051131,
+      "loss": 0.804,
+      "step": 1836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014332423405938758,
+      "loss": 0.8146,
+      "step": 1837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014326435039348125,
+      "loss": 0.7675,
+      "step": 1838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014320444763382242,
+      "loss": 0.8496,
+      "step": 1839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014314452580684782,
+      "loss": 0.8192,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014308458493900264,
+      "loss": 0.7706,
+      "step": 1841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014302462505674045,
+      "loss": 0.8617,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014296464618652314,
+      "loss": 0.7926,
+      "step": 1843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014290464835482112,
+      "loss": 0.7433,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014284463158811304,
+      "loss": 0.7531,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014278459591288592,
+      "loss": 0.7925,
+      "step": 1846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014272454135563526,
+      "loss": 0.8536,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014266446794286472,
+      "loss": 0.8067,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014260437570108638,
+      "loss": 0.8719,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001425442646568206,
+      "loss": 0.8328,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014248413483659603,
+      "loss": 0.9216,
+      "step": 1851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014242398626694965,
+      "loss": 0.7343,
+      "step": 1852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423638189744267,
+      "loss": 0.8146,
+      "step": 1853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423036329855806,
+      "loss": 0.8422,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014224342832697316,
+      "loss": 0.7749,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001421832050251743,
+      "loss": 0.8417,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014212296310676226,
+      "loss": 0.8031,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014206270259832344,
+      "loss": 0.8644,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420024235264525,
+      "loss": 0.8759,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014194212591775224,
+      "loss": 0.8764,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014188180979883363,
+      "loss": 0.8429,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014182147519631587,
+      "loss": 0.9202,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014176112213682627,
+      "loss": 0.8297,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014170075064700037,
+      "loss": 0.8026,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001416403607534817,
+      "loss": 0.8098,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014157995248292198,
+      "loss": 0.8093,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014151952586198113,
+      "loss": 0.8276,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141459080917327,
+      "loss": 0.8188,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014139861767563565,
+      "loss": 0.7905,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014133813616359117,
+      "loss": 0.8409,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014127763640788574,
+      "loss": 0.7761,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014121711843521954,
+      "loss": 0.825,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014115658227230085,
+      "loss": 0.7785,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014109602794584592,
+      "loss": 0.8457,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410354554825791,
+      "loss": 0.8042,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014097486490923265,
+      "loss": 0.8128,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014091425625254686,
+      "loss": 0.86,
+      "step": 1877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014085362953927003,
+      "loss": 0.8578,
+      "step": 1878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014079298479615837,
+      "loss": 0.8079,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014073232204997612,
+      "loss": 0.8471,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014067164132749538,
+      "loss": 0.8817,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014061094265549625,
+      "loss": 0.8079,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014055022606076673,
+      "loss": 0.8181,
+      "step": 1883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014048949157010273,
+      "loss": 0.8269,
+      "step": 1884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014042873921030804,
+      "loss": 0.772,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001403679690081944,
+      "loss": 0.85,
+      "step": 1886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014030718099058135,
+      "loss": 0.826,
+      "step": 1887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402463751842963,
+      "loss": 0.8149,
+      "step": 1888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014018555161617459,
+      "loss": 0.8156,
+      "step": 1889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014012471031305928,
+      "loss": 0.8689,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014006385130180138,
+      "loss": 0.826,
+      "step": 1891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014000297460925955,
+      "loss": 0.7873,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013994208026230044,
+      "loss": 0.7854,
+      "step": 1893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398811682877984,
+      "loss": 0.8035,
+      "step": 1894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398202387126355,
+      "loss": 0.7827,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013975929156370175,
+      "loss": 0.9185,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013969832686789473,
+      "loss": 0.8604,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013963734465211985,
+      "loss": 0.7801,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395763449432903,
+      "loss": 0.8313,
+      "step": 1899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395153277683269,
+      "loss": 0.8774,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013945429315415819,
+      "loss": 0.8106,
+      "step": 1901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013939324112772046,
+      "loss": 0.7142,
+      "step": 1902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013933217171595766,
+      "loss": 0.6898,
+      "step": 1903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013927108494582138,
+      "loss": 0.78,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392099808442709,
+      "loss": 0.8768,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013914885943827317,
+      "loss": 0.8261,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390877207548028,
+      "loss": 0.7509,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390265648208419,
+      "loss": 0.8654,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013896539166338028,
+      "loss": 0.7166,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013890420130941537,
+      "loss": 0.8445,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013884299378595215,
+      "loss": 0.8305,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001387817691200032,
+      "loss": 0.7673,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013872052733858868,
+      "loss": 0.7974,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013865926846873622,
+      "loss": 0.7299,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013859799253748104,
+      "loss": 0.8035,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013853669957186602,
+      "loss": 0.7982,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013847538959894132,
+      "loss": 0.823,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013841406264576475,
+      "loss": 0.8435,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013835271873940163,
+      "loss": 0.7836,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013829135790692466,
+      "loss": 0.8134,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013822998017541416,
+      "loss": 0.7877,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381685855719577,
+      "loss": 0.837,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381071741236505,
+      "loss": 0.8008,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013804574585759508,
+      "loss": 0.7847,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013798430080090148,
+      "loss": 0.8074,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013792283898068705,
+      "loss": 0.8329,
+      "step": 1926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001378613604240766,
+      "loss": 0.7745,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013779986515820235,
+      "loss": 0.7678,
+      "step": 1928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013773835321020383,
+      "loss": 0.7241,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013767682460722795,
+      "loss": 0.7872,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013761527937642903,
+      "loss": 0.8601,
+      "step": 1931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375537175449686,
+      "loss": 0.8264,
+      "step": 1932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013749213914001563,
+      "loss": 0.8572,
+      "step": 1933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013743054418874644,
+      "loss": 0.7941,
+      "step": 1934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013736893271834447,
+      "loss": 0.7487,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013730730475600066,
+      "loss": 0.7516,
+      "step": 1936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013724566032891303,
+      "loss": 0.8625,
+      "step": 1937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013718399946428708,
+      "loss": 0.8373,
+      "step": 1938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001371223221893354,
+      "loss": 0.8435,
+      "step": 1939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013706062853127785,
+      "loss": 0.841,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013699891851734158,
+      "loss": 0.7544,
+      "step": 1941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013693719217476092,
+      "loss": 0.7704,
+      "step": 1942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368754495307774,
+      "loss": 0.8715,
+      "step": 1943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013681369061263973,
+      "loss": 0.7894,
+      "step": 1944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013675191544760388,
+      "loss": 0.8696,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001366901240629329,
+      "loss": 0.8277,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013662831648589705,
+      "loss": 0.8602,
+      "step": 1947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001365664927437737,
+      "loss": 0.8066,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001365046528638474,
+      "loss": 0.7998,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013644279687340975,
+      "loss": 0.8658,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013638092479975958,
+      "loss": 0.873,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013631903667020266,
+      "loss": 0.737,
+      "step": 1952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000136257132512052,
+      "loss": 0.8431,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361952123526276,
+      "loss": 0.7612,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361332762192565,
+      "loss": 0.9143,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013607132413927288,
+      "loss": 0.7662,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013600935614001787,
+      "loss": 0.8923,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013594737224883964,
+      "loss": 0.7922,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013588537249309342,
+      "loss": 0.8053,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013582335690014144,
+      "loss": 0.8611,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013576132549735284,
+      "loss": 0.8531,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356992783121038,
+      "loss": 0.8607,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356372153717775,
+      "loss": 0.8807,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000135575136703764,
+      "loss": 0.7812,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001355130423354603,
+      "loss": 0.8314,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013545093229427047,
+      "loss": 0.8679,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013538880660760524,
+      "loss": 0.8193,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013532666530288246,
+      "loss": 0.861,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001352645084075268,
+      "loss": 0.8578,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013520233594896985,
+      "loss": 0.8963,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013514014795464998,
+      "loss": 0.8103,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013507794445201247,
+      "loss": 0.7701,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013501572546850948,
+      "loss": 0.8162,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013495349103159995,
+      "loss": 0.8875,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013489124116874965,
+      "loss": 0.7938,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013482897590743115,
+      "loss": 0.8124,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013476669527512385,
+      "loss": 0.8841,
+      "step": 1977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013470439929931394,
+      "loss": 0.819,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346420880074943,
+      "loss": 0.8334,
+      "step": 1979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345797614271646,
+      "loss": 0.8566,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013451741958583137,
+      "loss": 0.8218,
+      "step": 1981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013445506251100775,
+      "loss": 0.8146,
+      "step": 1982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343926902302136,
+      "loss": 0.8468,
+      "step": 1983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013433030277097563,
+      "loss": 0.8354,
+      "step": 1984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013426790016082702,
+      "loss": 0.8746,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013420548242730783,
+      "loss": 0.7732,
+      "step": 1986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013414304959796477,
+      "loss": 0.8184,
+      "step": 1987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013408060170035107,
+      "loss": 0.8167,
+      "step": 1988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340181387620268,
+      "loss": 0.8279,
+      "step": 1989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013395566081055848,
+      "loss": 0.8683,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013389316787351947,
+      "loss": 0.7844,
+      "step": 1991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001338306599784896,
+      "loss": 0.7725,
+      "step": 1992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001337681371530553,
+      "loss": 0.8135,
+      "step": 1993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013370559942480962,
+      "loss": 0.7626,
+      "step": 1994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336430468213522,
+      "loss": 0.7948,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013358047937028924,
+      "loss": 0.7968,
+      "step": 1996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013351789709923346,
+      "loss": 0.7878,
+      "step": 1997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013345530003580413,
+      "loss": 0.8068,
+      "step": 1998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013339268820762712,
+      "loss": 0.7649,
+      "step": 1999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013333006164233474,
+      "loss": 0.7844,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013326742036756577,
+      "loss": 0.7944,
+      "step": 2001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013320476441096562,
+      "loss": 0.8204,
+      "step": 2002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133142093800186,
+      "loss": 0.8074,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013307940856288524,
+      "loss": 0.7457,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001330167087267281,
+      "loss": 0.7845,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013295399431938564,
+      "loss": 0.7996,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013289126536853555,
+      "loss": 0.8353,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013282852190186182,
+      "loss": 0.8689,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013276576394705491,
+      "loss": 0.7987,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013270299153181155,
+      "loss": 0.8127,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013264020468383504,
+      "loss": 0.7187,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013257740343083487,
+      "loss": 0.8207,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132514587800527,
+      "loss": 0.8754,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013245175782063373,
+      "loss": 0.8594,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013238891351888363,
+      "loss": 0.799,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013232605492301163,
+      "loss": 0.742,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132263182060759,
+      "loss": 0.9198,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322002949598732,
+      "loss": 0.858,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321373936481081,
+      "loss": 0.7903,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013207447815322376,
+      "loss": 0.7921,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013201154850298651,
+      "loss": 0.8852,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000131948604725169,
+      "loss": 0.8421,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013188564684755004,
+      "loss": 0.8389,
+      "step": 2023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013182267489791463,
+      "loss": 0.8708,
+      "step": 2024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001317596889040541,
+      "loss": 0.8273,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316966888937658,
+      "loss": 0.8527,
+      "step": 2026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316336748948535,
+      "loss": 0.7922,
+      "step": 2027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013157064693512699,
+      "loss": 0.8234,
+      "step": 2028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013150760504240217,
+      "loss": 0.8018,
+      "step": 2029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013144454924450124,
+      "loss": 0.7841,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013138147956925244,
+      "loss": 0.8253,
+      "step": 2031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013131839604449018,
+      "loss": 0.74,
+      "step": 2032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013125529869805494,
+      "loss": 0.7791,
+      "step": 2033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013119218755779327,
+      "loss": 0.8083,
+      "step": 2034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000131129062651558,
+      "loss": 0.8265,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013106592400720778,
+      "loss": 0.89,
+      "step": 2036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013100277165260747,
+      "loss": 0.8132,
+      "step": 2037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013093960561562793,
+      "loss": 0.8279,
+      "step": 2038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013087642592414612,
+      "loss": 0.7466,
+      "step": 2039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013081323260604494,
+      "loss": 0.8562,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013075002568921338,
+      "loss": 0.7818,
+      "step": 2041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013068680520154636,
+      "loss": 0.8408,
+      "step": 2042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013062357117094484,
+      "loss": 0.7571,
+      "step": 2043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013056032362531575,
+      "loss": 0.8168,
+      "step": 2044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013049706259257198,
+      "loss": 0.8746,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001304337881006324,
+      "loss": 0.841,
+      "step": 2046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013037050017742168,
+      "loss": 0.7726,
+      "step": 2047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303071988508707,
+      "loss": 0.8145,
+      "step": 2048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013024388414891594,
+      "loss": 0.7847,
+      "step": 2049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013018055609949998,
+      "loss": 0.8683,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011721473057123,
+      "loss": 0.9093,
+      "step": 2051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000130053860070084,
+      "loss": 0.8158,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012999049214599843,
+      "loss": 0.8134,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012992711098628055,
+      "loss": 0.8131,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298637166189022,
+      "loss": 0.7751,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012980030907184107,
+      "loss": 0.7641,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001297368883730807,
+      "loss": 0.7945,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012967345455061033,
+      "loss": 0.8194,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001296100076324251,
+      "loss": 0.8485,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012954654764652592,
+      "loss": 0.8424,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012948307462091934,
+      "loss": 0.7485,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012941958858361783,
+      "loss": 0.8505,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293560895626395,
+      "loss": 0.8076,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012929257758600825,
+      "loss": 0.8013,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012922905268175362,
+      "loss": 0.8359,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012916551487791094,
+      "loss": 0.782,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001291019642025212,
+      "loss": 0.8493,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012903840068363107,
+      "loss": 0.8248,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012897482434929284,
+      "loss": 0.7849,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012891123522756459,
+      "loss": 0.7556,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001288476333465099,
+      "loss": 0.8442,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000128784018734198,
+      "loss": 0.7491,
+      "step": 2072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012872039141870385,
+      "loss": 0.8668,
+      "step": 2073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012865675142810788,
+      "loss": 0.7693,
+      "step": 2074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012859309879049624,
+      "loss": 0.7921,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012852943353396055,
+      "loss": 0.8211,
+      "step": 2076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000128465755686598,
+      "loss": 0.8881,
+      "step": 2077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012840206527651147,
+      "loss": 0.8098,
+      "step": 2078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012833836233180928,
+      "loss": 0.7997,
+      "step": 2079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001282746468806053,
+      "loss": 0.8528,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012821091895101887,
+      "loss": 0.8736,
+      "step": 2081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012814717857117494,
+      "loss": 0.8571,
+      "step": 2082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012808342576920383,
+      "loss": 0.7787,
+      "step": 2083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012801966057324147,
+      "loss": 0.8749,
+      "step": 2084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001279558830114292,
+      "loss": 0.7625,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012789209311191382,
+      "loss": 0.7988,
+      "step": 2086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012782829090284752,
+      "loss": 0.8212,
+      "step": 2087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012776447641238805,
+      "loss": 0.8358,
+      "step": 2088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012770064966869846,
+      "loss": 0.8187,
+      "step": 2089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001276368106999473,
+      "loss": 0.8611,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012757295953430838,
+      "loss": 0.776,
+      "step": 2091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275090961999611,
+      "loss": 0.8727,
+      "step": 2092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012744522072509002,
+      "loss": 0.829,
+      "step": 2093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001273813331378852,
+      "loss": 0.7859,
+      "step": 2094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012731743346654199,
+      "loss": 0.8077,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012725352173926106,
+      "loss": 0.7772,
+      "step": 2096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012718959798424843,
+      "loss": 0.8515,
+      "step": 2097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012712566222971541,
+      "loss": 0.8312,
+      "step": 2098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012706171450387865,
+      "loss": 0.8696,
+      "step": 2099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012699775483496,
+      "loss": 0.7903,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012693378325118667,
+      "loss": 0.8201,
+      "step": 2101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012686979978079107,
+      "loss": 0.8067,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012680580445201083,
+      "loss": 0.8164,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001267417972930889,
+      "loss": 0.8726,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012667777833227342,
+      "loss": 0.8257,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012661374759781767,
+      "loss": 0.7872,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012654970511798025,
+      "loss": 0.7803,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264856509210248,
+      "loss": 0.7904,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264215850352203,
+      "loss": 0.7344,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263575074888407,
+      "loss": 0.7934,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012629341831016527,
+      "loss": 0.8445,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012622931752747827,
+      "loss": 0.8083,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012616520516906918,
+      "loss": 0.8524,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012610108126323254,
+      "loss": 0.8245,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012603694583826803,
+      "loss": 0.8806,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012597279892248033,
+      "loss": 0.7739,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259086405441793,
+      "loss": 0.8166,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012584447073167975,
+      "loss": 0.8303,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257802895133016,
+      "loss": 0.8308,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012571609691736985,
+      "loss": 0.8086,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012565189297221444,
+      "loss": 0.8881,
+      "step": 2121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255876777061703,
+      "loss": 0.8437,
+      "step": 2122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012552345114757744,
+      "loss": 0.7814,
+      "step": 2123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012545921332478082,
+      "loss": 0.8374,
+      "step": 2124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012539496426613035,
+      "loss": 0.8947,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253307039999809,
+      "loss": 0.8278,
+      "step": 2126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012526643255469233,
+      "loss": 0.8514,
+      "step": 2127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012520214995862937,
+      "loss": 0.7877,
+      "step": 2128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012513785624016174,
+      "loss": 0.7721,
+      "step": 2129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012507355142766402,
+      "loss": 0.8224,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012500923554951564,
+      "loss": 0.828,
+      "step": 2131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012494490863410108,
+      "loss": 0.8224,
+      "step": 2132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012488057070980955,
+      "loss": 0.8202,
+      "step": 2133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248162218050351,
+      "loss": 0.8629,
+      "step": 2134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012475186194817675,
+      "loss": 0.7916,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012468749116763823,
+      "loss": 0.8845,
+      "step": 2136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012462310949182815,
+      "loss": 0.8333,
+      "step": 2137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012455871694915993,
+      "loss": 0.7953,
+      "step": 2138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012449431356805176,
+      "loss": 0.7526,
+      "step": 2139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012442989937692665,
+      "loss": 0.8089,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012436547440421237,
+      "loss": 0.8511,
+      "step": 2141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012430103867834143,
+      "loss": 0.7903,
+      "step": 2142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012423659222775111,
+      "loss": 0.781,
+      "step": 2143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012417213508088336,
+      "loss": 0.8571,
+      "step": 2144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012410766726618494,
+      "loss": 0.8354,
+      "step": 2145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012404318881210728,
+      "loss": 0.7613,
+      "step": 2146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239786997471065,
+      "loss": 0.7534,
+      "step": 2147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012391420009964335,
+      "loss": 0.7822,
+      "step": 2148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238496898981834,
+      "loss": 0.7625,
+      "step": 2149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001237851691711967,
+      "loss": 0.7539,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012372063794715806,
+      "loss": 0.8144,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001236560962545469,
+      "loss": 0.8461,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012359154412184723,
+      "loss": 0.763,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012352698157754773,
+      "loss": 0.8133,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012346240865014163,
+      "loss": 0.8002,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012339782536812668,
+      "loss": 0.7782,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001233332317600053,
+      "loss": 0.7864,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012326862785428446,
+      "loss": 0.8746,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012320401367947562,
+      "loss": 0.8533,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012313938926409484,
+      "loss": 0.7618,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012307475463666257,
+      "loss": 0.7852,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012301010982570395,
+      "loss": 0.8331,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012294545485974845,
+      "loss": 0.7369,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012288078976733016,
+      "loss": 0.8506,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012281611457698753,
+      "loss": 0.8797,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012275142931726348,
+      "loss": 0.859,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012268673401670546,
+      "loss": 0.8258,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012262202870386529,
+      "loss": 0.8036,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012255731340729914,
+      "loss": 0.8403,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012249258815556774,
+      "loss": 0.7883,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012242785297723608,
+      "loss": 0.7692,
+      "step": 2171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012236310790087363,
+      "loss": 0.7731,
+      "step": 2172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222983529550541,
+      "loss": 0.8427,
+      "step": 2173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222335881683557,
+      "loss": 0.8305,
+      "step": 2174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012216881356936091,
+      "loss": 0.8246,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012210402918665653,
+      "loss": 0.8463,
+      "step": 2176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012203923504883372,
+      "loss": 0.7621,
+      "step": 2177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219744311844879,
+      "loss": 0.8431,
+      "step": 2178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012190961762221882,
+      "loss": 0.8154,
+      "step": 2179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001218447943906305,
+      "loss": 0.8562,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001217799615183312,
+      "loss": 0.8058,
+      "step": 2181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012171511903393346,
+      "loss": 0.8703,
+      "step": 2182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012165026696605411,
+      "loss": 0.8292,
+      "step": 2183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012158540534331404,
+      "loss": 0.8266,
+      "step": 2184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001215205341943386,
+      "loss": 0.8892,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012145565354775717,
+      "loss": 0.7066,
+      "step": 2186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012139076343220337,
+      "loss": 0.8206,
+      "step": 2187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012132586387631501,
+      "loss": 0.8328,
+      "step": 2188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012126095490873404,
+      "loss": 0.7861,
+      "step": 2189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012119603655810658,
+      "loss": 0.8925,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001211311088530829,
+      "loss": 0.7843,
+      "step": 2191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012106617182231741,
+      "loss": 0.8351,
+      "step": 2192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100122549446857,
+      "loss": 0.85,
+      "step": 2193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000120936269898199,
+      "loss": 0.8258,
+      "step": 2194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001208713050621754,
+      "loss": 0.8074,
+      "step": 2195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012080633101506858,
+      "loss": 0.7891,
+      "step": 2196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012074134778555336,
+      "loss": 0.7853,
+      "step": 2197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012067635540230861,
+      "loss": 0.7101,
+      "step": 2198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012061135389401728,
+      "loss": 0.803,
+      "step": 2199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012054634328936634,
+      "loss": 0.799,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012048132361704676,
+      "loss": 0.7801,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001204162949057535,
+      "loss": 0.8002,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012035125718418556,
+      "loss": 0.8654,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012028621048104588,
+      "loss": 0.8176,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012022115482504131,
+      "loss": 0.7436,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012015609024488273,
+      "loss": 0.8676,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012009101676928498,
+      "loss": 0.8577,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012002593442696674,
+      "loss": 0.7603,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011996084324665068,
+      "loss": 0.7906,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011989574325706332,
+      "loss": 0.8179,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011983063448693507,
+      "loss": 0.772,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011976551696500023,
+      "loss": 0.8185,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011970039071999697,
+      "loss": 0.7811,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001196352557806673,
+      "loss": 0.8229,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011957011217575704,
+      "loss": 0.8327,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011950495993401587,
+      "loss": 0.7743,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011943979908419731,
+      "loss": 0.7838,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011937462965505863,
+      "loss": 0.7895,
+      "step": 2218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001193094516753609,
+      "loss": 0.7883,
+      "step": 2219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011924426517386893,
+      "loss": 0.8015,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011917907017935135,
+      "loss": 0.808,
+      "step": 2221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011911386672058049,
+      "loss": 0.8509,
+      "step": 2222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011904865482633247,
+      "loss": 0.8315,
+      "step": 2223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011898343452538709,
+      "loss": 0.8486,
+      "step": 2224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011891820584652784,
+      "loss": 0.7659,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011885296881854197,
+      "loss": 0.8,
+      "step": 2226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878772347022038,
+      "loss": 0.7861,
+      "step": 2227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001187224698303576,
+      "loss": 0.8394,
+      "step": 2228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011865720792775195,
+      "loss": 0.8268,
+      "step": 2229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011859193779120524,
+      "loss": 0.7941,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011852665944952297,
+      "loss": 0.7473,
+      "step": 2231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011846137293151434,
+      "loss": 0.6717,
+      "step": 2232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011839607826599204,
+      "loss": 0.7639,
+      "step": 2233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011833077548177243,
+      "loss": 0.8208,
+      "step": 2234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011826546460767542,
+      "loss": 0.7589,
+      "step": 2235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011820014567252453,
+      "loss": 0.8316,
+      "step": 2236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001181348187051468,
+      "loss": 0.7667,
+      "step": 2237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011806948373437279,
+      "loss": 0.783,
+      "step": 2238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011800414078903666,
+      "loss": 0.8005,
+      "step": 2239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011793878989797604,
+      "loss": 0.781,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011787343109003211,
+      "loss": 0.8294,
+      "step": 2241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001178080643940495,
+      "loss": 0.8539,
+      "step": 2242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011774268983887633,
+      "loss": 0.8273,
+      "step": 2243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001176773074533642,
+      "loss": 0.7776,
+      "step": 2244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011761191726636818,
+      "loss": 0.8889,
+      "step": 2245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011754651930674674,
+      "loss": 0.8448,
+      "step": 2246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011748111360336181,
+      "loss": 0.8498,
+      "step": 2247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011741570018507873,
+      "loss": 0.9065,
+      "step": 2248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011735027908076627,
+      "loss": 0.768,
+      "step": 2249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011728485031929657,
+      "loss": 0.828,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011721941392954512,
+      "loss": 0.8585,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001171539699403908,
+      "loss": 0.8114,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001170885183807159,
+      "loss": 0.814,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011702305927940595,
+      "loss": 0.7671,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011695759266534991,
+      "loss": 0.8182,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011689211856743996,
+      "loss": 0.7914,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011682663701457163,
+      "loss": 0.863,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011676114803564375,
+      "loss": 0.8605,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011669565165955842,
+      "loss": 0.7689,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011663014791522098,
+      "loss": 0.789,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011656463683154006,
+      "loss": 0.7747,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001164991184374275,
+      "loss": 0.8569,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011643359276179841,
+      "loss": 0.7639,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011636805983357103,
+      "loss": 0.8281,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011630251968166689,
+      "loss": 0.8212,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011623697233501063,
+      "loss": 0.7352,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011617141782253016,
+      "loss": 0.7955,
+      "step": 2267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011610585617315643,
+      "loss": 0.8039,
+      "step": 2268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011604028741582362,
+      "loss": 0.7772,
+      "step": 2269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011597471157946909,
+      "loss": 0.7907,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001159091286930332,
+      "loss": 0.8453,
+      "step": 2271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011584353878545952,
+      "loss": 0.8155,
+      "step": 2272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011577794188569471,
+      "loss": 0.7933,
+      "step": 2273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011571233802268843,
+      "loss": 0.8087,
+      "step": 2274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011564672722539354,
+      "loss": 0.8538,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011558110952276586,
+      "loss": 0.8035,
+      "step": 2276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001155154849437643,
+      "loss": 0.7902,
+      "step": 2277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011544985351735083,
+      "loss": 0.7664,
+      "step": 2278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011538421527249035,
+      "loss": 0.8209,
+      "step": 2279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011531857023815087,
+      "loss": 0.8021,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011525291844330337,
+      "loss": 0.8193,
+      "step": 2281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011518725991692175,
+      "loss": 0.7351,
+      "step": 2282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011512159468798298,
+      "loss": 0.7719,
+      "step": 2283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011505592278546691,
+      "loss": 0.8365,
+      "step": 2284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011499024423835638,
+      "loss": 0.7599,
+      "step": 2285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011492455907563714,
+      "loss": 0.8503,
+      "step": 2286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011485886732629789,
+      "loss": 0.81,
+      "step": 2287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011479316901933017,
+      "loss": 0.8451,
+      "step": 2288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011472746418372846,
+      "loss": 0.7463,
+      "step": 2289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011466175284849015,
+      "loss": 0.8518,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011459603504261543,
+      "loss": 0.7721,
+      "step": 2291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011453031079510741,
+      "loss": 0.8,
+      "step": 2292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011446458013497204,
+      "loss": 0.7821,
+      "step": 2293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011439884309121803,
+      "loss": 0.7783,
+      "step": 2294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011433309969285695,
+      "loss": 0.834,
+      "step": 2295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001142673499689032,
+      "loss": 0.6994,
+      "step": 2296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011420159394837395,
+      "loss": 0.7957,
+      "step": 2297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011413583166028914,
+      "loss": 0.7399,
+      "step": 2298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011407006313367148,
+      "loss": 0.8239,
+      "step": 2299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011400428839754642,
+      "loss": 0.806,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011393850748094222,
+      "loss": 0.7744,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011387272041288977,
+      "loss": 0.7872,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001138069272224227,
+      "loss": 0.7949,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011374112793857742,
+      "loss": 0.8232,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113675322590393,
+      "loss": 0.7941,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011360951120691103,
+      "loss": 0.8508,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113543693817176,
+      "loss": 0.8268,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011347787045023488,
+      "loss": 0.76,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341204113513739,
+      "loss": 0.8382,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011334620590093578,
+      "loss": 0.7704,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011328036477668497,
+      "loss": 0.8458,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001132145177914425,
+      "loss": 0.8619,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001131486649742684,
+      "loss": 0.7426,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011308280635422537,
+      "loss": 0.8408,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011301694196037867,
+      "loss": 0.7848,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011295107182179601,
+      "loss": 0.8729,
+      "step": 2316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011288519596754776,
+      "loss": 0.8374,
+      "step": 2317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011281931442670672,
+      "loss": 0.8348,
+      "step": 2318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011275342722834825,
+      "loss": 0.8596,
+      "step": 2319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011268753440155018,
+      "loss": 0.7778,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011262163597539284,
+      "loss": 0.8262,
+      "step": 2321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011255573197895901,
+      "loss": 0.7711,
+      "step": 2322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011248982244133395,
+      "loss": 0.7876,
+      "step": 2323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011242390739160538,
+      "loss": 0.7477,
+      "step": 2324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011235798685886341,
+      "loss": 0.783,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011229206087220059,
+      "loss": 0.7694,
+      "step": 2326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011222612946071189,
+      "loss": 0.764,
+      "step": 2327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011216019265349465,
+      "loss": 0.7699,
+      "step": 2328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001120942504796486,
+      "loss": 0.7765,
+      "step": 2329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011202830296827589,
+      "loss": 0.761,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011196235014848091,
+      "loss": 0.792,
+      "step": 2331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011189639204937051,
+      "loss": 0.7594,
+      "step": 2332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001118304287000538,
+      "loss": 0.7874,
+      "step": 2333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011176446012964222,
+      "loss": 0.8226,
+      "step": 2334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011169848636724951,
+      "loss": 0.7999,
+      "step": 2335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011163250744199176,
+      "loss": 0.7869,
+      "step": 2336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011156652338298725,
+      "loss": 0.8376,
+      "step": 2337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001115005342193566,
+      "loss": 0.8365,
+      "step": 2338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011143453998022264,
+      "loss": 0.836,
+      "step": 2339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011136854069471044,
+      "loss": 0.7707,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113025363919473,
+      "loss": 0.8171,
+      "step": 2341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011123652710106278,
+      "loss": 0.8566,
+      "step": 2342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011117051285118854,
+      "loss": 0.8603,
+      "step": 2343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011110449367145854,
+      "loss": 0.7778,
+      "step": 2344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011103846959100884,
+      "loss": 0.8279,
+      "step": 2345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011097244063897769,
+      "loss": 0.7771,
+      "step": 2346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011090640684450553,
+      "loss": 0.8196,
+      "step": 2347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011084036823673488,
+      "loss": 0.7783,
+      "step": 2348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011077432484481036,
+      "loss": 0.8081,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011070827669787877,
+      "loss": 0.8387,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011064222382508903,
+      "loss": 0.7709,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011057616625559202,
+      "loss": 0.8061,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011051010401854078,
+      "loss": 0.7608,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011044403714309046,
+      "loss": 0.8713,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011037796565839812,
+      "loss": 0.7864,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011031188959362299,
+      "loss": 0.7995,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011024580897792622,
+      "loss": 0.7946,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011017972384047105,
+      "loss": 0.7985,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011011363421042264,
+      "loss": 0.866,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001100475401169482,
+      "loss": 0.7653,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010998144158921686,
+      "loss": 0.832,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010991533865639974,
+      "loss": 0.7601,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010984923134766985,
+      "loss": 0.7786,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010978311969220221,
+      "loss": 0.7645,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010971700371917371,
+      "loss": 0.9218,
+      "step": 2365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001096508834577631,
+      "loss": 0.8013,
+      "step": 2366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958475893715119,
+      "loss": 0.7451,
+      "step": 2367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010951863018652047,
+      "loss": 0.7972,
+      "step": 2368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010945249723505541,
+      "loss": 0.7687,
+      "step": 2369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010938636011194233,
+      "loss": 0.7344,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010932021884636933,
+      "loss": 0.8513,
+      "step": 2371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010925407346752638,
+      "loss": 0.7393,
+      "step": 2372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010918792400460532,
+      "loss": 0.792,
+      "step": 2373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010912177048679963,
+      "loss": 0.7735,
+      "step": 2374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001090556129433048,
+      "loss": 0.7675,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010898945140331789,
+      "loss": 0.8312,
+      "step": 2376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010892328589603784,
+      "loss": 0.7665,
+      "step": 2377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010885711645066535,
+      "loss": 0.7634,
+      "step": 2378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010879094309640276,
+      "loss": 0.7966,
+      "step": 2379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010872476586245428,
+      "loss": 0.7778,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001086585847780257,
+      "loss": 0.784,
+      "step": 2381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010859239987232456,
+      "loss": 0.8141,
+      "step": 2382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010852621117456008,
+      "loss": 0.7717,
+      "step": 2383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001084600187139432,
+      "loss": 0.7999,
+      "step": 2384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010839382251968641,
+      "loss": 0.8091,
+      "step": 2385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000108327622621004,
+      "loss": 0.8564,
+      "step": 2386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010826141904711177,
+      "loss": 0.7911,
+      "step": 2387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010819521182722716,
+      "loss": 0.8259,
+      "step": 2388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001081290009905693,
+      "loss": 0.7935,
+      "step": 2389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010806278656635884,
+      "loss": 0.8667,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010799656858381804,
+      "loss": 0.7706,
+      "step": 2391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010793034707217069,
+      "loss": 0.7622,
+      "step": 2392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001078641220606422,
+      "loss": 0.7953,
+      "step": 2393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001077978935784595,
+      "loss": 0.8169,
+      "step": 2394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010773166165485103,
+      "loss": 0.7528,
+      "step": 2395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010766542631904679,
+      "loss": 0.8008,
+      "step": 2396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010759918760027823,
+      "loss": 0.7046,
+      "step": 2397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753294552777835,
+      "loss": 0.7163,
+      "step": 2398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010746670013078161,
+      "loss": 0.8254,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010740045143852393,
+      "loss": 0.8335,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010733419948024265,
+      "loss": 0.7716,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010726794428517666,
+      "loss": 0.7337,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010720168588256613,
+      "loss": 0.7735,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010713542430165277,
+      "loss": 0.8033,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010706915957167964,
+      "loss": 0.7866,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001070028917218912,
+      "loss": 0.8359,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010693662078153325,
+      "loss": 0.8582,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010687034677985301,
+      "loss": 0.7218,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010680406974609905,
+      "loss": 0.8547,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010673778970952117,
+      "loss": 0.7638,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010667150669937069,
+      "loss": 0.8514,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010660522074490007,
+      "loss": 0.8325,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010653893187536315,
+      "loss": 0.8032,
+      "step": 2413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010647264012001504,
+      "loss": 0.7928,
+      "step": 2414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010640634550811208,
+      "loss": 0.8223,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010634004806891198,
+      "loss": 0.8323,
+      "step": 2416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010627374783167358,
+      "loss": 0.8446,
+      "step": 2417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010620744482565706,
+      "loss": 0.795,
+      "step": 2418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001061411390801237,
+      "loss": 0.8131,
+      "step": 2419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010607483062433608,
+      "loss": 0.8141,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010600851948755797,
+      "loss": 0.902,
+      "step": 2421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001059422056990543,
+      "loss": 0.8086,
+      "step": 2422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010587588928809118,
+      "loss": 0.78,
+      "step": 2423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010580957028393585,
+      "loss": 0.7928,
+      "step": 2424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010574324871585674,
+      "loss": 0.7974,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010567692461312339,
+      "loss": 0.7958,
+      "step": 2426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010561059800500644,
+      "loss": 0.8216,
+      "step": 2427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010554426892077767,
+      "loss": 0.77,
+      "step": 2428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010547793738970993,
+      "loss": 0.8134,
+      "step": 2429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010541160344107715,
+      "loss": 0.8359,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010534526710415435,
+      "loss": 0.7818,
+      "step": 2431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010527892840821756,
+      "loss": 0.7863,
+      "step": 2432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001052125873825439,
+      "loss": 0.88,
+      "step": 2433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010514624405641149,
+      "loss": 0.756,
+      "step": 2434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010507989845909949,
+      "loss": 0.798,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010501355061988803,
+      "loss": 0.7651,
+      "step": 2436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010494720056805825,
+      "loss": 0.7897,
+      "step": 2437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010488084833289224,
+      "loss": 0.7866,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010481449394367309,
+      "loss": 0.8503,
+      "step": 2439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010474813742968479,
+      "loss": 0.808,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010468177882021235,
+      "loss": 0.8344,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010461541814454163,
+      "loss": 0.7868,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010454905543195946,
+      "loss": 0.7984,
+      "step": 2443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010448269071175349,
+      "loss": 0.8401,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010441632401321234,
+      "loss": 0.7433,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010434995536562544,
+      "loss": 0.7525,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010428358479828313,
+      "loss": 0.8668,
+      "step": 2447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010421721234047659,
+      "loss": 0.8051,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010415083802149776,
+      "loss": 0.7392,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010408446187063951,
+      "loss": 0.8519,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010401808391719545,
+      "loss": 0.8268,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010395170419045998,
+      "loss": 0.8417,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010388532271972837,
+      "loss": 0.7926,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010381893953429653,
+      "loss": 0.8505,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010375255466346126,
+      "loss": 0.8516,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010368616813651996,
+      "loss": 0.7798,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010361977998277089,
+      "loss": 0.7809,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355339023151296,
+      "loss": 0.843,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010348699891204579,
+      "loss": 0.8329,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342060605366972,
+      "loss": 0.8007,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010335421168568573,
+      "loss": 0.7234,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010328781583739553,
+      "loss": 0.7498,
+      "step": 2462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010322141853810137,
+      "loss": 0.8457,
+      "step": 2463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010315501981710629,
+      "loss": 0.7133,
+      "step": 2464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010308861970371384,
+      "loss": 0.8693,
+      "step": 2465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010302221822722824,
+      "loss": 0.8507,
+      "step": 2466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010295581541695427,
+      "loss": 0.8163,
+      "step": 2467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010288941130219734,
+      "loss": 0.8324,
+      "step": 2468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010282300591226341,
+      "loss": 0.7732,
+      "step": 2469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010275659927645903,
+      "loss": 0.8402,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010269019142409127,
+      "loss": 0.8802,
+      "step": 2471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010262378238446774,
+      "loss": 0.794,
+      "step": 2472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001025573721868966,
+      "loss": 0.7766,
+      "step": 2473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001024909608606865,
+      "loss": 0.7569,
+      "step": 2474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010242454843514657,
+      "loss": 0.8324,
+      "step": 2475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010235813493958648,
+      "loss": 0.7679,
+      "step": 2476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010229172040331633,
+      "loss": 0.822,
+      "step": 2477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001022253048556467,
+      "loss": 0.7895,
+      "step": 2478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010215888832588856,
+      "loss": 0.7534,
+      "step": 2479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001020924708433534,
+      "loss": 0.7578,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010202605243735309,
+      "loss": 0.819,
+      "step": 2481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010195963313719987,
+      "loss": 0.796,
+      "step": 2482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010189321297220645,
+      "loss": 0.8037,
+      "step": 2483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010182679197168585,
+      "loss": 0.8189,
+      "step": 2484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010176037016495151,
+      "loss": 0.7778,
+      "step": 2485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010169394758131721,
+      "loss": 0.7361,
+      "step": 2486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010162752425009708,
+      "loss": 0.736,
+      "step": 2487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010156110020060554,
+      "loss": 0.8588,
+      "step": 2488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149467546215741,
+      "loss": 0.7547,
+      "step": 2489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010142825006406768,
+      "loss": 0.7824,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001013618240356518,
+      "loss": 0.8001,
+      "step": 2491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010129539740622533,
+      "loss": 0.7963,
+      "step": 2492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010122897020510427,
+      "loss": 0.7461,
+      "step": 2493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010116254246160468,
+      "loss": 0.6709,
+      "step": 2494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010109611420504299,
+      "loss": 0.808,
+      "step": 2495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010102968546473586,
+      "loss": 0.8171,
+      "step": 2496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001009632562700001,
+      "loss": 0.7603,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010089682665015274,
+      "loss": 0.8814,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010083039663451101,
+      "loss": 0.8463,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010076396625239233,
+      "loss": 0.7669,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010069753553311425,
+      "loss": 0.7634,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010063110450599447,
+      "loss": 0.7906,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010056467320035083,
+      "loss": 0.7377,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010049824164550135,
+      "loss": 0.7531,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010043180987076404,
+      "loss": 0.8568,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010036537790545712,
+      "loss": 0.7686,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010029894577889882,
+      "loss": 0.7893,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010023251352040752,
+      "loss": 0.7756,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010016608115930157,
+      "loss": 0.8707,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010009964872489943,
+      "loss": 0.7953,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010003321624651956,
+      "loss": 0.8405,
+      "step": 2511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996678375348045e-05,
+      "loss": 0.8362,
+      "step": 2512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990035127510059e-05,
+      "loss": 0.8078,
+      "step": 2513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983391884069842e-05,
+      "loss": 0.7834,
+      "step": 2514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97674864795925e-05,
+      "loss": 0.8227,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970105422110117e-05,
+      "loss": 0.7242,
+      "step": 2516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96346220945429e-05,
+      "loss": 0.8229,
+      "step": 2517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956819012923598e-05,
+      "loss": 0.8209,
+      "step": 2518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950175835449869e-05,
+      "loss": 0.826,
+      "step": 2519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943532679964916e-05,
+      "loss": 0.7882,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936889549400557e-05,
+      "loss": 0.8144,
+      "step": 2521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93024644668858e-05,
+      "loss": 0.7955,
+      "step": 2522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923603374760769e-05,
+      "loss": 0.7469,
+      "step": 2523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916960336548903e-05,
+      "loss": 0.7249,
+      "step": 2524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910317334984728e-05,
+      "loss": 0.7708,
+      "step": 2525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903674372999993e-05,
+      "loss": 0.7832,
+      "step": 2526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897031453526416e-05,
+      "loss": 0.7579,
+      "step": 2527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890388579495704e-05,
+      "loss": 0.8008,
+      "step": 2528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883745753839533e-05,
+      "loss": 0.7774,
+      "step": 2529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877102979489578e-05,
+      "loss": 0.7774,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870460259377465e-05,
+      "loss": 0.7772,
+      "step": 2531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863817596434822e-05,
+      "loss": 0.8405,
+      "step": 2532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857174993593231e-05,
+      "loss": 0.8101,
+      "step": 2533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850532453784262e-05,
+      "loss": 0.7648,
+      "step": 2534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843889979939444e-05,
+      "loss": 0.7864,
+      "step": 2535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837247574990295e-05,
+      "loss": 0.8192,
+      "step": 2536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830605241868279e-05,
+      "loss": 0.7615,
+      "step": 2537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823962983504852e-05,
+      "loss": 0.8123,
+      "step": 2538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817320802831416e-05,
+      "loss": 0.8206,
+      "step": 2539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810678702779358e-05,
+      "loss": 0.9144,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804036686280018e-05,
+      "loss": 0.808,
+      "step": 2541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797394756264695e-05,
+      "loss": 0.7952,
+      "step": 2542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790752915664665e-05,
+      "loss": 0.8186,
+      "step": 2543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784111167411146e-05,
+      "loss": 0.8236,
+      "step": 2544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777469514435335e-05,
+      "loss": 0.7673,
+      "step": 2545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770827959668368e-05,
+      "loss": 0.7928,
+      "step": 2546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764186506041356e-05,
+      "loss": 0.8271,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757545156485344e-05,
+      "loss": 0.8057,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750903913931353e-05,
+      "loss": 0.753,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744262781310343e-05,
+      "loss": 0.8096,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737621761553228e-05,
+      "loss": 0.8153,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730980857590873e-05,
+      "loss": 0.7892,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724340072354098e-05,
+      "loss": 0.8228,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717699408773659e-05,
+      "loss": 0.7852,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711058869780269e-05,
+      "loss": 0.8118,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704418458304574e-05,
+      "loss": 0.8251,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697778177277179e-05,
+      "loss": 0.8301,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691138029628616e-05,
+      "loss": 0.8074,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684498018289374e-05,
+      "loss": 0.8341,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677858146189864e-05,
+      "loss": 0.757,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67121841626045e-05,
+      "loss": 0.8115,
+      "step": 2561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66457883143143e-05,
+      "loss": 0.7401,
+      "step": 2562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657939394633031e-05,
+      "loss": 0.7726,
+      "step": 2563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651300108795426e-05,
+      "loss": 0.7546,
+      "step": 2564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644660976848708e-05,
+      "loss": 0.8024,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638022001722916e-05,
+      "loss": 0.803,
+      "step": 2566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631383186348006e-05,
+      "loss": 0.8013,
+      "step": 2567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624744533653879e-05,
+      "loss": 0.813,
+      "step": 2568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618106046570348e-05,
+      "loss": 0.8193,
+      "step": 2569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611467728027164e-05,
+      "loss": 0.7482,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604829580954003e-05,
+      "loss": 0.8098,
+      "step": 2571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598191608280459e-05,
+      "loss": 0.8628,
+      "step": 2572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59155381293605e-05,
+      "loss": 0.8186,
+      "step": 2573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584916197850226e-05,
+      "loss": 0.7996,
+      "step": 2574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578278765952342e-05,
+      "loss": 0.8644,
+      "step": 2575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571641520171688e-05,
+      "loss": 0.7376,
+      "step": 2576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565004463437456e-05,
+      "loss": 0.7331,
+      "step": 2577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558367598678769e-05,
+      "loss": 0.7797,
+      "step": 2578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551730928824656e-05,
+      "loss": 0.7759,
+      "step": 2579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545094456804058e-05,
+      "loss": 0.7518,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53845818554584e-05,
+      "loss": 0.8033,
+      "step": 2581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531822117978766e-05,
+      "loss": 0.7092,
+      "step": 2582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525186257031524e-05,
+      "loss": 0.7408,
+      "step": 2583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518550605632695e-05,
+      "loss": 0.737,
+      "step": 2584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511915166710781e-05,
+      "loss": 0.8848,
+      "step": 2585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505279943194176e-05,
+      "loss": 0.7836,
+      "step": 2586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4986449380112e-05,
+      "loss": 0.8433,
+      "step": 2587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49201015409005e-05,
+      "loss": 0.8132,
+      "step": 2588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485375594358852e-05,
+      "loss": 0.8203,
+      "step": 2589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478741261745611e-05,
+      "loss": 0.8252,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472107159178247e-05,
+      "loss": 0.8648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465473289584567e-05,
+      "loss": 0.7993,
+      "step": 2592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458839655892286e-05,
+      "loss": 0.7941,
+      "step": 2593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45220626102901e-05,
+      "loss": 0.7613,
+      "step": 2594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445573107922235e-05,
+      "loss": 0.788,
+      "step": 2595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438940199499355e-05,
+      "loss": 0.6853,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432307538687663e-05,
+      "loss": 0.843,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42567512841433e-05,
+      "loss": 0.826,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419042971606417e-05,
+      "loss": 0.7686,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.412411071190886e-05,
+      "loss": 0.7611,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40577943009457e-05,
+      "loss": 0.8182,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.399148051244204e-05,
+      "loss": 0.8339,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392516937566393e-05,
+      "loss": 0.8589,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385886091987633e-05,
+      "loss": 0.7984,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379255517434295e-05,
+      "loss": 0.7596,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372625216832644e-05,
+      "loss": 0.7943,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365995193108803e-05,
+      "loss": 0.8435,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359365449188794e-05,
+      "loss": 0.7997,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.352735987998497e-05,
+      "loss": 0.8317,
+      "step": 2609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346106812463688e-05,
+      "loss": 0.7722,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339477925509994e-05,
+      "loss": 0.7592,
+      "step": 2611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332849330062933e-05,
+      "loss": 0.7953,
+      "step": 2612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326221029047883e-05,
+      "loss": 0.8416,
+      "step": 2613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319593025390097e-05,
+      "loss": 0.755,
+      "step": 2614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3129653220147e-05,
+      "loss": 0.7826,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306337921846677e-05,
+      "loss": 0.7745,
+      "step": 2616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299710827810884e-05,
+      "loss": 0.8647,
+      "step": 2617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293084042832038e-05,
+      "loss": 0.7932,
+      "step": 2618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286457569834725e-05,
+      "loss": 0.7719,
+      "step": 2619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279831411743389e-05,
+      "loss": 0.8088,
+      "step": 2620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27320557148234e-05,
+      "loss": 0.7589,
+      "step": 2621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266580051975736e-05,
+      "loss": 0.8127,
+      "step": 2622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259954856147612e-05,
+      "loss": 0.7812,
+      "step": 2623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253329986921841e-05,
+      "loss": 0.7853,
+      "step": 2624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246705447222166e-05,
+      "loss": 0.811,
+      "step": 2625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240081239972176e-05,
+      "loss": 0.7142,
+      "step": 2626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233457368095323e-05,
+      "loss": 0.7852,
+      "step": 2627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226833834514897e-05,
+      "loss": 0.8169,
+      "step": 2628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220210642154052e-05,
+      "loss": 0.7826,
+      "step": 2629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21358779393578e-05,
+      "loss": 0.7932,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206965292782933e-05,
+      "loss": 0.862,
+      "step": 2631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200343141618197e-05,
+      "loss": 0.8091,
+      "step": 2632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193721343364117e-05,
+      "loss": 0.7403,
+      "step": 2633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187099900943071e-05,
+      "loss": 0.8486,
+      "step": 2634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180478817277286e-05,
+      "loss": 0.8141,
+      "step": 2635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173858095288828e-05,
+      "loss": 0.801,
+      "step": 2636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167237737899601e-05,
+      "loss": 0.8203,
+      "step": 2637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160617748031361e-05,
+      "loss": 0.8191,
+      "step": 2638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153998128605683e-05,
+      "loss": 0.8177,
+      "step": 2639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147378882543996e-05,
+      "loss": 0.836,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140760012767546e-05,
+      "loss": 0.8011,
+      "step": 2641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134141522197435e-05,
+      "loss": 0.8273,
+      "step": 2642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127523413754573e-05,
+      "loss": 0.7533,
+      "step": 2643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120905690359726e-05,
+      "loss": 0.7817,
+      "step": 2644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114288354933466e-05,
+      "loss": 0.7916,
+      "step": 2645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107671410396217e-05,
+      "loss": 0.8481,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101054859668213e-05,
+      "loss": 0.7955,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094438705669524e-05,
+      "loss": 0.7928,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087822951320035e-05,
+      "loss": 0.8938,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081207599539472e-05,
+      "loss": 0.795,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07459265324736e-05,
+      "loss": 0.7547,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067978115363069e-05,
+      "loss": 0.8223,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061363988805771e-05,
+      "loss": 0.7689,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05475027649446e-05,
+      "loss": 0.8355,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048136981347954e-05,
+      "loss": 0.8116,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041524106284885e-05,
+      "loss": 0.7803,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03491165422369e-05,
+      "loss": 0.7759,
+      "step": 2657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028299628082632e-05,
+      "loss": 0.7271,
+      "step": 2658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021688030779784e-05,
+      "loss": 0.8569,
+      "step": 2659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015076865233018e-05,
+      "loss": 0.8213,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008466134360031e-05,
+      "loss": 0.7346,
+      "step": 2661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001855841078317e-05,
+      "loss": 0.7793,
+      "step": 2662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995245988305183e-05,
+      "loss": 0.7682,
+      "step": 2663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988636578957737e-05,
+      "loss": 0.7529,
+      "step": 2664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982027615952898e-05,
+      "loss": 0.798,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975419102207379e-05,
+      "loss": 0.8319,
+      "step": 2666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968811040637702e-05,
+      "loss": 0.8823,
+      "step": 2667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962203434160189e-05,
+      "loss": 0.7741,
+      "step": 2668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955596285690957e-05,
+      "loss": 0.8561,
+      "step": 2669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94898959814592e-05,
+      "loss": 0.8016,
+      "step": 2670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942383374440801e-05,
+      "loss": 0.7232,
+      "step": 2671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935777617491102e-05,
+      "loss": 0.756,
+      "step": 2672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929172330212124e-05,
+      "loss": 0.7805,
+      "step": 2673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922567515518968e-05,
+      "loss": 0.7527,
+      "step": 2674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915963176326516e-05,
+      "loss": 0.8038,
+      "step": 2675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909359315549451e-05,
+      "loss": 0.7269,
+      "step": 2676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902755936102233e-05,
+      "loss": 0.7873,
+      "step": 2677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89615304089912e-05,
+      "loss": 0.7602,
+      "step": 2678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889550632854148e-05,
+      "loss": 0.7301,
+      "step": 2679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88294871488115e-05,
+      "loss": 0.7996,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876347289893724e-05,
+      "loss": 0.8006,
+      "step": 2681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869746360805273e-05,
+      "loss": 0.8199,
+      "step": 2682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863145930528956e-05,
+      "loss": 0.761,
+      "step": 2683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856546001977737e-05,
+      "loss": 0.8078,
+      "step": 2684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849946578064339e-05,
+      "loss": 0.7666,
+      "step": 2685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843347661701276e-05,
+      "loss": 0.7833,
+      "step": 2686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836749255800825e-05,
+      "loss": 0.826,
+      "step": 2687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830151363275052e-05,
+      "loss": 0.8194,
+      "step": 2688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823553987035781e-05,
+      "loss": 0.7363,
+      "step": 2689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816957129994624e-05,
+      "loss": 0.7908,
+      "step": 2690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810360795062953e-05,
+      "loss": 0.795,
+      "step": 2691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80376498515191e-05,
+      "loss": 0.7003,
+      "step": 2692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797169703172416e-05,
+      "loss": 0.7966,
+      "step": 2693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79057495203514e-05,
+      "loss": 0.8293,
+      "step": 2694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78398073465054e-05,
+      "loss": 0.8213,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777387053928815e-05,
+      "loss": 0.7653,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770793912779946e-05,
+      "loss": 0.7881,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764201314113662e-05,
+      "loss": 0.7407,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757609260839464e-05,
+      "loss": 0.8308,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751017755866607e-05,
+      "loss": 0.8116,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744426802104101e-05,
+      "loss": 0.8135,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737836402460717e-05,
+      "loss": 0.8409,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731246559844984e-05,
+      "loss": 0.8184,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724657277165175e-05,
+      "loss": 0.8556,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71806855732933e-05,
+      "loss": 0.7692,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711480403245224e-05,
+      "loss": 0.7674,
+      "step": 2706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.7048928178204e-05,
+      "loss": 0.8226,
+      "step": 2707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698305803962134e-05,
+      "loss": 0.7935,
+      "step": 2708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691719364577465e-05,
+      "loss": 0.7328,
+      "step": 2709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685133502573164e-05,
+      "loss": 0.8554,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678548220855754e-05,
+      "loss": 0.7816,
+      "step": 2711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671963522331506e-05,
+      "loss": 0.7551,
+      "step": 2712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665379409906424e-05,
+      "loss": 0.8435,
+      "step": 2713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658795886486266e-05,
+      "loss": 0.7203,
+      "step": 2714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652212954976513e-05,
+      "loss": 0.7651,
+      "step": 2715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645630618282405e-05,
+      "loss": 0.7802,
+      "step": 2716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639048879308898e-05,
+      "loss": 0.8116,
+      "step": 2717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632467740960706e-05,
+      "loss": 0.8255,
+      "step": 2718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625887206142257e-05,
+      "loss": 0.8119,
+      "step": 2719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61930727775773e-05,
+      "loss": 0.7187,
+      "step": 2720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612727958711026e-05,
+      "loss": 0.724,
+      "step": 2721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60614925190578e-05,
+      "loss": 0.8055,
+      "step": 2722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599571160245356e-05,
+      "loss": 0.8085,
+      "step": 2723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592993686632856e-05,
+      "loss": 0.8028,
+      "step": 2724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586416833971087e-05,
+      "loss": 0.7701,
+      "step": 2725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579840605162607e-05,
+      "loss": 0.7786,
+      "step": 2726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57326500310968e-05,
+      "loss": 0.7973,
+      "step": 2727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566690030714309e-05,
+      "loss": 0.7848,
+      "step": 2728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560115690878204e-05,
+      "loss": 0.8526,
+      "step": 2729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553541986502799e-05,
+      "loss": 0.752,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54696892048926e-05,
+      "loss": 0.8519,
+      "step": 2731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540396495738458e-05,
+      "loss": 0.7943,
+      "step": 2732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533824715150989e-05,
+      "loss": 0.8297,
+      "step": 2733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527253581627155e-05,
+      "loss": 0.8698,
+      "step": 2734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520683098066988e-05,
+      "loss": 0.8285,
+      "step": 2735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514113267370213e-05,
+      "loss": 0.7369,
+      "step": 2736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507544092436288e-05,
+      "loss": 0.7805,
+      "step": 2737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500975576164361e-05,
+      "loss": 0.6899,
+      "step": 2738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494407721453311e-05,
+      "loss": 0.7617,
+      "step": 2739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487840531201703e-05,
+      "loss": 0.7906,
+      "step": 2740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481274008307828e-05,
+      "loss": 0.8167,
+      "step": 2741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474708155669667e-05,
+      "loss": 0.7541,
+      "step": 2742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468142976184914e-05,
+      "loss": 0.8025,
+      "step": 2743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461578472750967e-05,
+      "loss": 0.8179,
+      "step": 2744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455014648264921e-05,
+      "loss": 0.7932,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448451505623568e-05,
+      "loss": 0.7855,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441889047723416e-05,
+      "loss": 0.8138,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43532727746065e-05,
+      "loss": 0.7661,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428766197731158e-05,
+      "loss": 0.799,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422205811430534e-05,
+      "loss": 0.7693,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415646121454049e-05,
+      "loss": 0.7939,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409087130696682e-05,
+      "loss": 0.824,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402528842053094e-05,
+      "loss": 0.8232,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39597125841764e-05,
+      "loss": 0.8411,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389414382684359e-05,
+      "loss": 0.8108,
+      "step": 2755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382858217746989e-05,
+      "loss": 0.7888,
+      "step": 2756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376302766498936e-05,
+      "loss": 0.7833,
+      "step": 2757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369748031833313e-05,
+      "loss": 0.8271,
+      "step": 2758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363194016642896e-05,
+      "loss": 0.7803,
+      "step": 2759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356640723820161e-05,
+      "loss": 0.8024,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350088156257248e-05,
+      "loss": 0.7753,
+      "step": 2761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343536316845996e-05,
+      "loss": 0.6948,
+      "step": 2762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336985208477902e-05,
+      "loss": 0.8126,
+      "step": 2763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33043483404416e-05,
+      "loss": 0.819,
+      "step": 2764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323885196435627e-05,
+      "loss": 0.8599,
+      "step": 2765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317336298542839e-05,
+      "loss": 0.8085,
+      "step": 2766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31078814325601e-05,
+      "loss": 0.7359,
+      "step": 2767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304240733465012e-05,
+      "loss": 0.7778,
+      "step": 2768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297694072059407e-05,
+      "loss": 0.847,
+      "step": 2769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291148161928412e-05,
+      "loss": 0.8116,
+      "step": 2770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284603005960923e-05,
+      "loss": 0.8192,
+      "step": 2771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278058607045492e-05,
+      "loss": 0.8446,
+      "step": 2772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271514968070348e-05,
+      "loss": 0.8013,
+      "step": 2773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264972091923375e-05,
+      "loss": 0.7866,
+      "step": 2774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258429981492128e-05,
+      "loss": 0.7631,
+      "step": 2775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251888639663822e-05,
+      "loss": 0.7963,
+      "step": 2776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245348069325329e-05,
+      "loss": 0.8264,
+      "step": 2777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238808273363183e-05,
+      "loss": 0.7624,
+      "step": 2778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232269254663582e-05,
+      "loss": 0.8776,
+      "step": 2779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225731016112367e-05,
+      "loss": 0.7874,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219193560595052e-05,
+      "loss": 0.7301,
+      "step": 2781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212656890996789e-05,
+      "loss": 0.8527,
+      "step": 2782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206121010202398e-05,
+      "loss": 0.8074,
+      "step": 2783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199585921096334e-05,
+      "loss": 0.8261,
+      "step": 2784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193051626562725e-05,
+      "loss": 0.752,
+      "step": 2785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186518129485324e-05,
+      "loss": 0.7645,
+      "step": 2786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.179985432747548e-05,
+      "loss": 0.7901,
+      "step": 2787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17345353923246e-05,
+      "loss": 0.8123,
+      "step": 2788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166922451822758e-05,
+      "loss": 0.7634,
+      "step": 2789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1603921734008e-05,
+      "loss": 0.7581,
+      "step": 2790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153862706848568e-05,
+      "loss": 0.8328,
+      "step": 2791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147334055047706e-05,
+      "loss": 0.8648,
+      "step": 2792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14080622087948e-05,
+      "loss": 0.7741,
+      "step": 2793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13427920722481e-05,
+      "loss": 0.8008,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12775301696424e-05,
+      "loss": 0.7533,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121227652977965e-05,
+      "loss": 0.7398,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114703118145804e-05,
+      "loss": 0.8493,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108179415347217e-05,
+      "loss": 0.8217,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101656547461292e-05,
+      "loss": 0.8643,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095134517366755e-05,
+      "loss": 0.8177,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08861332794195e-05,
+      "loss": 0.7449,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082092982064867e-05,
+      "loss": 0.704,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075573482613108e-05,
+      "loss": 0.8128,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069054832463914e-05,
+      "loss": 0.7565,
+      "step": 2804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062537034494138e-05,
+      "loss": 0.7352,
+      "step": 2805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05602009158027e-05,
+      "loss": 0.7643,
+      "step": 2806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049504006598413e-05,
+      "loss": 0.8178,
+      "step": 2807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042988782424298e-05,
+      "loss": 0.8103,
+      "step": 2808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036474421933276e-05,
+      "loss": 0.8212,
+      "step": 2809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029960928000305e-05,
+      "loss": 0.8466,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023448303499981e-05,
+      "loss": 0.6953,
+      "step": 2811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016936551306495e-05,
+      "loss": 0.7328,
+      "step": 2812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010425674293672e-05,
+      "loss": 0.7909,
+      "step": 2813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003915675334933e-05,
+      "loss": 0.8515,
+      "step": 2814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997406557303327e-05,
+      "loss": 0.817,
+      "step": 2815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990898323071503e-05,
+      "loss": 0.7076,
+      "step": 2816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98439097551173e-05,
+      "loss": 0.7664,
+      "step": 2817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977884517495873e-05,
+      "loss": 0.8751,
+      "step": 2818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971378951895416e-05,
+      "loss": 0.7664,
+      "step": 2819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964874281581444e-05,
+      "loss": 0.7281,
+      "step": 2820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958370509424651e-05,
+      "loss": 0.8848,
+      "step": 2821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951867638295323e-05,
+      "loss": 0.8072,
+      "step": 2822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945365671063367e-05,
+      "loss": 0.8116,
+      "step": 2823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938864610598275e-05,
+      "loss": 0.8239,
+      "step": 2824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932364459769141e-05,
+      "loss": 0.7604,
+      "step": 2825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92586522144467e-05,
+      "loss": 0.7843,
+      "step": 2826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919366898493143e-05,
+      "loss": 0.7543,
+      "step": 2827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91286949378246e-05,
+      "loss": 0.82,
+      "step": 2828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906373010180104e-05,
+      "loss": 0.8223,
+      "step": 2829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899877450553146e-05,
+      "loss": 0.8322,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893382817768261e-05,
+      "loss": 0.7388,
+      "step": 2831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886889114691713e-05,
+      "loss": 0.8156,
+      "step": 2832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880396344189343e-05,
+      "loss": 0.7915,
+      "step": 2833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8739045091266e-05,
+      "loss": 0.7991,
+      "step": 2834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8674136123685e-05,
+      "loss": 0.7744,
+      "step": 2835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860923656779665e-05,
+      "loss": 0.779,
+      "step": 2836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854434645224283e-05,
+      "loss": 0.8009,
+      "step": 2837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847946580566141e-05,
+      "loss": 0.77,
+      "step": 2838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841459465668595e-05,
+      "loss": 0.8105,
+      "step": 2839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834973303394593e-05,
+      "loss": 0.7814,
+      "step": 2840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828488096606655e-05,
+      "loss": 0.7729,
+      "step": 2841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822003848166881e-05,
+      "loss": 0.7957,
+      "step": 2842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815520560936954e-05,
+      "loss": 0.8068,
+      "step": 2843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809038237778119e-05,
+      "loss": 0.795,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802556881551213e-05,
+      "loss": 0.7639,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79607649511663e-05,
+      "loss": 0.7122,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78959708133435e-05,
+      "loss": 0.7832,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783118643063911e-05,
+      "loss": 0.8263,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.776641183164432e-05,
+      "loss": 0.8191,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770164704494592e-05,
+      "loss": 0.8527,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763689209912641e-05,
+      "loss": 0.7926,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757214702276391e-05,
+      "loss": 0.7709,
+      "step": 2852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750741184443228e-05,
+      "loss": 0.8252,
+      "step": 2853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744268659270085e-05,
+      "loss": 0.8375,
+      "step": 2854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737797129613474e-05,
+      "loss": 0.8147,
+      "step": 2855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731326598329452e-05,
+      "loss": 0.7878,
+      "step": 2856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724857068273653e-05,
+      "loss": 0.7411,
+      "step": 2857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718388542301248e-05,
+      "loss": 0.8836,
+      "step": 2858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711921023266987e-05,
+      "loss": 0.7939,
+      "step": 2859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705454514025154e-05,
+      "loss": 0.8821,
+      "step": 2860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698989017429607e-05,
+      "loss": 0.8063,
+      "step": 2861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692524536333746e-05,
+      "loss": 0.8456,
+      "step": 2862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68606107359052e-05,
+      "loss": 0.7887,
+      "step": 2863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679598632052441e-05,
+      "loss": 0.7985,
+      "step": 2864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673137214571555e-05,
+      "loss": 0.7839,
+      "step": 2865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666676823999473e-05,
+      "loss": 0.8051,
+      "step": 2866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660217463187334e-05,
+      "loss": 0.7555,
+      "step": 2867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653759134985843e-05,
+      "loss": 0.7565,
+      "step": 2868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647301842245227e-05,
+      "loss": 0.7917,
+      "step": 2869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640845587815279e-05,
+      "loss": 0.8168,
+      "step": 2870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63439037454531e-05,
+      "loss": 0.7956,
+      "step": 2871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627936205284195e-05,
+      "loss": 0.7688,
+      "step": 2872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621483082880332e-05,
+      "loss": 0.7558,
+      "step": 2873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615031010181663e-05,
+      "loss": 0.7264,
+      "step": 2874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608579990035665e-05,
+      "loss": 0.765,
+      "step": 2875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602130025289353e-05,
+      "loss": 0.8702,
+      "step": 2876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595681118789271e-05,
+      "loss": 0.7714,
+      "step": 2877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589233273381508e-05,
+      "loss": 0.7719,
+      "step": 2878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582786491911669e-05,
+      "loss": 0.7399,
+      "step": 2879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576340777224892e-05,
+      "loss": 0.8627,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56989613216586e-05,
+      "loss": 0.7509,
+      "step": 2881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563452559578765e-05,
+      "loss": 0.7668,
+      "step": 2882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557010062307337e-05,
+      "loss": 0.8157,
+      "step": 2883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550568643194826e-05,
+      "loss": 0.7538,
+      "step": 2884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544128305084013e-05,
+      "loss": 0.8638,
+      "step": 2885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537689050817189e-05,
+      "loss": 0.7429,
+      "step": 2886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531250883236182e-05,
+      "loss": 0.7679,
+      "step": 2887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524813805182326e-05,
+      "loss": 0.7737,
+      "step": 2888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518377819496492e-05,
+      "loss": 0.712,
+      "step": 2889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511942929019046e-05,
+      "loss": 0.9028,
+      "step": 2890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505509136589893e-05,
+      "loss": 0.8162,
+      "step": 2891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499076445048434e-05,
+      "loss": 0.7605,
+      "step": 2892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492644857233601e-05,
+      "loss": 0.8087,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486214375983828e-05,
+      "loss": 0.7872,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479785004137065e-05,
+      "loss": 0.8281,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473356744530768e-05,
+      "loss": 0.78,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466929600001912e-05,
+      "loss": 0.7691,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46050357338697e-05,
+      "loss": 0.8076,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45407866752192e-05,
+      "loss": 0.82,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447654885242259e-05,
+      "loss": 0.7916,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441232229382972e-05,
+      "loss": 0.8009,
+      "step": 2901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43481070277856e-05,
+      "loss": 0.7922,
+      "step": 2902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428390308263015e-05,
+      "loss": 0.8072,
+      "step": 2903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421971048669841e-05,
+      "loss": 0.8392,
+      "step": 2904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415552926832026e-05,
+      "loss": 0.8037,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409135945582075e-05,
+      "loss": 0.7925,
+      "step": 2906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402720107751968e-05,
+      "loss": 0.7955,
+      "step": 2907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396305416173202e-05,
+      "loss": 0.8051,
+      "step": 2908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389891873676747e-05,
+      "loss": 0.8064,
+      "step": 2909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383479483093085e-05,
+      "loss": 0.7762,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377068247252174e-05,
+      "loss": 0.7824,
+      "step": 2911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370658168983476e-05,
+      "loss": 0.7462,
+      "step": 2912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36424925111593e-05,
+      "loss": 0.7749,
+      "step": 2913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357841496477973e-05,
+      "loss": 0.8303,
+      "step": 2914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351434907897519e-05,
+      "loss": 0.7821,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345029488201976e-05,
+      "loss": 0.7731,
+      "step": 2916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338625240218236e-05,
+      "loss": 0.7614,
+      "step": 2917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332222166772661e-05,
+      "loss": 0.8607,
+      "step": 2918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325820270691115e-05,
+      "loss": 0.7756,
+      "step": 2919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319419554798921e-05,
+      "loss": 0.739,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3130200219209e-05,
+      "loss": 0.7738,
+      "step": 2921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306621674881335e-05,
+      "loss": 0.7896,
+      "step": 2922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300224516504003e-05,
+      "loss": 0.7758,
+      "step": 2923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293828549612136e-05,
+      "loss": 0.6691,
+      "step": 2924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28743377702846e-05,
+      "loss": 0.7672,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281040201575158e-05,
+      "loss": 0.7237,
+      "step": 2926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274647826073897e-05,
+      "loss": 0.7758,
+      "step": 2927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268256653345801e-05,
+      "loss": 0.8073,
+      "step": 2928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261866686211482e-05,
+      "loss": 0.7613,
+      "step": 2929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255477927490997e-05,
+      "loss": 0.8085,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249090380003892e-05,
+      "loss": 0.8154,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242704046569159e-05,
+      "loss": 0.8637,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236318930005275e-05,
+      "loss": 0.8127,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229935033130155e-05,
+      "loss": 0.7563,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223552358761199e-05,
+      "loss": 0.7752,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217170909715251e-05,
+      "loss": 0.7967,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210790688808621e-05,
+      "loss": 0.7099,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204411698857083e-05,
+      "loss": 0.8229,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198033942675855e-05,
+      "loss": 0.7664,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191657423079621e-05,
+      "loss": 0.8983,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18528214288251e-05,
+      "loss": 0.8258,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178908104898117e-05,
+      "loss": 0.7652,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172535311939472e-05,
+      "loss": 0.7076,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166163766819074e-05,
+      "loss": 0.714,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159793472348852e-05,
+      "loss": 0.7533,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1534244313402e-05,
+      "loss": 0.8069,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147056646603949e-05,
+      "loss": 0.7449,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140690120950379e-05,
+      "loss": 0.7023,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134324857189211e-05,
+      "loss": 0.7645,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127960858129618e-05,
+      "loss": 0.7974,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1215981265802e-05,
+      "loss": 0.8218,
+      "step": 2951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.115236665349015e-05,
+      "loss": 0.8451,
+      "step": 2952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108876477243541e-05,
+      "loss": 0.7572,
+      "step": 2953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102517565070716e-05,
+      "loss": 0.8065,
+      "step": 2954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0961599316369e-05,
+      "loss": 0.7947,
+      "step": 2955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089803579747883e-05,
+      "loss": 0.7709,
+      "step": 2956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08344851220891e-05,
+      "loss": 0.7463,
+      "step": 2957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07709473182464e-05,
+      "loss": 0.8273,
+      "step": 2958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070742241399182e-05,
+      "loss": 0.8233,
+      "step": 2959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064391043736053e-05,
+      "loss": 0.7306,
+      "step": 2960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058041141638222e-05,
+      "loss": 0.7263,
+      "step": 2961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051692537908068e-05,
+      "loss": 0.7875,
+      "step": 2962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045345235347413e-05,
+      "loss": 0.8357,
+      "step": 2963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038999236757488e-05,
+      "loss": 0.7958,
+      "step": 2964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.032654544938968e-05,
+      "loss": 0.7519,
+      "step": 2965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026311162691931e-05,
+      "loss": 0.8224,
+      "step": 2966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019969092815895e-05,
+      "loss": 0.8365,
+      "step": 2967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013628338109781e-05,
+      "loss": 0.7422,
+      "step": 2968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007288901371947e-05,
+      "loss": 0.7822,
+      "step": 2969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000950785400158e-05,
+      "loss": 0.7232,
+      "step": 2970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994613992991603e-05,
+      "loss": 0.797,
+      "step": 2971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988278526942878e-05,
+      "loss": 0.7664,
+      "step": 2972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981944390050005e-05,
+      "loss": 0.8751,
+      "step": 2973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975611585108411e-05,
+      "loss": 0.7999,
+      "step": 2974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969280114912935e-05,
+      "loss": 0.7725,
+      "step": 2975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962949982257834e-05,
+      "loss": 0.8288,
+      "step": 2976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956621189936765e-05,
+      "loss": 0.8224,
+      "step": 2977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950293740742803e-05,
+      "loss": 0.814,
+      "step": 2978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943967637468427e-05,
+      "loss": 0.7692,
+      "step": 2979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93764288290552e-05,
+      "loss": 0.8002,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931319479845366e-05,
+      "loss": 0.8042,
+      "step": 2981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924997431078666e-05,
+      "loss": 0.7378,
+      "step": 2982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918676739395506e-05,
+      "loss": 0.758,
+      "step": 2983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91235740758539e-05,
+      "loss": 0.8483,
+      "step": 2984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906039438437206e-05,
+      "loss": 0.8054,
+      "step": 2985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899722834739257e-05,
+      "loss": 0.7619,
+      "step": 2986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893407599279222e-05,
+      "loss": 0.8257,
+      "step": 2987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887093734844204e-05,
+      "loss": 0.8229,
+      "step": 2988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880781244220671e-05,
+      "loss": 0.7833,
+      "step": 2989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87447013019451e-05,
+      "loss": 0.83,
+      "step": 2990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868160395550985e-05,
+      "loss": 0.783,
+      "step": 2991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861852043074758e-05,
+      "loss": 0.8177,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85554507554988e-05,
+      "loss": 0.8401,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849239495759784e-05,
+      "loss": 0.7783,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842935306487308e-05,
+      "loss": 0.7535,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836632510514651e-05,
+      "loss": 0.7795,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830331110623423e-05,
+      "loss": 0.7564,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824031109594596e-05,
+      "loss": 0.7838,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817732510208541e-05,
+      "loss": 0.7802,
+      "step": 2999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811435315245e-05,
+      "loss": 0.8255,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8051395274831e-05,
+      "loss": 0.7517,
+      "step": 3001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798845149701347e-05,
+      "loss": 0.817,
+      "step": 3002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792552184677628e-05,
+      "loss": 0.7399,
+      "step": 3003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786260635189192e-05,
+      "loss": 0.7719,
+      "step": 3004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779970504012683e-05,
+      "loss": 0.7556,
+      "step": 3005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773681793924103e-05,
+      "loss": 0.7543,
+      "step": 3006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767394507698839e-05,
+      "loss": 0.7826,
+      "step": 3007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761108648111638e-05,
+      "loss": 0.8114,
+      "step": 3008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754824217936628e-05,
+      "loss": 0.7889,
+      "step": 3009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748541219947298e-05,
+      "loss": 0.7586,
+      "step": 3010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742259656916517e-05,
+      "loss": 0.7852,
+      "step": 3011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7359795316165e-05,
+      "loss": 0.7524,
+      "step": 3012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729700846818845e-05,
+      "loss": 0.7602,
+      "step": 3013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723423605294515e-05,
+      "loss": 0.8428,
+      "step": 3014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717147809813817e-05,
+      "loss": 0.7379,
+      "step": 3015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710873463146447e-05,
+      "loss": 0.7849,
+      "step": 3016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704600568061437e-05,
+      "loss": 0.8348,
+      "step": 3017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698329127327196e-05,
+      "loss": 0.8122,
+      "step": 3018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692059143711477e-05,
+      "loss": 0.8769,
+      "step": 3019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685790619981404e-05,
+      "loss": 0.8227,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679523558903442e-05,
+      "loss": 0.7887,
+      "step": 3021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673257963243425e-05,
+      "loss": 0.8046,
+      "step": 3022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66699383576653e-05,
+      "loss": 0.7249,
+      "step": 3023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660731179237289e-05,
+      "loss": 0.7562,
+      "step": 3024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654469996419585e-05,
+      "loss": 0.7889,
+      "step": 3025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648210290076658e-05,
+      "loss": 0.7521,
+      "step": 3026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641952062971077e-05,
+      "loss": 0.7083,
+      "step": 3027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635695317864783e-05,
+      "loss": 0.7863,
+      "step": 3028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629440057519038e-05,
+      "loss": 0.7222,
+      "step": 3029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623186284694474e-05,
+      "loss": 0.8035,
+      "step": 3030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616934002151043e-05,
+      "loss": 0.7327,
+      "step": 3031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610683212648055e-05,
+      "loss": 0.8119,
+      "step": 3032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604433918944153e-05,
+      "loss": 0.8098,
+      "step": 3033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598186123797323e-05,
+      "loss": 0.7983,
+      "step": 3034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591939829964898e-05,
+      "loss": 0.7286,
+      "step": 3035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585695040203528e-05,
+      "loss": 0.7912,
+      "step": 3036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57945175726922e-05,
+      "loss": 0.8664,
+      "step": 3037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5732099839173e-05,
+      "loss": 0.7706,
+      "step": 3038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566969722902442e-05,
+      "loss": 0.7909,
+      "step": 3039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560730976978637e-05,
+      "loss": 0.7599,
+      "step": 3040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554493748899227e-05,
+      "loss": 0.8077,
+      "step": 3041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548258041416863e-05,
+      "loss": 0.804,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542023857283539e-05,
+      "loss": 0.7789,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535791199250574e-05,
+      "loss": 0.7598,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529560070068608e-05,
+      "loss": 0.7918,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523330472487613e-05,
+      "loss": 0.8109,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517102409256886e-05,
+      "loss": 0.7951,
+      "step": 3047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510875883125035e-05,
+      "loss": 0.7304,
+      "step": 3048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504650896840007e-05,
+      "loss": 0.7758,
+      "step": 3049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498427453149056e-05,
+      "loss": 0.7888,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492205554798754e-05,
+      "loss": 0.8413,
+      "step": 3051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485985204535007e-05,
+      "loss": 0.7448,
+      "step": 3052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479766405103018e-05,
+      "loss": 0.8016,
+      "step": 3053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473549159247321e-05,
+      "loss": 0.7921,
+      "step": 3054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467333469711756e-05,
+      "loss": 0.7558,
+      "step": 3055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461119339239479e-05,
+      "loss": 0.8489,
+      "step": 3056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454906770572955e-05,
+      "loss": 0.8145,
+      "step": 3057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448695766453969e-05,
+      "loss": 0.8462,
+      "step": 3058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4424863296236e-05,
+      "loss": 0.8607,
+      "step": 3059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436278462822252e-05,
+      "loss": 0.7556,
+      "step": 3060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430072168789619e-05,
+      "loss": 0.8658,
+      "step": 3061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42386745026472e-05,
+      "loss": 0.7468,
+      "step": 3062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.417664309985858e-05,
+      "loss": 0.7397,
+      "step": 3063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41146275069066e-05,
+      "loss": 0.7833,
+      "step": 3064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405262775116038e-05,
+      "loss": 0.7336,
+      "step": 3065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399064385998216e-05,
+      "loss": 0.8262,
+      "step": 3066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392867586072714e-05,
+      "loss": 0.7177,
+      "step": 3067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386672378074351e-05,
+      "loss": 0.8287,
+      "step": 3068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380478764737244e-05,
+      "loss": 0.8417,
+      "step": 3069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374286748794801e-05,
+      "loss": 0.7849,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368096332979737e-05,
+      "loss": 0.8217,
+      "step": 3071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361907520024046e-05,
+      "loss": 0.7885,
+      "step": 3072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355720312659028e-05,
+      "loss": 0.7813,
+      "step": 3073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349534713615264e-05,
+      "loss": 0.8045,
+      "step": 3074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343350725622632e-05,
+      "loss": 0.762,
+      "step": 3075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337168351410297e-05,
+      "loss": 0.7574,
+      "step": 3076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330987593706712e-05,
+      "loss": 0.7594,
+      "step": 3077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324808455239612e-05,
+      "loss": 0.784,
+      "step": 3078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318630938736028e-05,
+      "loss": 0.7944,
+      "step": 3079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312455046922261e-05,
+      "loss": 0.8365,
+      "step": 3080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30628078252391e-05,
+      "loss": 0.734,
+      "step": 3081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30010814826584e-05,
+      "loss": 0.8038,
+      "step": 3082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293937146872216e-05,
+      "loss": 0.765,
+      "step": 3083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28776778106646e-05,
+      "loss": 0.8044,
+      "step": 3084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281600053571293e-05,
+      "loss": 0.8001,
+      "step": 3085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275433967108694e-05,
+      "loss": 0.744,
+      "step": 3086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269269524399936e-05,
+      "loss": 0.7117,
+      "step": 3087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263106728165555e-05,
+      "loss": 0.8038,
+      "step": 3088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25694558112536e-05,
+      "loss": 0.8505,
+      "step": 3089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250786085998439e-05,
+      "loss": 0.7213,
+      "step": 3090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244628245503143e-05,
+      "loss": 0.7877,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238472062357103e-05,
+      "loss": 0.846,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232317539277206e-05,
+      "loss": 0.8367,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22616467897962e-05,
+      "loss": 0.778,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220013484179766e-05,
+      "loss": 0.7661,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213863957592343e-05,
+      "loss": 0.7937,
+      "step": 3096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207716101931296e-05,
+      "loss": 0.7364,
+      "step": 3097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201569919909854e-05,
+      "loss": 0.7642,
+      "step": 3098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19542541424049e-05,
+      "loss": 0.7936,
+      "step": 3099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189282587634952e-05,
+      "loss": 0.772,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18314144280423e-05,
+      "loss": 0.7409,
+      "step": 3101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177001982458587e-05,
+      "loss": 0.876,
+      "step": 3102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170864209307532e-05,
+      "loss": 0.7647,
+      "step": 3103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16472812605984e-05,
+      "loss": 0.8039,
+      "step": 3104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158593735423527e-05,
+      "loss": 0.8263,
+      "step": 3105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152461040105871e-05,
+      "loss": 0.8118,
+      "step": 3106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146330042813402e-05,
+      "loss": 0.7618,
+      "step": 3107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140200746251896e-05,
+      "loss": 0.7765,
+      "step": 3108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134073153126383e-05,
+      "loss": 0.7012,
+      "step": 3109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127947266141135e-05,
+      "loss": 0.8005,
+      "step": 3110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121823087999682e-05,
+      "loss": 0.6527,
+      "step": 3111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115700621404786e-05,
+      "loss": 0.7461,
+      "step": 3112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109579869058465e-05,
+      "loss": 0.769,
+      "step": 3113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103460833661974e-05,
+      "loss": 0.7966,
+      "step": 3114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0973435179158145e-05,
+      "loss": 0.8162,
+      "step": 3115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091227924519721e-05,
+      "loss": 0.762,
+      "step": 3116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0851140561726826e-05,
+      "loss": 0.8984,
+      "step": 3117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079001915572909e-05,
+      "loss": 0.7825,
+      "step": 3118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0728915054178635e-05,
+      "loss": 0.7805,
+      "step": 3119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066782828404236e-05,
+      "loss": 0.7943,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060675887227956e-05,
+      "loss": 0.7621,
+      "step": 3121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054570684584181e-05,
+      "loss": 0.8328,
+      "step": 3122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048467223167314e-05,
+      "loss": 0.8286,
+      "step": 3123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0423655056709736e-05,
+      "loss": 0.7413,
+      "step": 3124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0362655347880146e-05,
+      "loss": 0.7948,
+      "step": 3125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.030167313210531e-05,
+      "loss": 0.8008,
+      "step": 3126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024070843629828e-05,
+      "loss": 0.757,
+      "step": 3127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017976128736451e-05,
+      "loss": 0.7792,
+      "step": 3128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011883171220165e-05,
+      "loss": 0.7878,
+      "step": 3129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0057919737699586e-05,
+      "loss": 0.8074,
+      "step": 3130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9997025390740456e-05,
+      "loss": 0.8401,
+      "step": 3131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993614869819868e-05,
+      "loss": 0.7763,
+      "step": 3132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9875289686940714e-05,
+      "loss": 0.7565,
+      "step": 3133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981444838382544e-05,
+      "loss": 0.7895,
+      "step": 3134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975362481570368e-05,
+      "loss": 0.7554,
+      "step": 3135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969281900941868e-05,
+      "loss": 0.7745,
+      "step": 3136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96320309918056e-05,
+      "loss": 0.8332,
+      "step": 3137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9571260789691975e-05,
+      "loss": 0.734,
+      "step": 3138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9510508429897296e-05,
+      "loss": 0.7469,
+      "step": 3139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94497739392333e-05,
+      "loss": 0.7237,
+      "step": 3140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938905734450378e-05,
+      "loss": 0.7471,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932835867250466e-05,
+      "loss": 0.7917,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9267677950023946e-05,
+      "loss": 0.7719,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920701520384166e-05,
+      "loss": 0.7887,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914637046073003e-05,
+      "loss": 0.7487,
+      "step": 3145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908574374745317e-05,
+      "loss": 0.7766,
+      "step": 3146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9025135090767394e-05,
+      "loss": 0.7366,
+      "step": 3147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8964544517420926e-05,
+      "loss": 0.843,
+      "step": 3148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8903972054154097e-05,
+      "loss": 0.7979,
+      "step": 3149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884341772769917e-05,
+      "loss": 0.7816,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878288156478049e-05,
+      "loss": 0.8268,
+      "step": 3151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872236359211429e-05,
+      "loss": 0.8058,
+      "step": 3152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866186383640885e-05,
+      "loss": 0.7856,
+      "step": 3153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860138232436435e-05,
+      "loss": 0.7797,
+      "step": 3154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854091908267303e-05,
+      "loss": 0.8419,
+      "step": 3155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848047413801887e-05,
+      "loss": 0.7969,
+      "step": 3156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8420047517078013e-05,
+      "loss": 0.7699,
+      "step": 3157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83596392465183e-05,
+      "loss": 0.7881,
+      "step": 3158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829924935299964e-05,
+      "loss": 0.7791,
+      "step": 3159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82388778631737e-05,
+      "loss": 0.7684,
+      "step": 3160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817852480368416e-05,
+      "loss": 0.7601,
+      "step": 3161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811819020116644e-05,
+      "loss": 0.8682,
+      "step": 3162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805787408224782e-05,
+      "loss": 0.7896,
+      "step": 3163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7997576473547575e-05,
+      "loss": 0.827,
+      "step": 3164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793729740167661e-05,
+      "loss": 0.7601,
+      "step": 3165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787703689323778e-05,
+      "loss": 0.7184,
+      "step": 3166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781679497482575e-05,
+      "loss": 0.7305,
+      "step": 3167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7756571673026883e-05,
+      "loss": 0.7809,
+      "step": 3168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7696367014419396e-05,
+      "loss": 0.7702,
+      "step": 3169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763618102557333e-05,
+      "loss": 0.7794,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757601373305033e-05,
+      "loss": 0.806,
+      "step": 3171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751586516340397e-05,
+      "loss": 0.7789,
+      "step": 3172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7455735343179405e-05,
+      "loss": 0.8276,
+      "step": 3173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7395624298913653e-05,
+      "loss": 0.7943,
+      "step": 3174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733553205713529e-05,
+      "loss": 0.8101,
+      "step": 3175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7275458644364766e-05,
+      "loss": 0.7857,
+      "step": 3176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7215404087114064e-05,
+      "loss": 0.7529,
+      "step": 3177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7155368411887e-05,
+      "loss": 0.7764,
+      "step": 3178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70953516451789e-05,
+      "loss": 0.7791,
+      "step": 3179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703535381347687e-05,
+      "loss": 0.697,
+      "step": 3180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6975374943259596e-05,
+      "loss": 0.7391,
+      "step": 3181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6915415060997365e-05,
+      "loss": 0.7883,
+      "step": 3182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685547419315222e-05,
+      "loss": 0.7041,
+      "step": 3183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679555236617762e-05,
+      "loss": 0.733,
+      "step": 3184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67356496065188e-05,
+      "loss": 0.7444,
+      "step": 3185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667576594061246e-05,
+      "loss": 0.7683,
+      "step": 3186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661590139488695e-05,
+      "loss": 0.7783,
+      "step": 3187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655605599576209e-05,
+      "loss": 0.7355,
+      "step": 3188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649622976964935e-05,
+      "loss": 0.7227,
+      "step": 3189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643642274295168e-05,
+      "loss": 0.8243,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6376634942063625e-05,
+      "loss": 0.7827,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63168663933711e-05,
+      "loss": 0.841,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6257117123251703e-05,
+      "loss": 0.6536,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619738715807437e-05,
+      "loss": 0.7927,
+      "step": 3194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613767652419963e-05,
+      "loss": 0.8005,
+      "step": 3195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6077985247979405e-05,
+      "loss": 0.8031,
+      "step": 3196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601831335575713e-05,
+      "loss": 0.7553,
+      "step": 3197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595866087386763e-05,
+      "loss": 0.8162,
+      "step": 3198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589902782863723e-05,
+      "loss": 0.7905,
+      "step": 3199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583941424638361e-05,
+      "loss": 0.7487,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577982015341586e-05,
+      "loss": 0.8196,
+      "step": 3201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572024557603458e-05,
+      "loss": 0.7735,
+      "step": 3202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566069054053159e-05,
+      "loss": 0.806,
+      "step": 3203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560115507319025e-05,
+      "loss": 0.7503,
+      "step": 3204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5541639200285125e-05,
+      "loss": 0.7773,
+      "step": 3205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5482142948082294e-05,
+      "loss": 0.7604,
+      "step": 3206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5422666342839026e-05,
+      "loss": 0.7778,
+      "step": 3207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536320941080405e-05,
+      "loss": 0.7715,
+      "step": 3208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530377217821729e-05,
+      "loss": 0.8382,
+      "step": 3209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524435467131007e-05,
+      "loss": 0.7604,
+      "step": 3210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518495691630502e-05,
+      "loss": 0.7841,
+      "step": 3211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512557893941592e-05,
+      "loss": 0.8486,
+      "step": 3212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506622076684796e-05,
+      "loss": 0.7542,
+      "step": 3213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500688242479759e-05,
+      "loss": 0.7855,
+      "step": 3214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4947563939452354e-05,
+      "loss": 0.7663,
+      "step": 3215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488826533699123e-05,
+      "loss": 0.8217,
+      "step": 3216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4828986643584254e-05,
+      "loss": 0.7709,
+      "step": 3217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476972788539283e-05,
+      "loss": 0.8151,
+      "step": 3218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471048908856944e-05,
+      "loss": 0.772,
+      "step": 3219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465127027925775e-05,
+      "loss": 0.7788,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4592071483592756e-05,
+      "loss": 0.7849,
+      "step": 3221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4532892727700424e-05,
+      "loss": 0.7138,
+      "step": 3222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447373403769807e-05,
+      "loss": 0.7416,
+      "step": 3223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441459543969396e-05,
+      "loss": 0.7701,
+      "step": 3224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4355476959787675e-05,
+      "loss": 0.8075,
+      "step": 3225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429637862406976e-05,
+      "loss": 0.7101,
+      "step": 3226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423730045862203e-05,
+      "loss": 0.7682,
+      "step": 3227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4178242489517204e-05,
+      "loss": 0.8119,
+      "step": 3228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4119204742819305e-05,
+      "loss": 0.7751,
+      "step": 3229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4060187244583215e-05,
+      "loss": 0.8478,
+      "step": 3230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400119002085504e-05,
+      "loss": 0.8506,
+      "step": 3231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3942213097671914e-05,
+      "loss": 0.7705,
+      "step": 3232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388325650106192e-05,
+      "loss": 0.7666,
+      "step": 3233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382432025704425e-05,
+      "loss": 0.754,
+      "step": 3234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376540439162916e-05,
+      "loss": 0.7801,
+      "step": 3235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370650893081775e-05,
+      "loss": 0.8467,
+      "step": 3236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364763390060228e-05,
+      "loss": 0.7601,
+      "step": 3237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358877932696591e-05,
+      "loss": 0.834,
+      "step": 3238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352994523588275e-05,
+      "loss": 0.8399,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3471131653317985e-05,
+      "loss": 0.7657,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3412338605227586e-05,
+      "loss": 0.709,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3353566117558616e-05,
+      "loss": 0.758,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3294814216248956e-05,
+      "loss": 0.8392,
+      "step": 3243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323608292722748e-05,
+      "loss": 0.8035,
+      "step": 3244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317737227641388e-05,
+      "loss": 0.7718,
+      "step": 3245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311868228971886e-05,
+      "loss": 0.8453,
+      "step": 3246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306001299304384e-05,
+      "loss": 0.8057,
+      "step": 3247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300136441228127e-05,
+      "loss": 0.7021,
+      "step": 3248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294273657331435e-05,
+      "loss": 0.7823,
+      "step": 3249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28841295020172e-05,
+      "loss": 0.8446,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282554322425469e-05,
+      "loss": 0.8094,
+      "step": 3251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276697776588262e-05,
+      "loss": 0.7906,
+      "step": 3252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270843315274748e-05,
+      "loss": 0.751,
+      "step": 3253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264990941068666e-05,
+      "loss": 0.8131,
+      "step": 3254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2591406565528346e-05,
+      "loss": 0.8318,
+      "step": 3255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2532924643091366e-05,
+      "loss": 0.761,
+      "step": 3256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247446366918549e-05,
+      "loss": 0.7962,
+      "step": 3257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241602366961109e-05,
+      "loss": 0.8295,
+      "step": 3258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235760467015944e-05,
+      "loss": 0.7815,
+      "step": 3259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2299206696612356e-05,
+      "loss": 0.7645,
+      "step": 3260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2240829774742564e-05,
+      "loss": 0.8022,
+      "step": 3261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218247393031334e-05,
+      "loss": 0.8039,
+      "step": 3262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212413918907879e-05,
+      "loss": 0.8181,
+      "step": 3263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206582557678359e-05,
+      "loss": 0.7559,
+      "step": 3264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200753311916322e-05,
+      "loss": 0.7384,
+      "step": 3265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1949261841943656e-05,
+      "loss": 0.8327,
+      "step": 3266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1891011770841715e-05,
+      "loss": 0.7946,
+      "step": 3267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183278293156468e-05,
+      "loss": 0.7994,
+      "step": 3268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177457534981063e-05,
+      "loss": 0.8123,
+      "step": 3269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171638905126809e-05,
+      "loss": 0.7013,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165822406161635e-05,
+      "loss": 0.7927,
+      "step": 3271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160008040652519e-05,
+      "loss": 0.7444,
+      "step": 3272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154195811165503e-05,
+      "loss": 0.8547,
+      "step": 3273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148385720265681e-05,
+      "loss": 0.7376,
+      "step": 3274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1425777705172095e-05,
+      "loss": 0.8037,
+      "step": 3275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1367719644833e-05,
+      "loss": 0.7897,
+      "step": 3276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130968304726207e-05,
+      "loss": 0.8639,
+      "step": 3277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125166793807254e-05,
+      "loss": 0.7857,
+      "step": 3278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119367434286801e-05,
+      "loss": 0.8182,
+      "step": 3279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1135702287242734e-05,
+      "loss": 0.7829,
+      "step": 3280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1077751796781295e-05,
+      "loss": 0.8032,
+      "step": 3281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101982289705891e-05,
+      "loss": 0.7359,
+      "step": 3282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0961915613641155e-05,
+      "loss": 0.729,
+      "step": 3283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090402997208417e-05,
+      "loss": 0.7359,
+      "step": 3284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0846165997934413e-05,
+      "loss": 0.7368,
+      "step": 3285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078832371672893e-05,
+      "loss": 0.7892,
+      "step": 3286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0730503153995035e-05,
+      "loss": 0.7468,
+      "step": 3287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067270433525062e-05,
+      "loss": 0.7995,
+      "step": 3288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0614927286003835e-05,
+      "loss": 0.796,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0557172031753355e-05,
+      "loss": 0.6701,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0499438597988093e-05,
+      "loss": 0.7765,
+      "step": 3291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0441727010187464e-05,
+      "loss": 0.79,
+      "step": 3292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038403729382114e-05,
+      "loss": 0.7675,
+      "step": 3293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0326369474349254e-05,
+      "loss": 0.7402,
+      "step": 3294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026872357722211e-05,
+      "loss": 0.7466,
+      "step": 3295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021109962788053e-05,
+      "loss": 0.8112,
+      "step": 3296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015349765175549e-05,
+      "loss": 0.7614,
+      "step": 3297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0095917674268335e-05,
+      "loss": 0.7805,
+      "step": 3298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003835972083078e-05,
+      "loss": 0.7476,
+      "step": 3299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9980823816844616e-05,
+      "loss": 0.7632,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9923309987702136e-05,
+      "loss": 0.809,
+      "step": 3301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986581825878569e-05,
+      "loss": 0.8361,
+      "step": 3302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980834865546805e-05,
+      "loss": 0.7051,
+      "step": 3303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9750901203112064e-05,
+      "loss": 0.701,
+      "step": 3304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969347592707094e-05,
+      "loss": 0.7337,
+      "step": 3305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9636072852687984e-05,
+      "loss": 0.7433,
+      "step": 3306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957869200529682e-05,
+      "loss": 0.8115,
+      "step": 3307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9521333410221126e-05,
+      "loss": 0.8051,
+      "step": 3308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946399709277491e-05,
+      "loss": 0.7491,
+      "step": 3309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9406683078262195e-05,
+      "loss": 0.7157,
+      "step": 3310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9349391391977304e-05,
+      "loss": 0.7876,
+      "step": 3311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929212205920459e-05,
+      "loss": 0.8046,
+      "step": 3312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923487510521866e-05,
+      "loss": 0.7014,
+      "step": 3313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9177650555284116e-05,
+      "loss": 0.7363,
+      "step": 3314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912044843465572e-05,
+      "loss": 0.7161,
+      "step": 3315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9063268768578364e-05,
+      "loss": 0.7085,
+      "step": 3316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900611158228706e-05,
+      "loss": 0.7366,
+      "step": 3317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894897690100679e-05,
+      "loss": 0.7912,
+      "step": 3318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8891864749952685e-05,
+      "loss": 0.6435,
+      "step": 3319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883477515432996e-05,
+      "loss": 0.7731,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8777708139333756e-05,
+      "loss": 0.7633,
+      "step": 3321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8720663730149395e-05,
+      "loss": 0.8093,
+      "step": 3322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866364195195208e-05,
+      "loss": 0.7174,
+      "step": 3323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860664282990716e-05,
+      "loss": 0.7028,
+      "step": 3324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854966638916986e-05,
+      "loss": 0.7873,
+      "step": 3325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849271265488552e-05,
+      "loss": 0.7823,
+      "step": 3326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8435781652189324e-05,
+      "loss": 0.7548,
+      "step": 3327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837887340620658e-05,
+      "loss": 0.7968,
+      "step": 3328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8321987942052385e-05,
+      "loss": 0.7533,
+      "step": 3329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826512528483193e-05,
+      "loss": 0.7726,
+      "step": 3330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820828545964022e-05,
+      "loss": 0.7212,
+      "step": 3331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815146849156231e-05,
+      "loss": 0.8085,
+      "step": 3332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809467440567305e-05,
+      "loss": 0.7274,
+      "step": 3333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8037903227037205e-05,
+      "loss": 0.7903,
+      "step": 3334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798115498070954e-05,
+      "loss": 0.6953,
+      "step": 3335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792442969173455e-05,
+      "loss": 0.7895,
+      "step": 3336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7867727385146745e-05,
+      "loss": 0.8626,
+      "step": 3337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7811048085970346e-05,
+      "loss": 0.776,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775439181921952e-05,
+      "loss": 0.7931,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769775860989829e-05,
+      "loss": 0.8127,
+      "step": 3340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7641148483000384e-05,
+      "loss": 0.8313,
+      "step": 3341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758456146350944e-05,
+      "loss": 0.8122,
+      "step": 3342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752799757639891e-05,
+      "loss": 0.7694,
+      "step": 3343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7471456846631935e-05,
+      "loss": 0.7877,
+      "step": 3344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741493929916157e-05,
+      "loss": 0.7272,
+      "step": 3345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735844495893048e-05,
+      "loss": 0.8285,
+      "step": 3346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7301973850871275e-05,
+      "loss": 0.7649,
+      "step": 3347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724552599990614e-05,
+      "loss": 0.8206,
+      "step": 3348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7189101430947116e-05,
+      "loss": 0.7538,
+      "step": 3349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7132700168895914e-05,
+      "loss": 0.783,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707632223864392e-05,
+      "loss": 0.7569,
+      "step": 3351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7019967665072364e-05,
+      "loss": 0.7556,
+      "step": 3352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696363647305197e-05,
+      "loss": 0.7373,
+      "step": 3353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690732868744333e-05,
+      "loss": 0.7569,
+      "step": 3354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685104433309657e-05,
+      "loss": 0.8116,
+      "step": 3355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67947834348516e-05,
+      "loss": 0.6931,
+      "step": 3356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673854601753781e-05,
+      "loss": 0.8091,
+      "step": 3357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6682332105974426e-05,
+      "loss": 0.7914,
+      "step": 3358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662614172497011e-05,
+      "loss": 0.8486,
+      "step": 3359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656997489932328e-05,
+      "loss": 0.7484,
+      "step": 3360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6513831653821927e-05,
+      "loss": 0.7417,
+      "step": 3361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645771201324355e-05,
+      "loss": 0.8055,
+      "step": 3362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640161600235533e-05,
+      "loss": 0.7938,
+      "step": 3363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634554364591402e-05,
+      "loss": 0.8053,
+      "step": 3364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6289494968665836e-05,
+      "loss": 0.8521,
+      "step": 3365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623346999534668e-05,
+      "loss": 0.7446,
+      "step": 3366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617746875068183e-05,
+      "loss": 0.7205,
+      "step": 3367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6121491259386274e-05,
+      "loss": 0.7274,
+      "step": 3368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606553754616438e-05,
+      "loss": 0.7607,
+      "step": 3369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600960763571003e-05,
+      "loss": 0.7289,
+      "step": 3370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595370155270671e-05,
+      "loss": 0.7535,
+      "step": 3371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589781932182724e-05,
+      "loss": 0.7768,
+      "step": 3372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5841960967734075e-05,
+      "loss": 0.7965,
+      "step": 3373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5786126515078977e-05,
+      "loss": 0.6854,
+      "step": 3374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5730315988503305e-05,
+      "loss": 0.808,
+      "step": 3375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567452941263769e-05,
+      "loss": 0.7485,
+      "step": 3376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5618766812102384e-05,
+      "loss": 0.7698,
+      "step": 3377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5563028211506885e-05,
+      "loss": 0.7578,
+      "step": 3378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5507313635450256e-05,
+      "loss": 0.7913,
+      "step": 3379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545162310852079e-05,
+      "loss": 0.7641,
+      "step": 3380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5395956655296337e-05,
+      "loss": 0.7878,
+      "step": 3381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5340314300343967e-05,
+      "loss": 0.7433,
+      "step": 3382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528469606822021e-05,
+      "loss": 0.7995,
+      "step": 3383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5229101983470954e-05,
+      "loss": 0.7644,
+      "step": 3384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517353207063142e-05,
+      "loss": 0.7026,
+      "step": 3385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5117986354226074e-05,
+      "loss": 0.7791,
+      "step": 3386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506246485876885e-05,
+      "loss": 0.8009,
+      "step": 3387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5006967608762885e-05,
+      "loss": 0.7701,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4951494628700605e-05,
+      "loss": 0.8154,
+      "step": 3389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489604594306386e-05,
+      "loss": 0.7335,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4840621576323585e-05,
+      "loss": 0.7585,
+      "step": 3391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478522155294016e-05,
+      "loss": 0.7819,
+      "step": 3392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4729845897363086e-05,
+      "loss": 0.8217,
+      "step": 3393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467449463403125e-05,
+      "loss": 0.7385,
+      "step": 3394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461916778737259e-05,
+      "loss": 0.8086,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456386538180447e-05,
+      "loss": 0.8384,
+      "step": 3396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450858744173327e-05,
+      "loss": 0.7555,
+      "step": 3397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445333399155478e-05,
+      "loss": 0.7556,
+      "step": 3398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439810505565377e-05,
+      "loss": 0.8203,
+      "step": 3399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434290065840439e-05,
+      "loss": 0.7979,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428772082416977e-05,
+      "loss": 0.8192,
+      "step": 3401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423256557730238e-05,
+      "loss": 0.7288,
+      "step": 3402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417743494214368e-05,
+      "loss": 0.6868,
+      "step": 3403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412232894302437e-05,
+      "loss": 0.7371,
+      "step": 3404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406724760426429e-05,
+      "loss": 0.7683,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401219095017229e-05,
+      "loss": 0.7801,
+      "step": 3406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395715900504647e-05,
+      "loss": 0.7843,
+      "step": 3407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390215179317385e-05,
+      "loss": 0.7899,
+      "step": 3408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384716933883073e-05,
+      "loss": 0.8093,
+      "step": 3409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379221166628229e-05,
+      "loss": 0.8541,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373727879978294e-05,
+      "loss": 0.8124,
+      "step": 3411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368237076357603e-05,
+      "loss": 0.7941,
+      "step": 3412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362748758189404e-05,
+      "loss": 0.7286,
+      "step": 3413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3572629278958364e-05,
+      "loss": 0.7676,
+      "step": 3414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351779587897956e-05,
+      "loss": 0.7574,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346298740615705e-05,
+      "loss": 0.7029,
+      "step": 3416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34082038846794e-05,
+      "loss": 0.7323,
+      "step": 3417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335344533872401e-05,
+      "loss": 0.7529,
+      "step": 3418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3298711792457425e-05,
+      "loss": 0.7748,
+      "step": 3419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324400327003498e-05,
+      "loss": 0.7998,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318931979560115e-05,
+      "loss": 0.8434,
+      "step": 3421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313466139328918e-05,
+      "loss": 0.7777,
+      "step": 3422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308002808722139e-05,
+      "loss": 0.7703,
+      "step": 3423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3025419901508924e-05,
+      "loss": 0.7712,
+      "step": 3424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297083686025193e-05,
+      "loss": 0.746,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291627898753935e-05,
+      "loss": 0.7952,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286174630744911e-05,
+      "loss": 0.7757,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280723884404803e-05,
+      "loss": 0.7395,
+      "step": 3428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2752756621391686e-05,
+      "loss": 0.7526,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2698299663524657e-05,
+      "loss": 0.7793,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2643867994480224e-05,
+      "loss": 0.7859,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258946163828067e-05,
+      "loss": 0.8005,
+      "step": 3432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253508061893695e-05,
+      "loss": 0.7977,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248072496044897e-05,
+      "loss": 0.8145,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2426394686805316e-05,
+      "loss": 0.804,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237208982198351e-05,
+      "loss": 0.7621,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231781038994971e-05,
+      "loss": 0.8172,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226355641465901e-05,
+      "loss": 0.7987,
+      "step": 3438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220932792005512e-05,
+      "loss": 0.7976,
+      "step": 3439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2155124930070634e-05,
+      "loss": 0.7353,
+      "step": 3440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210094746862676e-05,
+      "loss": 0.8274,
+      "step": 3441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2046795559633576e-05,
+      "loss": 0.8171,
+      "step": 3442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199266922698974e-05,
+      "loss": 0.7701,
+      "step": 3443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1938568494582786e-05,
+      "loss": 0.8259,
+      "step": 3444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188449338628878e-05,
+      "loss": 0.7332,
+      "step": 3445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183044392597262e-05,
+      "loss": 0.7563,
+      "step": 3446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177642013748777e-05,
+      "loss": 0.8182,
+      "step": 3447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172242204467645e-05,
+      "loss": 0.8083,
+      "step": 3448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166844967136954e-05,
+      "loss": 0.7464,
+      "step": 3449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1614503041386464e-05,
+      "loss": 0.7894,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1560582178535444e-05,
+      "loss": 0.7167,
+      "step": 3451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1506687106613175e-05,
+      "loss": 0.7732,
+      "step": 3452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1452817849405104e-05,
+      "loss": 0.6221,
+      "step": 3453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139897443068515e-05,
+      "loss": 0.7849,
+      "step": 3454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1345156874215974e-05,
+      "loss": 0.8247,
+      "step": 3455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129136520374868e-05,
+      "loss": 0.7728,
+      "step": 3456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123759944302308e-05,
+      "loss": 0.8575,
+      "step": 3457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118385961576743e-05,
+      "loss": 0.8455,
+      "step": 3458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113014574569865e-05,
+      "loss": 0.6944,
+      "step": 3459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107645785652212e-05,
+      "loss": 0.7981,
+      "step": 3460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102279597193184e-05,
+      "loss": 0.7806,
+      "step": 3461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09691601156102e-05,
+      "loss": 0.8102,
+      "step": 3462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091555031122826e-05,
+      "loss": 0.8033,
+      "step": 3463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0861966582445465e-05,
+      "loss": 0.7442,
+      "step": 3464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0808408952909784e-05,
+      "loss": 0.7349,
+      "step": 3465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0754877446257725e-05,
+      "loss": 0.7076,
+      "step": 3466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070137208611417e-05,
+      "loss": 0.8121,
+      "step": 3467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064789289609252e-05,
+      "loss": 0.8314,
+      "step": 3468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0594439899794635e-05,
+      "loss": 0.769,
+      "step": 3469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054101312081083e-05,
+      "loss": 0.7964,
+      "step": 3470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048761258271976e-05,
+      "loss": 0.8573,
+      "step": 3471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0434238309088615e-05,
+      "loss": 0.7637,
+      "step": 3472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0380890323472854e-05,
+      "loss": 0.7601,
+      "step": 3473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032756864941651e-05,
+      "loss": 0.8009,
+      "step": 3474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027427331045183e-05,
+      "loss": 0.8089,
+      "step": 3475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022100433009958e-05,
+      "loss": 0.8159,
+      "step": 3476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016776173186877e-05,
+      "loss": 0.853,
+      "step": 3477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011454553925691e-05,
+      "loss": 0.8367,
+      "step": 3478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006135577574969e-05,
+      "loss": 0.8015,
+      "step": 3479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000819246482128e-05,
+      "loss": 0.7332,
+      "step": 3480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9955055629934076e-05,
+      "loss": 0.6914,
+      "step": 3481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990194529453889e-05,
+      "loss": 0.8044,
+      "step": 3482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984886148207474e-05,
+      "loss": 0.6815,
+      "step": 3483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979580421596893e-05,
+      "loss": 0.7318,
+      "step": 3484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97427735196372e-05,
+      "loss": 0.7753,
+      "step": 3485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968976941648337e-05,
+      "loss": 0.7509,
+      "step": 3486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963679192989967e-05,
+      "loss": 0.8409,
+      "step": 3487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9583841083266494e-05,
+      "loss": 0.6923,
+      "step": 3488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953091689995252e-05,
+      "loss": 0.7538,
+      "step": 3489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947801940331469e-05,
+      "loss": 0.7314,
+      "step": 3490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942514861669806e-05,
+      "loss": 0.7111,
+      "step": 3491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9372304563436023e-05,
+      "loss": 0.7154,
+      "step": 3492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9319487266850133e-05,
+      "loss": 0.7314,
+      "step": 3493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9266696750250045e-05,
+      "loss": 0.7933,
+      "step": 3494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921393303693376e-05,
+      "loss": 0.7593,
+      "step": 3495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9161196150187286e-05,
+      "loss": 0.8184,
+      "step": 3496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910848611328494e-05,
+      "loss": 0.7843,
+      "step": 3497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905580294948904e-05,
+      "loss": 0.8831,
+      "step": 3498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900314668205019e-05,
+      "loss": 0.8171,
+      "step": 3499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895051733420699e-05,
+      "loss": 0.7325,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889791492918631e-05,
+      "loss": 0.7861,
+      "step": 3501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884533949020297e-05,
+      "loss": 0.7466,
+      "step": 3502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8792791040459966e-05,
+      "loss": 0.7939,
+      "step": 3503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874026960314842e-05,
+      "loss": 0.8072,
+      "step": 3504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868777520144744e-05,
+      "loss": 0.7328,
+      "step": 3505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863530785852433e-05,
+      "loss": 0.7656,
+      "step": 3506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8582867597534286e-05,
+      "loss": 0.6871,
+      "step": 3507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853045444162073e-05,
+      "loss": 0.7411,
+      "step": 3508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847806841391497e-05,
+      "loss": 0.7337,
+      "step": 3509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8425709537536424e-05,
+      "loss": 0.7937,
+      "step": 3510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837337783559256e-05,
+      "loss": 0.7315,
+      "step": 3511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832107333117872e-05,
+      "loss": 0.8062,
+      "step": 3512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826879604737837e-05,
+      "loss": 0.7481,
+      "step": 3513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821654600726296e-05,
+      "loss": 0.7849,
+      "step": 3514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816432323389178e-05,
+      "loss": 0.7611,
+      "step": 3515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8112127750312286e-05,
+      "loss": 0.8288,
+      "step": 3516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805995957955969e-05,
+      "loss": 0.7981,
+      "step": 3517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800781874465733e-05,
+      "loss": 0.8629,
+      "step": 3518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795570526861631e-05,
+      "loss": 0.7729,
+      "step": 3519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7903619174435844e-05,
+      "loss": 0.7409,
+      "step": 3520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78515604851029e-05,
+      "loss": 0.7564,
+      "step": 3521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.77995292235924e-05,
+      "loss": 0.7738,
+      "step": 3522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7747525412867236e-05,
+      "loss": 0.8096,
+      "step": 3523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769554907587806e-05,
+      "loss": 0.7822,
+      "step": 3524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7643600235563535e-05,
+      "loss": 0.7732,
+      "step": 3525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7591678914850045e-05,
+      "loss": 0.7983,
+      "step": 3526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753978513665197e-05,
+      "loss": 0.8187,
+      "step": 3527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748791892387141e-05,
+      "loss": 0.7376,
+      "step": 3528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7436080299398416e-05,
+      "loss": 0.7707,
+      "step": 3529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738426928611073e-05,
+      "loss": 0.7481,
+      "step": 3530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733248590687407e-05,
+      "loss": 0.7626,
+      "step": 3531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728073018454178e-05,
+      "loss": 0.8324,
+      "step": 3532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722900214195515e-05,
+      "loss": 0.8049,
+      "step": 3533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71773018019432e-05,
+      "loss": 0.8046,
+      "step": 3534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712562918732267e-05,
+      "loss": 0.7937,
+      "step": 3535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7073984320898135e-05,
+      "loss": 0.765,
+      "step": 3536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702236722546194e-05,
+      "loss": 0.7998,
+      "step": 3537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697077792379407e-05,
+      "loss": 0.7004,
+      "step": 3538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6919216438662366e-05,
+      "loss": 0.6786,
+      "step": 3539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68676827928223e-05,
+      "loss": 0.784,
+      "step": 3540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681617700901707e-05,
+      "loss": 0.8113,
+      "step": 3541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.676469910997766e-05,
+      "loss": 0.7579,
+      "step": 3542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671324911842263e-05,
+      "loss": 0.8264,
+      "step": 3543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666182705705834e-05,
+      "loss": 0.7656,
+      "step": 3544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661043294857871e-05,
+      "loss": 0.797,
+      "step": 3545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6559066815665434e-05,
+      "loss": 0.7559,
+      "step": 3546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650772868098773e-05,
+      "loss": 0.735,
+      "step": 3547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645641856720263e-05,
+      "loss": 0.7797,
+      "step": 3548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.64051364969546e-05,
+      "loss": 0.7899,
+      "step": 3549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6353882492875944e-05,
+      "loss": 0.7272,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6302656577586366e-05,
+      "loss": 0.725,
+      "step": 3551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625145877369337e-05,
+      "loss": 0.7332,
+      "step": 3552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6200289103791876e-05,
+      "loss": 0.7714,
+      "step": 3553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6149147590464504e-05,
+      "loss": 0.7364,
+      "step": 3554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609803425628145e-05,
+      "loss": 0.8369,
+      "step": 3555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60469491238004e-05,
+      "loss": 0.6547,
+      "step": 3556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.599589221556663e-05,
+      "loss": 0.755,
+      "step": 3557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594486355411302e-05,
+      "loss": 0.7473,
+      "step": 3558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5893863161959887e-05,
+      "loss": 0.7452,
+      "step": 3559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584289106161508e-05,
+      "loss": 0.7736,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579194727557406e-05,
+      "loss": 0.7842,
+      "step": 3561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574103182631965e-05,
+      "loss": 0.7853,
+      "step": 3562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5690144736322325e-05,
+      "loss": 0.7866,
+      "step": 3563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56392860280399e-05,
+      "loss": 0.7463,
+      "step": 3564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558845572391778e-05,
+      "loss": 0.7277,
+      "step": 3565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5537653846388716e-05,
+      "loss": 0.7705,
+      "step": 3566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548688041787306e-05,
+      "loss": 0.7848,
+      "step": 3567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543613546077844e-05,
+      "loss": 0.672,
+      "step": 3568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5385418997500094e-05,
+      "loss": 0.7499,
+      "step": 3569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.533473105042052e-05,
+      "loss": 0.8541,
+      "step": 3570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528407164190979e-05,
+      "loss": 0.7864,
+      "step": 3571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.52334407943252e-05,
+      "loss": 0.7488,
+      "step": 3572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5182838530011644e-05,
+      "loss": 0.7478,
+      "step": 3573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513226487130121e-05,
+      "loss": 0.8116,
+      "step": 3574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508171984051353e-05,
+      "loss": 0.8462,
+      "step": 3575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5031203459955444e-05,
+      "loss": 0.7672,
+      "step": 3576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4980715751921254e-05,
+      "loss": 0.8085,
+      "step": 3577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4930256738692635e-05,
+      "loss": 0.7729,
+      "step": 3578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487982644253846e-05,
+      "loss": 0.8414,
+      "step": 3579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482942488571508e-05,
+      "loss": 0.7541,
+      "step": 3580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4779052090466035e-05,
+      "loss": 0.8245,
+      "step": 3581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472870807902228e-05,
+      "loss": 0.6868,
+      "step": 3582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467839287360197e-05,
+      "loss": 0.7256,
+      "step": 3583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462810649641065e-05,
+      "loss": 0.7998,
+      "step": 3584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4577848969641025e-05,
+      "loss": 0.8094,
+      "step": 3585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45276203154732e-05,
+      "loss": 0.7615,
+      "step": 3586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447742055607438e-05,
+      "loss": 0.8221,
+      "step": 3587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.44272497135992e-05,
+      "loss": 0.7521,
+      "step": 3588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4377107810189344e-05,
+      "loss": 0.8173,
+      "step": 3589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432699486797391e-05,
+      "loss": 0.8043,
+      "step": 3590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4276910909069035e-05,
+      "loss": 0.7678,
+      "step": 3591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422685595557824e-05,
+      "loss": 0.8164,
+      "step": 3592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4176830029592075e-05,
+      "loss": 0.6799,
+      "step": 3593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.412683315318844e-05,
+      "loss": 0.767,
+      "step": 3594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4076865348432265e-05,
+      "loss": 0.7166,
+      "step": 3595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402692663737579e-05,
+      "loss": 0.7933,
+      "step": 3596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397701704205828e-05,
+      "loss": 0.8169,
+      "step": 3597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392713658450625e-05,
+      "loss": 0.6809,
+      "step": 3598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387728528673335e-05,
+      "loss": 0.8026,
+      "step": 3599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382746317074029e-05,
+      "loss": 0.7639,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3777670258514996e-05,
+      "loss": 0.7767,
+      "step": 3601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3727906572032374e-05,
+      "loss": 0.8041,
+      "step": 3602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3678172133254616e-05,
+      "loss": 0.7865,
+      "step": 3603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.36284669641308e-05,
+      "loss": 0.7873,
+      "step": 3604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357879108659728e-05,
+      "loss": 0.7924,
+      "step": 3605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352914452257731e-05,
+      "loss": 0.7093,
+      "step": 3606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347952729398138e-05,
+      "loss": 0.8201,
+      "step": 3607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3429939422706856e-05,
+      "loss": 0.7288,
+      "step": 3608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33803809306383e-05,
+      "loss": 0.8122,
+      "step": 3609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3330851839647195e-05,
+      "loss": 0.7722,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328135217159214e-05,
+      "loss": 0.8033,
+      "step": 3611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3231881948318644e-05,
+      "loss": 0.731,
+      "step": 3612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3182441191659366e-05,
+      "loss": 0.7248,
+      "step": 3613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.313302992343383e-05,
+      "loss": 0.8274,
+      "step": 3614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308364816544857e-05,
+      "loss": 0.7961,
+      "step": 3615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303429593949716e-05,
+      "loss": 0.8416,
+      "step": 3616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2984973267360064e-05,
+      "loss": 0.7927,
+      "step": 3617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293568017080476e-05,
+      "loss": 0.7371,
+      "step": 3618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2886416671585686e-05,
+      "loss": 0.7873,
+      "step": 3619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283718279144411e-05,
+      "loss": 0.7835,
+      "step": 3620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278797855210836e-05,
+      "loss": 0.722,
+      "step": 3621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273880397529363e-05,
+      "loss": 0.8772,
+      "step": 3622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2689659082701974e-05,
+      "loss": 0.8104,
+      "step": 3623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2640543896022444e-05,
+      "loss": 0.6987,
+      "step": 3624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2591458436930866e-05,
+      "loss": 0.7511,
+      "step": 3625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2542402727090074e-05,
+      "loss": 0.8011,
+      "step": 3626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249337678814963e-05,
+      "loss": 0.7598,
+      "step": 3627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244438064174613e-05,
+      "loss": 0.7776,
+      "step": 3628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2395414309502833e-05,
+      "loss": 0.7714,
+      "step": 3629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234647781303002e-05,
+      "loss": 0.7394,
+      "step": 3630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229757117392464e-05,
+      "loss": 0.7737,
+      "step": 3631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22486944137706e-05,
+      "loss": 0.8202,
+      "step": 3632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.219984755413854e-05,
+      "loss": 0.7772,
+      "step": 3633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21510306165859e-05,
+      "loss": 0.7498,
+      "step": 3634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2102243622657e-05,
+      "loss": 0.7175,
+      "step": 3635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205348659388284e-05,
+      "loss": 0.7605,
+      "step": 3636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200475955178128e-05,
+      "loss": 0.7282,
+      "step": 3637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1956062517856875e-05,
+      "loss": 0.7971,
+      "step": 3638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190739551360098e-05,
+      "loss": 0.8015,
+      "step": 3639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185875856049173e-05,
+      "loss": 0.7487,
+      "step": 3640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1810151679993884e-05,
+      "loss": 0.7834,
+      "step": 3641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1761574893559044e-05,
+      "loss": 0.7564,
+      "step": 3642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171302822262551e-05,
+      "loss": 0.814,
+      "step": 3643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16645116886182e-05,
+      "loss": 0.6958,
+      "step": 3644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1616025312948884e-05,
+      "loss": 0.8148,
+      "step": 3645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156756911701585e-05,
+      "loss": 0.7233,
+      "step": 3646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151914312220424e-05,
+      "loss": 0.7602,
+      "step": 3647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14707473498857e-05,
+      "loss": 0.7329,
+      "step": 3648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14223818214187e-05,
+      "loss": 0.8281,
+      "step": 3649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137404655814822e-05,
+      "loss": 0.8161,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1325741581405986e-05,
+      "loss": 0.8094,
+      "step": 3651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127746691251032e-05,
+      "loss": 0.7907,
+      "step": 3652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1229222572766114e-05,
+      "loss": 0.7626,
+      "step": 3653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1181008583465e-05,
+      "loss": 0.7548,
+      "step": 3654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1132824965885067e-05,
+      "loss": 0.8334,
+      "step": 3655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.108467174129115e-05,
+      "loss": 0.783,
+      "step": 3656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103654893093453e-05,
+      "loss": 0.7739,
+      "step": 3657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098845655605319e-05,
+      "loss": 0.7529,
+      "step": 3658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0940394637871574e-05,
+      "loss": 0.7533,
+      "step": 3659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0892363197600794e-05,
+      "loss": 0.8244,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0844362256438366e-05,
+      "loss": 0.7372,
+      "step": 3661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0796391835568496e-05,
+      "loss": 0.7376,
+      "step": 3662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074845195616182e-05,
+      "loss": 0.7527,
+      "step": 3663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0700542639375593e-05,
+      "loss": 0.7487,
+      "step": 3664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065266390635344e-05,
+      "loss": 0.8323,
+      "step": 3665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0604815778225625e-05,
+      "loss": 0.6517,
+      "step": 3666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055699827610881e-05,
+      "loss": 0.7974,
+      "step": 3667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050921142110622e-05,
+      "loss": 0.8404,
+      "step": 3668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0461455234307457e-05,
+      "loss": 0.7349,
+      "step": 3669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0413729736788698e-05,
+      "loss": 0.7681,
+      "step": 3670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036603494961252e-05,
+      "loss": 0.8245,
+      "step": 3671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0318370893827884e-05,
+      "loss": 0.7776,
+      "step": 3672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0270737590470343e-05,
+      "loss": 0.6975,
+      "step": 3673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0223135060561715e-05,
+      "loss": 0.8408,
+      "step": 3674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0175563325110366e-05,
+      "loss": 0.7425,
+      "step": 3675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0128022405110966e-05,
+      "loss": 0.7501,
+      "step": 3676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0080512321544695e-05,
+      "loss": 0.7741,
+      "step": 3677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0033033095379005e-05,
+      "loss": 0.775,
+      "step": 3678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9985584747567853e-05,
+      "loss": 0.8087,
+      "step": 3679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9938167299051423e-05,
+      "loss": 0.7676,
+      "step": 3680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989078077075643e-05,
+      "loss": 0.8153,
+      "step": 3681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9843425183595795e-05,
+      "loss": 0.7951,
+      "step": 3682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9796100558468864e-05,
+      "loss": 0.7664,
+      "step": 3683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974880691626134e-05,
+      "loss": 0.753,
+      "step": 3684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970154427784514e-05,
+      "loss": 0.7555,
+      "step": 3685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965431266407862e-05,
+      "loss": 0.8149,
+      "step": 3686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9607112095806412e-05,
+      "loss": 0.725,
+      "step": 3687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955994259385938e-05,
+      "loss": 0.7082,
+      "step": 3688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9512804179054777e-05,
+      "loss": 0.7755,
+      "step": 3689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.946569687219607e-05,
+      "loss": 0.7758,
+      "step": 3690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9418620694072974e-05,
+      "loss": 0.8193,
+      "step": 3691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937157566546156e-05,
+      "loss": 0.8048,
+      "step": 3692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9324561807124052e-05,
+      "loss": 0.7919,
+      "step": 3693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927757913980903e-05,
+      "loss": 0.7505,
+      "step": 3694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9230627684251156e-05,
+      "loss": 0.7403,
+      "step": 3695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9183707461171495e-05,
+      "loss": 0.7508,
+      "step": 3696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9136818491277152e-05,
+      "loss": 0.688,
+      "step": 3697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9089960795261594e-05,
+      "loss": 0.6792,
+      "step": 3698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904313439380435e-05,
+      "loss": 0.771,
+      "step": 3699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8996339307571286e-05,
+      "loss": 0.7044,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894957555721428e-05,
+      "loss": 0.7694,
+      "step": 3701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890284316337153e-05,
+      "loss": 0.7778,
+      "step": 3702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8856142146667287e-05,
+      "loss": 0.715,
+      "step": 3703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8809472527712046e-05,
+      "loss": 0.7493,
+      "step": 3704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876283432710234e-05,
+      "loss": 0.7707,
+      "step": 3705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871622756542094e-05,
+      "loss": 0.761,
+      "step": 3706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8669652263236667e-05,
+      "loss": 0.7994,
+      "step": 3707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862310844110454e-05,
+      "loss": 0.7971,
+      "step": 3708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85765961195656e-05,
+      "loss": 0.7684,
+      "step": 3709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853011531914699e-05,
+      "loss": 0.7906,
+      "step": 3710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8483666060362024e-05,
+      "loss": 0.7791,
+      "step": 3711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843724836370999e-05,
+      "loss": 0.7964,
+      "step": 3712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8390862249676353e-05,
+      "loss": 0.7017,
+      "step": 3713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834450773873253e-05,
+      "loss": 0.6917,
+      "step": 3714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8298184851336097e-05,
+      "loss": 0.8441,
+      "step": 3715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825189360793058e-05,
+      "loss": 0.7926,
+      "step": 3716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820563402894564e-05,
+      "loss": 0.7642,
+      "step": 3717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815940613479683e-05,
+      "loss": 0.783,
+      "step": 3718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8113209945885865e-05,
+      "loss": 0.767,
+      "step": 3719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806704548260034e-05,
+      "loss": 0.7979,
+      "step": 3720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8020912765313977e-05,
+      "loss": 0.7664,
+      "step": 3721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7974811814386348e-05,
+      "loss": 0.6722,
+      "step": 3722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792874265016313e-05,
+      "loss": 0.7044,
+      "step": 3723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7882705292975876e-05,
+      "loss": 0.7969,
+      "step": 3724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7836699763142182e-05,
+      "loss": 0.7414,
+      "step": 3725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77907260809655e-05,
+      "loss": 0.7238,
+      "step": 3726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7744784266735336e-05,
+      "loss": 0.7453,
+      "step": 3727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76988743407271e-05,
+      "loss": 0.7733,
+      "step": 3728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765299632320204e-05,
+      "loss": 0.7524,
+      "step": 3729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7607150234407463e-05,
+      "loss": 0.7362,
+      "step": 3730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756133609457645e-05,
+      "loss": 0.7349,
+      "step": 3731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7515553923928115e-05,
+      "loss": 0.7115,
+      "step": 3732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746980374266732e-05,
+      "loss": 0.8268,
+      "step": 3733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7424085570984947e-05,
+      "loss": 0.6717,
+      "step": 3734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7378399429057644e-05,
+      "loss": 0.8533,
+      "step": 3735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7332745337047993e-05,
+      "loss": 0.8015,
+      "step": 3736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728712331510438e-05,
+      "loss": 0.7147,
+      "step": 3737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7241533383361105e-05,
+      "loss": 0.7617,
+      "step": 3738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7195975561938215e-05,
+      "loss": 0.7657,
+      "step": 3739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7150449870941673e-05,
+      "loss": 0.8054,
+      "step": 3740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7104956330463184e-05,
+      "loss": 0.8029,
+      "step": 3741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7059494960580368e-05,
+      "loss": 0.8351,
+      "step": 3742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70140657813565e-05,
+      "loss": 0.7542,
+      "step": 3743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6968668812840803e-05,
+      "loss": 0.786,
+      "step": 3744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6923304075068157e-05,
+      "loss": 0.7887,
+      "step": 3745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6877971588059325e-05,
+      "loss": 0.7135,
+      "step": 3746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683267137182074e-05,
+      "loss": 0.7443,
+      "step": 3747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678740344634465e-05,
+      "loss": 0.7245,
+      "step": 3748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6742167831609076e-05,
+      "loss": 0.7697,
+      "step": 3749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6696964547577697e-05,
+      "loss": 0.7562,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6651793614200014e-05,
+      "loss": 0.8226,
+      "step": 3751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6606655051411157e-05,
+      "loss": 0.7957,
+      "step": 3752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.656154887913207e-05,
+      "loss": 0.8302,
+      "step": 3753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65164751172693e-05,
+      "loss": 0.7256,
+      "step": 3754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6471433785715205e-05,
+      "loss": 0.8717,
+      "step": 3755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6426424904347703e-05,
+      "loss": 0.8506,
+      "step": 3756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6381448493030515e-05,
+      "loss": 0.7602,
+      "step": 3757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633650457161292e-05,
+      "loss": 0.7841,
+      "step": 3758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6291593159929973e-05,
+      "loss": 0.7388,
+      "step": 3759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6246714277802252e-05,
+      "loss": 0.7335,
+      "step": 3760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6201867945036106e-05,
+      "loss": 0.8071,
+      "step": 3761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61570541814234e-05,
+      "loss": 0.7955,
+      "step": 3762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6112273006741762e-05,
+      "loss": 0.7885,
+      "step": 3763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6067524440754275e-05,
+      "loss": 0.822,
+      "step": 3764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6022808503209795e-05,
+      "loss": 0.8276,
+      "step": 3765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5978125213842676e-05,
+      "loss": 0.8014,
+      "step": 3766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593347459237283e-05,
+      "loss": 0.8029,
+      "step": 3767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5888856658505868e-05,
+      "loss": 0.7477,
+      "step": 3768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584427143193292e-05,
+      "loss": 0.8077,
+      "step": 3769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579971893233064e-05,
+      "loss": 0.7646,
+      "step": 3770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57551991793613e-05,
+      "loss": 0.7452,
+      "step": 3771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5710712192672725e-05,
+      "loss": 0.8072,
+      "step": 3772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566625799189819e-05,
+      "loss": 0.785,
+      "step": 3773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5621836596656613e-05,
+      "loss": 0.761,
+      "step": 3774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5577448026552343e-05,
+      "loss": 0.7403,
+      "step": 3775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553309230117532e-05,
+      "loss": 0.7628,
+      "step": 3776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5488769440100902e-05,
+      "loss": 0.7294,
+      "step": 3777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5444479462890047e-05,
+      "loss": 0.7201,
+      "step": 3778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540022238908909e-05,
+      "loss": 0.8292,
+      "step": 3779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.535599823822995e-05,
+      "loss": 0.7213,
+      "step": 3780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5311807029829915e-05,
+      "loss": 0.8507,
+      "step": 3781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526764878339185e-05,
+      "loss": 0.7641,
+      "step": 3782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5223523518403936e-05,
+      "loss": 0.768,
+      "step": 3783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5179431254339935e-05,
+      "loss": 0.7717,
+      "step": 3784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5135372010658964e-05,
+      "loss": 0.7694,
+      "step": 3785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5091345806805534e-05,
+      "loss": 0.8168,
+      "step": 3786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504735266220969e-05,
+      "loss": 0.7693,
+      "step": 3787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5003392596286766e-05,
+      "loss": 0.7252,
+      "step": 3788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4959465628437584e-05,
+      "loss": 0.8239,
+      "step": 3789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.491557177804835e-05,
+      "loss": 0.7533,
+      "step": 3790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4871711064490576e-05,
+      "loss": 0.7303,
+      "step": 3791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.482788350712124e-05,
+      "loss": 0.7313,
+      "step": 3792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4784089125282684e-05,
+      "loss": 0.7491,
+      "step": 3793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474032793830251e-05,
+      "loss": 0.7935,
+      "step": 3794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4696599965493806e-05,
+      "loss": 0.7357,
+      "step": 3795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4652905226154866e-05,
+      "loss": 0.7866,
+      "step": 3796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4609243739569453e-05,
+      "loss": 0.7517,
+      "step": 3797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4565615525006523e-05,
+      "loss": 0.7761,
+      "step": 3798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4522020601720474e-05,
+      "loss": 0.7235,
+      "step": 3799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447845898895088e-05,
+      "loss": 0.7855,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4434930705922755e-05,
+      "loss": 0.8413,
+      "step": 3801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439143577184627e-05,
+      "loss": 0.8096,
+      "step": 3802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4347974205917003e-05,
+      "loss": 0.7632,
+      "step": 3803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43045460273157e-05,
+      "loss": 0.7289,
+      "step": 3804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426115125520839e-05,
+      "loss": 0.7539,
+      "step": 3805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4217789908746457e-05,
+      "loss": 0.7574,
+      "step": 3806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.41744620070664e-05,
+      "loss": 0.7805,
+      "step": 3807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4131167569290058e-05,
+      "loss": 0.6957,
+      "step": 3808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4087906614524412e-05,
+      "loss": 0.7795,
+      "step": 3809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4044679161861796e-05,
+      "loss": 0.7786,
+      "step": 3810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4001485230379585e-05,
+      "loss": 0.8131,
+      "step": 3811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39583248391405e-05,
+      "loss": 0.7597,
+      "step": 3812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391519800719245e-05,
+      "loss": 0.7731,
+      "step": 3813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3872104753568424e-05,
+      "loss": 0.7828,
+      "step": 3814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3829045097286706e-05,
+      "loss": 0.771,
+      "step": 3815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3786019057350728e-05,
+      "loss": 0.8161,
+      "step": 3816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3743026652749025e-05,
+      "loss": 0.7905,
+      "step": 3817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370006790245538e-05,
+      "loss": 0.7045,
+      "step": 3818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3657142825428626e-05,
+      "loss": 0.7749,
+      "step": 3819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3614251440612822e-05,
+      "loss": 0.8277,
+      "step": 3820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3571393766937122e-05,
+      "loss": 0.7417,
+      "step": 3821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352856982331576e-05,
+      "loss": 0.7079,
+      "step": 3822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3485779628648176e-05,
+      "loss": 0.7926,
+      "step": 3823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3443023201818802e-05,
+      "loss": 0.7513,
+      "step": 3824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340030056169731e-05,
+      "loss": 0.7706,
+      "step": 3825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3357611727138295e-05,
+      "loss": 0.7579,
+      "step": 3826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3314956716981605e-05,
+      "loss": 0.841,
+      "step": 3827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3272335550051995e-05,
+      "loss": 0.7929,
+      "step": 3828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3229748245159433e-05,
+      "loss": 0.7558,
+      "step": 3829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31871948210988e-05,
+      "loss": 0.6642,
+      "step": 3830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3144675296650166e-05,
+      "loss": 0.748,
+      "step": 3831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3102189690578512e-05,
+      "loss": 0.8395,
+      "step": 3832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3059738021633946e-05,
+      "loss": 0.7419,
+      "step": 3833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301732030855157e-05,
+      "loss": 0.8179,
+      "step": 3834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2974936570051453e-05,
+      "loss": 0.8004,
+      "step": 3835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2932586824838732e-05,
+      "loss": 0.7751,
+      "step": 3836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2890271091603544e-05,
+      "loss": 0.7785,
+      "step": 3837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284798938902094e-05,
+      "loss": 0.7579,
+      "step": 3838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2805741735751053e-05,
+      "loss": 0.7723,
+      "step": 3839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2763528150438916e-05,
+      "loss": 0.8281,
+      "step": 3840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2721348651714512e-05,
+      "loss": 0.719,
+      "step": 3841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2679203258192894e-05,
+      "loss": 0.7704,
+      "step": 3842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2637091988473913e-05,
+      "loss": 0.8041,
+      "step": 3843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2595014861142505e-05,
+      "loss": 0.7433,
+      "step": 3844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2552971894768393e-05,
+      "loss": 0.835,
+      "step": 3845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251096310790638e-05,
+      "loss": 0.7507,
+      "step": 3846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2468988519096036e-05,
+      "loss": 0.7963,
+      "step": 3847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2427048146861952e-05,
+      "loss": 0.7659,
+      "step": 3848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238514200971352e-05,
+      "loss": 0.7407,
+      "step": 3849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2343270126145155e-05,
+      "loss": 0.7381,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230143251463599e-05,
+      "loss": 0.77,
+      "step": 3851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225962919365019e-05,
+      "loss": 0.8268,
+      "step": 3852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221786018163665e-05,
+      "loss": 0.7681,
+      "step": 3853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2176125497029244e-05,
+      "loss": 0.7668,
+      "step": 3854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2134425158246596e-05,
+      "loss": 0.8084,
+      "step": 3855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2092759183692248e-05,
+      "loss": 0.7661,
+      "step": 3856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205112759175455e-05,
+      "loss": 0.7511,
+      "step": 3857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.200953040080663e-05,
+      "loss": 0.7387,
+      "step": 3858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1967967629206543e-05,
+      "loss": 0.7615,
+      "step": 3859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1926439295297018e-05,
+      "loss": 0.7161,
+      "step": 3860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188494541740572e-05,
+      "loss": 0.78,
+      "step": 3861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1843486013844984e-05,
+      "loss": 0.7649,
+      "step": 3862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1802061102912042e-05,
+      "loss": 0.7198,
+      "step": 3863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1760670702888798e-05,
+      "loss": 0.7965,
+      "step": 3864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1719314832042025e-05,
+      "loss": 0.7408,
+      "step": 3865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1677993508623172e-05,
+      "loss": 0.7662,
+      "step": 3866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163670675086853e-05,
+      "loss": 0.7401,
+      "step": 3867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159545457699903e-05,
+      "loss": 0.7257,
+      "step": 3868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1554237005220446e-05,
+      "loss": 0.8055,
+      "step": 3869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151305405372318e-05,
+      "loss": 0.7458,
+      "step": 3870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1471905740682452e-05,
+      "loss": 0.7325,
+      "step": 3871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14307920842581e-05,
+      "loss": 0.8363,
+      "step": 3872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1389713102594777e-05,
+      "loss": 0.716,
+      "step": 3873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1348668813821714e-05,
+      "loss": 0.7472,
+      "step": 3874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1307659236052935e-05,
+      "loss": 0.7656,
+      "step": 3875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126668438738705e-05,
+      "loss": 0.8195,
+      "step": 3876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1225744285907423e-05,
+      "loss": 0.7215,
+      "step": 3877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1184838949682063e-05,
+      "loss": 0.786,
+      "step": 3878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114396839676359e-05,
+      "loss": 0.765,
+      "step": 3879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1103132645189362e-05,
+      "loss": 0.7855,
+      "step": 3880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1062331712981243e-05,
+      "loss": 0.7355,
+      "step": 3881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1021565618145898e-05,
+      "loss": 0.8221,
+      "step": 3882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098083437867445e-05,
+      "loss": 0.7226,
+      "step": 3883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0940138012542786e-05,
+      "loss": 0.7583,
+      "step": 3884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0899476537711294e-05,
+      "loss": 0.7257,
+      "step": 3885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0858849972125038e-05,
+      "loss": 0.7208,
+      "step": 3886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0818258333713592e-05,
+      "loss": 0.7148,
+      "step": 3887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0777701640391235e-05,
+      "loss": 0.8192,
+      "step": 3888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0737179910056692e-05,
+      "loss": 0.8003,
+      "step": 3889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0696693160593373e-05,
+      "loss": 0.7466,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065624140986915e-05,
+      "loss": 0.8193,
+      "step": 3891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0615824675736552e-05,
+      "loss": 0.8176,
+      "step": 3892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0575442976032543e-05,
+      "loss": 0.7263,
+      "step": 3893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.053509632857874e-05,
+      "loss": 0.8004,
+      "step": 3894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0494784751181173e-05,
+      "loss": 0.7546,
+      "step": 3895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045450826163051e-05,
+      "loss": 0.7545,
+      "step": 3896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0414266877701827e-05,
+      "loss": 0.7851,
+      "step": 3897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037406061715481e-05,
+      "loss": 0.7002,
+      "step": 3898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0333889497733526e-05,
+      "loss": 0.7835,
+      "step": 3899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029375353716666e-05,
+      "loss": 0.8349,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0253652753167308e-05,
+      "loss": 0.7097,
+      "step": 3901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0213587163433022e-05,
+      "loss": 0.8582,
+      "step": 3902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0173556785645897e-05,
+      "loss": 0.734,
+      "step": 3903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0133561637472398e-05,
+      "loss": 0.6972,
+      "step": 3904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009360173656354e-05,
+      "loss": 0.7287,
+      "step": 3905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0053677100554667e-05,
+      "loss": 0.7526,
+      "step": 3906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0013787747065694e-05,
+      "loss": 0.7161,
+      "step": 3907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997393369370083e-05,
+      "loss": 0.8209,
+      "step": 3908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993411495804882e-05,
+      "loss": 0.7577,
+      "step": 3909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9894331557682733e-05,
+      "loss": 0.8579,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985458351016012e-05,
+      "loss": 0.73,
+      "step": 3911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9814870833022836e-05,
+      "loss": 0.7745,
+      "step": 3912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9775193543797254e-05,
+      "loss": 0.7802,
+      "step": 3913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973555165999399e-05,
+      "loss": 0.772,
+      "step": 3914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9695945199108158e-05,
+      "loss": 0.7797,
+      "step": 3915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9656374178619153e-05,
+      "loss": 0.7674,
+      "step": 3916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961683861599074e-05,
+      "loss": 0.8578,
+      "step": 3917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9577338528671074e-05,
+      "loss": 0.7125,
+      "step": 3918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9537873934092675e-05,
+      "loss": 0.7119,
+      "step": 3919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949844484967228e-05,
+      "loss": 0.7859,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945905129281108e-05,
+      "loss": 0.7284,
+      "step": 3921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9419693280894557e-05,
+      "loss": 0.7888,
+      "step": 3922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938037083129245e-05,
+      "loss": 0.7352,
+      "step": 3923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9341083961358865e-05,
+      "loss": 0.8398,
+      "step": 3924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9301832688432153e-05,
+      "loss": 0.7273,
+      "step": 3925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926261702983504e-05,
+      "loss": 0.727,
+      "step": 3926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922343700287441e-05,
+      "loss": 0.7014,
+      "step": 3927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9184292624841548e-05,
+      "loss": 0.8056,
+      "step": 3928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914518391301191e-05,
+      "loss": 0.8125,
+      "step": 3929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910611088464529e-05,
+      "loss": 0.8032,
+      "step": 3930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906707355698564e-05,
+      "loss": 0.7054,
+      "step": 3931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902807194726126e-05,
+      "loss": 0.7783,
+      "step": 3932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8989106072684603e-05,
+      "loss": 0.8293,
+      "step": 3933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895017595045243e-05,
+      "loss": 0.7566,
+      "step": 3934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8911281597745646e-05,
+      "loss": 0.7811,
+      "step": 3935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8872423031729382e-05,
+      "loss": 0.7328,
+      "step": 3936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8833600269553052e-05,
+      "loss": 0.8196,
+      "step": 3937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8794813328350147e-05,
+      "loss": 0.7469,
+      "step": 3938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875606222523848e-05,
+      "loss": 0.7263,
+      "step": 3939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8717346977319927e-05,
+      "loss": 0.7875,
+      "step": 3940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8678667601680622e-05,
+      "loss": 0.8012,
+      "step": 3941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864002411539084e-05,
+      "loss": 0.6835,
+      "step": 3942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8601416535505056e-05,
+      "loss": 0.7905,
+      "step": 3943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8562844879061792e-05,
+      "loss": 0.7869,
+      "step": 3944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.852430916308384e-05,
+      "loss": 0.7553,
+      "step": 3945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8485809404578036e-05,
+      "loss": 0.7655,
+      "step": 3946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8447345620535428e-05,
+      "loss": 0.7552,
+      "step": 3947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8408917827931093e-05,
+      "loss": 0.7241,
+      "step": 3948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8370526043724333e-05,
+      "loss": 0.8014,
+      "step": 3949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8332170284858452e-05,
+      "loss": 0.7391,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.829385056826095e-05,
+      "loss": 0.7898,
+      "step": 3951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8255566910843336e-05,
+      "loss": 0.7421,
+      "step": 3952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8217319329501293e-05,
+      "loss": 0.6977,
+      "step": 3953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8179107841114507e-05,
+      "loss": 0.8245,
+      "step": 3954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8140932462546724e-05,
+      "loss": 0.833,
+      "step": 3955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8102793210645862e-05,
+      "loss": 0.7822,
+      "step": 3956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8064690102243774e-05,
+      "loss": 0.6932,
+      "step": 3957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8026623154156463e-05,
+      "loss": 0.7394,
+      "step": 3958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7988592383183865e-05,
+      "loss": 0.7967,
+      "step": 3959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.795059780611007e-05,
+      "loss": 0.7719,
+      "step": 3960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7912639439703082e-05,
+      "loss": 0.792,
+      "step": 3961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7874717300714992e-05,
+      "loss": 0.8007,
+      "step": 3962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7836831405881915e-05,
+      "loss": 0.69,
+      "step": 3963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7798981771923905e-05,
+      "loss": 0.8494,
+      "step": 3964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776116841554507e-05,
+      "loss": 0.719,
+      "step": 3965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7723391353433505e-05,
+      "loss": 0.7249,
+      "step": 3966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7685650602261228e-05,
+      "loss": 0.7828,
+      "step": 3967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7647946178684318e-05,
+      "loss": 0.7212,
+      "step": 3968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7610278099342738e-05,
+      "loss": 0.7429,
+      "step": 3969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757264638086049e-05,
+      "loss": 0.7166,
+      "step": 3970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7535051039845463e-05,
+      "loss": 0.8126,
+      "step": 3971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7497492092889534e-05,
+      "loss": 0.7768,
+      "step": 3972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745996955656851e-05,
+      "loss": 0.7469,
+      "step": 3973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7422483447442083e-05,
+      "loss": 0.7154,
+      "step": 3974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738503378205396e-05,
+      "loss": 0.8,
+      "step": 3975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7347620576931655e-05,
+      "loss": 0.8215,
+      "step": 3976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7310243848586717e-05,
+      "loss": 0.7878,
+      "step": 3977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7272903613514468e-05,
+      "loss": 0.7118,
+      "step": 3978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723559988819422e-05,
+      "loss": 0.7923,
+      "step": 3979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7198332689089114e-05,
+      "loss": 0.6905,
+      "step": 3980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7161102032646225e-05,
+      "loss": 0.7327,
+      "step": 3981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7123907935296436e-05,
+      "loss": 0.8322,
+      "step": 3982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086750413454556e-05,
+      "loss": 0.7566,
+      "step": 3983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7049629483519193e-05,
+      "loss": 0.8212,
+      "step": 3984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701254516187286e-05,
+      "loss": 0.7619,
+      "step": 3985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6975497464881896e-05,
+      "loss": 0.7902,
+      "step": 3986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.693848640889649e-05,
+      "loss": 0.8317,
+      "step": 3987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6901512010250596e-05,
+      "loss": 0.8104,
+      "step": 3988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686457428526208e-05,
+      "loss": 0.7617,
+      "step": 3989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6827673250232544e-05,
+      "loss": 0.6968,
+      "step": 3990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6790808921447476e-05,
+      "loss": 0.6908,
+      "step": 3991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675398131517609e-05,
+      "loss": 0.7593,
+      "step": 3992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6717190447671426e-05,
+      "loss": 0.7694,
+      "step": 3993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6680436335170323e-05,
+      "loss": 0.7467,
+      "step": 3994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6643718993893363e-05,
+      "loss": 0.7056,
+      "step": 3995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6607038440044976e-05,
+      "loss": 0.7689,
+      "step": 3996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6570394689813227e-05,
+      "loss": 0.7819,
+      "step": 3997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653378775937009e-05,
+      "loss": 0.7974,
+      "step": 3998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6497217664871144e-05,
+      "loss": 0.7663,
+      "step": 3999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6460684422455842e-05,
+      "loss": 0.8099,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.642418804824727e-05,
+      "loss": 0.769,
+      "step": 4001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387728558352324e-05,
+      "loss": 0.708,
+      "step": 4002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635130596886153e-05,
+      "loss": 0.761,
+      "step": 4003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6314920295849255e-05,
+      "loss": 0.7595,
+      "step": 4004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6278571555373423e-05,
+      "loss": 0.7837,
+      "step": 4005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6242259763475786e-05,
+      "loss": 0.7432,
+      "step": 4006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6205984936181752e-05,
+      "loss": 0.7388,
+      "step": 4007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.616974708950034e-05,
+      "loss": 0.7274,
+      "step": 4008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613354623942438e-05,
+      "loss": 0.7487,
+      "step": 4009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6097382401930283e-05,
+      "loss": 0.7923,
+      "step": 4010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6061255592978163e-05,
+      "loss": 0.8091,
+      "step": 4011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6025165828511723e-05,
+      "loss": 0.7799,
+      "step": 4012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5989113124458442e-05,
+      "loss": 0.7702,
+      "step": 4013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.595309749672933e-05,
+      "loss": 0.7613,
+      "step": 4014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59171189612191e-05,
+      "loss": 0.7492,
+      "step": 4015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5881177533806045e-05,
+      "loss": 0.7695,
+      "step": 4016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5845273230352164e-05,
+      "loss": 0.7374,
+      "step": 4017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5809406066702948e-05,
+      "loss": 0.7111,
+      "step": 4018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5773576058687635e-05,
+      "loss": 0.7353,
+      "step": 4019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573778322211893e-05,
+      "loss": 0.742,
+      "step": 4020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5702027572793254e-05,
+      "loss": 0.7717,
+      "step": 4021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566630912649052e-05,
+      "loss": 0.7966,
+      "step": 4022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5630627898974315e-05,
+      "loss": 0.7679,
+      "step": 4023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.55949839059917e-05,
+      "loss": 0.8281,
+      "step": 4024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5559377163273392e-05,
+      "loss": 0.7776,
+      "step": 4025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5523807686533588e-05,
+      "loss": 0.8373,
+      "step": 4026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5488275491470116e-05,
+      "loss": 0.6608,
+      "step": 4027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5452780593764317e-05,
+      "loss": 0.7558,
+      "step": 4028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5417323009081032e-05,
+      "loss": 0.7419,
+      "step": 4029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538190275306872e-05,
+      "loss": 0.7798,
+      "step": 4030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5346519841359275e-05,
+      "loss": 0.7494,
+      "step": 4031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5311174289568197e-05,
+      "loss": 0.7904,
+      "step": 4032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.52758661132944e-05,
+      "loss": 0.7063,
+      "step": 4033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5240595328120422e-05,
+      "loss": 0.8042,
+      "step": 4034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5205361949612163e-05,
+      "loss": 0.7271,
+      "step": 4035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5170165993319153e-05,
+      "loss": 0.7912,
+      "step": 4036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135007474774277e-05,
+      "loss": 0.7155,
+      "step": 4037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5099886409494014e-05,
+      "loss": 0.8358,
+      "step": 4038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5064802812978219e-05,
+      "loss": 0.8001,
+      "step": 4039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5029756700710295e-05,
+      "loss": 0.7583,
+      "step": 4040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994748088157007e-05,
+      "loss": 0.7893,
+      "step": 4041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.495977699076867e-05,
+      "loss": 0.7663,
+      "step": 4042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4924843423978963e-05,
+      "loss": 0.7731,
+      "step": 4043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889947403205062e-05,
+      "loss": 0.7175,
+      "step": 4044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855088943847495e-05,
+      "loss": 0.8021,
+      "step": 4045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4820268061290332e-05,
+      "loss": 0.7488,
+      "step": 4046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4785484770900925e-05,
+      "loss": 0.7832,
+      "step": 4047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4750739088030151e-05,
+      "loss": 0.7286,
+      "step": 4048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.47160310280122e-05,
+      "loss": 0.7173,
+      "step": 4049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4681360606164707e-05,
+      "loss": 0.8404,
+      "step": 4050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4646727837788721e-05,
+      "loss": 0.8292,
+      "step": 4051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4612132738168583e-05,
+      "loss": 0.759,
+      "step": 4052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577575322572124e-05,
+      "loss": 0.7487,
+      "step": 4053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4543055606250434e-05,
+      "loss": 0.7276,
+      "step": 4054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4508573604438059e-05,
+      "loss": 0.7507,
+      "step": 4055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4474129332352815e-05,
+      "loss": 0.791,
+      "step": 4056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4439722805195966e-05,
+      "loss": 0.7543,
+      "step": 4057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4405354038151997e-05,
+      "loss": 0.711,
+      "step": 4058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4371023046388853e-05,
+      "loss": 0.7841,
+      "step": 4059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4336729845057706e-05,
+      "loss": 0.7528,
+      "step": 4060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4302474449293136e-05,
+      "loss": 0.7402,
+      "step": 4061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4268256874212937e-05,
+      "loss": 0.8081,
+      "step": 4062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4234077134918322e-05,
+      "loss": 0.7196,
+      "step": 4063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4199935246493723e-05,
+      "loss": 0.7418,
+      "step": 4064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4165831224006932e-05,
+      "loss": 0.7477,
+      "step": 4065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4131765082508963e-05,
+      "loss": 0.7302,
+      "step": 4066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4097736837034137e-05,
+      "loss": 0.7733,
+      "step": 4067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4063746502600072e-05,
+      "loss": 0.7191,
+      "step": 4068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4029794094207671e-05,
+      "loss": 0.7697,
+      "step": 4069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995879626841014e-05,
+      "loss": 0.8015,
+      "step": 4070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962003115467537e-05,
+      "loss": 0.7448,
+      "step": 4071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3928164575037872e-05,
+      "loss": 0.7947,
+      "step": 4072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3894364020485873e-05,
+      "loss": 0.7823,
+      "step": 4073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3860601466728707e-05,
+      "loss": 0.7548,
+      "step": 4074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3826876928666665e-05,
+      "loss": 0.7506,
+      "step": 4075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3793190421183377e-05,
+      "loss": 0.7631,
+      "step": 4076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3759541959145572e-05,
+      "loss": 0.8414,
+      "step": 4077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3725931557403282e-05,
+      "loss": 0.7099,
+      "step": 4078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3692359230789675e-05,
+      "loss": 0.7839,
+      "step": 4079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3658824994121189e-05,
+      "loss": 0.7822,
+      "step": 4080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3625328862197363e-05,
+      "loss": 0.7775,
+      "step": 4081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3591870849801014e-05,
+      "loss": 0.7494,
+      "step": 4082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3558450971698022e-05,
+      "loss": 0.7777,
+      "step": 4083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3525069242637577e-05,
+      "loss": 0.7483,
+      "step": 4084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3491725677351907e-05,
+      "loss": 0.7679,
+      "step": 4085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3458420290556439e-05,
+      "loss": 0.6973,
+      "step": 4086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3425153096949817e-05,
+      "loss": 0.7333,
+      "step": 4087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3391924111213705e-05,
+      "loss": 0.7756,
+      "step": 4088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3358733348013019e-05,
+      "loss": 0.7396,
+      "step": 4089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3325580821995731e-05,
+      "loss": 0.7826,
+      "step": 4090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.329246654779297e-05,
+      "loss": 0.765,
+      "step": 4091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3259390540019023e-05,
+      "loss": 0.798,
+      "step": 4092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3226352813271181e-05,
+      "loss": 0.8014,
+      "step": 4093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3193353382129937e-05,
+      "loss": 0.7491,
+      "step": 4094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3160392261158872e-05,
+      "loss": 0.7393,
+      "step": 4095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127469464904597e-05,
+      "loss": 0.7516,
+      "step": 4096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3094585007896887e-05,
+      "loss": 0.7253,
+      "step": 4097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061738904648524e-05,
+      "loss": 0.7725,
+      "step": 4098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3028931169655446e-05,
+      "loss": 0.6878,
+      "step": 4099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2996161817396568e-05,
+      "loss": 0.7899,
+      "step": 4100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2963430862333948e-05,
+      "loss": 0.7601,
+      "step": 4101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2930738318912616e-05,
+      "loss": 0.7891,
+      "step": 4102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2898084201560745e-05,
+      "loss": 0.791,
+      "step": 4103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2865468524689462e-05,
+      "loss": 0.7381,
+      "step": 4104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2832891302692972e-05,
+      "loss": 0.7387,
+      "step": 4105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2800352549948513e-05,
+      "loss": 0.7812,
+      "step": 4106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2767852280816317e-05,
+      "loss": 0.8075,
+      "step": 4107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2735390509639688e-05,
+      "loss": 0.7852,
+      "step": 4108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2702967250744858e-05,
+      "loss": 0.7333,
+      "step": 4109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2670582518441131e-05,
+      "loss": 0.8182,
+      "step": 4110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263823632702077e-05,
+      "loss": 0.7395,
+      "step": 4111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260592869075905e-05,
+      "loss": 0.7664,
+      "step": 4112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2573659623914236e-05,
+      "loss": 0.7046,
+      "step": 4113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2541429140727534e-05,
+      "loss": 0.8055,
+      "step": 4114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2509237255423144e-05,
+      "loss": 0.8113,
+      "step": 4115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2477083982208282e-05,
+      "loss": 0.7604,
+      "step": 4116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2444969335273015e-05,
+      "loss": 0.7072,
+      "step": 4117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2412893328790475e-05,
+      "loss": 0.7406,
+      "step": 4118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238085597691664e-05,
+      "loss": 0.7404,
+      "step": 4119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348857293790527e-05,
+      "loss": 0.7667,
+      "step": 4120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2316897293533992e-05,
+      "loss": 0.7332,
+      "step": 4121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.22849759902519e-05,
+      "loss": 0.8297,
+      "step": 4122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2253093398032e-05,
+      "loss": 0.7949,
+      "step": 4123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2221249530944922e-05,
+      "loss": 0.8011,
+      "step": 4124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2189444403044293e-05,
+      "loss": 0.735,
+      "step": 4125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215767802836656e-05,
+      "loss": 0.7842,
+      "step": 4126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2125950420931131e-05,
+      "loss": 0.7658,
+      "step": 4127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2094261594740242e-05,
+      "loss": 0.7623,
+      "step": 4128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.206261156377909e-05,
+      "loss": 0.7757,
+      "step": 4129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2031000342015652e-05,
+      "loss": 0.7922,
+      "step": 4130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1999427943400886e-05,
+      "loss": 0.8198,
+      "step": 4131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1967894381868527e-05,
+      "loss": 0.7727,
+      "step": 4132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.193639967133523e-05,
+      "loss": 0.7613,
+      "step": 4133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1904943825700466e-05,
+      "loss": 0.752,
+      "step": 4134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1873526858846562e-05,
+      "loss": 0.7447,
+      "step": 4135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1842148784638717e-05,
+      "loss": 0.753,
+      "step": 4136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1810809616924923e-05,
+      "loss": 0.7233,
+      "step": 4137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1779509369536012e-05,
+      "loss": 0.7886,
+      "step": 4138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1748248056285693e-05,
+      "loss": 0.7463,
+      "step": 4139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1717025690970384e-05,
+      "loss": 0.7567,
+      "step": 4140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1685842287369431e-05,
+      "loss": 0.7855,
+      "step": 4141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1654697859244911e-05,
+      "loss": 0.7929,
+      "step": 4142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.16235924203417e-05,
+      "loss": 0.8253,
+      "step": 4143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1592525984387515e-05,
+      "loss": 0.7016,
+      "step": 4144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156149856509281e-05,
+      "loss": 0.8046,
+      "step": 4145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1530510176150866e-05,
+      "loss": 0.7733,
+      "step": 4146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1499560831237688e-05,
+      "loss": 0.7753,
+      "step": 4147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468650544012105e-05,
+      "loss": 0.7086,
+      "step": 4148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437779328115639e-05,
+      "loss": 0.7749,
+      "step": 4149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1406947197172668e-05,
+      "loss": 0.7075,
+      "step": 4150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137615416479021e-05,
+      "loss": 0.7771,
+      "step": 4151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1345400244558114e-05,
+      "loss": 0.7501,
+      "step": 4152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1314685450048913e-05,
+      "loss": 0.8099,
+      "step": 4153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1284009794817918e-05,
+      "loss": 0.8123,
+      "step": 4154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1253373292403103e-05,
+      "loss": 0.7437,
+      "step": 4155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1222775956325226e-05,
+      "loss": 0.7644,
+      "step": 4156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192217800087768e-05,
+      "loss": 0.7543,
+      "step": 4157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1161698837176826e-05,
+      "loss": 0.7807,
+      "step": 4158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1131219081061295e-05,
+      "loss": 0.7561,
+      "step": 4159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1100778545192747e-05,
+      "loss": 0.797,
+      "step": 4160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070377243005415e-05,
+      "loss": 0.7318,
+      "step": 4161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104001518791622e-05,
+      "loss": 0.7566,
+      "step": 4162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10096923933248e-05,
+      "loss": 0.7673,
+      "step": 4163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0979408872613405e-05,
+      "loss": 0.7762,
+      "step": 4164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0949164639147037e-05,
+      "loss": 0.7275,
+      "step": 4165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0918959706273269e-05,
+      "loss": 0.6871,
+      "step": 4166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0888794087322407e-05,
+      "loss": 0.7154,
+      "step": 4167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0858667795607335e-05,
+      "loss": 0.7435,
+      "step": 4168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0828580844423652e-05,
+      "loss": 0.7513,
+      "step": 4169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0798533247049536e-05,
+      "loss": 0.7693,
+      "step": 4170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0768525016745833e-05,
+      "loss": 0.794,
+      "step": 4171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738556166755997e-05,
+      "loss": 0.7312,
+      "step": 4172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0708626710306136e-05,
+      "loss": 0.6817,
+      "step": 4173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0678736660604894e-05,
+      "loss": 0.726,
+      "step": 4174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064888603084362e-05,
+      "loss": 0.774,
+      "step": 4175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0619074834196197e-05,
+      "loss": 0.7617,
+      "step": 4176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0589303083819147e-05,
+      "loss": 0.7568,
+      "step": 4177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0559570792851537e-05,
+      "loss": 0.7786,
+      "step": 4178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.052987797441507e-05,
+      "loss": 0.7647,
+      "step": 4179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0500224641614031e-05,
+      "loss": 0.7265,
+      "step": 4180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470610807535198e-05,
+      "loss": 0.7978,
+      "step": 4181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0441036485248024e-05,
+      "loss": 0.7578,
+      "step": 4182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0411501687804437e-05,
+      "loss": 0.8065,
+      "step": 4183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0382006428238999e-05,
+      "loss": 0.7545,
+      "step": 4184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0352550719568732e-05,
+      "loss": 0.7303,
+      "step": 4185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0323134574793303e-05,
+      "loss": 0.7966,
+      "step": 4186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293758006894816e-05,
+      "loss": 0.6583,
+      "step": 4187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0264421028838023e-05,
+      "loss": 0.7875,
+      "step": 4188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.023512365357009e-05,
+      "loss": 0.7758,
+      "step": 4189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0205865894020795e-05,
+      "loss": 0.7483,
+      "step": 4190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0176647763102354e-05,
+      "loss": 0.7966,
+      "step": 4191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147469273709587e-05,
+      "loss": 0.825,
+      "step": 4192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0118330438719714e-05,
+      "loss": 0.7094,
+      "step": 4193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.008923127099256e-05,
+      "loss": 0.7941,
+      "step": 4194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0060171783370342e-05,
+      "loss": 0.7688,
+      "step": 4195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031151988677856e-05,
+      "loss": 0.8156,
+      "step": 4196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0002171899722313e-05,
+      "loss": 0.735,
+      "step": 4197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.973231529293436e-06,
+      "loss": 0.776,
+      "step": 4198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94433089016339e-06,
+      "loss": 0.755,
+      "step": 4199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.915469995086846e-06,
+      "loss": 0.7689,
+      "step": 4200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.886648856800929e-06,
+      "loss": 0.7632,
+      "step": 4201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.857867488025164e-06,
+      "loss": 0.7171,
+      "step": 4202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8291259014616e-06,
+      "loss": 0.791,
+      "step": 4203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.80042410979466e-06,
+      "loss": 0.7132,
+      "step": 4204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.771762125691275e-06,
+      "loss": 0.756,
+      "step": 4205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74313996180073e-06,
+      "loss": 0.8038,
+      "step": 4206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.714557630754806e-06,
+      "loss": 0.8245,
+      "step": 4207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.68601514516766e-06,
+      "loss": 0.7458,
+      "step": 4208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.657512517635891e-06,
+      "loss": 0.8116,
+      "step": 4209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.629049760738462e-06,
+      "loss": 0.7596,
+      "step": 4210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.600626887036835e-06,
+      "loss": 0.8029,
+      "step": 4211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.57224390907474e-06,
+      "loss": 0.7774,
+      "step": 4212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.543900839378428e-06,
+      "loss": 0.8057,
+      "step": 4213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.515597690456435e-06,
+      "loss": 0.7603,
+      "step": 4214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.487334474799758e-06,
+      "loss": 0.7962,
+      "step": 4215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.459111204881699e-06,
+      "loss": 0.7346,
+      "step": 4216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.430927893157993e-06,
+      "loss": 0.8123,
+      "step": 4217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.402784552066712e-06,
+      "loss": 0.6847,
+      "step": 4218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374681194028257e-06,
+      "loss": 0.7302,
+      "step": 4219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.346617831445448e-06,
+      "loss": 0.7933,
+      "step": 4220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.318594476703413e-06,
+      "loss": 0.7484,
+      "step": 4221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290611142169648e-06,
+      "loss": 0.7081,
+      "step": 4222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.262667840193939e-06,
+      "loss": 0.7401,
+      "step": 4223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.234764583108479e-06,
+      "loss": 0.7048,
+      "step": 4224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.206901383227695e-06,
+      "loss": 0.7608,
+      "step": 4225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.179078252848438e-06,
+      "loss": 0.7171,
+      "step": 4226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151295204249777e-06,
+      "loss": 0.7428,
+      "step": 4227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123552249693179e-06,
+      "loss": 0.7908,
+      "step": 4228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09584940142234e-06,
+      "loss": 0.7502,
+      "step": 4229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.068186671663326e-06,
+      "loss": 0.745,
+      "step": 4230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.040564072624435e-06,
+      "loss": 0.7225,
+      "step": 4231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.012981616496308e-06,
+      "loss": 0.7352,
+      "step": 4232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985439315451804e-06,
+      "loss": 0.7935,
+      "step": 4233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957937181646159e-06,
+      "loss": 0.8429,
+      "step": 4234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.930475227216761e-06,
+      "loss": 0.8121,
+      "step": 4235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.903053464283394e-06,
+      "loss": 0.7763,
+      "step": 4236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.875671904948001e-06,
+      "loss": 0.7762,
+      "step": 4237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.848330561294804e-06,
+      "loss": 0.7314,
+      "step": 4238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.821029445390327e-06,
+      "loss": 0.6953,
+      "step": 4239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79376856928329e-06,
+      "loss": 0.7552,
+      "step": 4240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.766547945004656e-06,
+      "loss": 0.8251,
+      "step": 4241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.739367584567682e-06,
+      "loss": 0.7087,
+      "step": 4242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.712227499967773e-06,
+      "loss": 0.7654,
+      "step": 4243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.685127703182605e-06,
+      "loss": 0.8153,
+      "step": 4244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.658068206172087e-06,
+      "loss": 0.7621,
+      "step": 4245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.631049020878302e-06,
+      "loss": 0.7486,
+      "step": 4246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.604070159225574e-06,
+      "loss": 0.7121,
+      "step": 4247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577131633120395e-06,
+      "loss": 0.7876,
+      "step": 4248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.550233454451518e-06,
+      "loss": 0.795,
+      "step": 4249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523375635089814e-06,
+      "loss": 0.7963,
+      "step": 4250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496558186888426e-06,
+      "loss": 0.7766,
+      "step": 4251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.469781121682584e-06,
+      "loss": 0.7788,
+      "step": 4252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443044451289783e-06,
+      "loss": 0.773,
+      "step": 4253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.416348187509626e-06,
+      "loss": 0.816,
+      "step": 4254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.389692342123945e-06,
+      "loss": 0.7462,
+      "step": 4255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363076926896674e-06,
+      "loss": 0.7435,
+      "step": 4256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.336501953573927e-06,
+      "loss": 0.782,
+      "step": 4257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.309967433884003e-06,
+      "loss": 0.7589,
+      "step": 4258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.283473379537287e-06,
+      "loss": 0.8375,
+      "step": 4259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.257019802226362e-06,
+      "loss": 0.7935,
+      "step": 4260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.230606713625888e-06,
+      "loss": 0.7535,
+      "step": 4261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20423412539274e-06,
+      "loss": 0.7706,
+      "step": 4262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.17790204916582e-06,
+      "loss": 0.7508,
+      "step": 4263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15161049656622e-06,
+      "loss": 0.7596,
+      "step": 4264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.125359479197126e-06,
+      "loss": 0.7903,
+      "step": 4265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.099149008643869e-06,
+      "loss": 0.7684,
+      "step": 4266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.072979096473798e-06,
+      "loss": 0.6749,
+      "step": 4267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.046849754236474e-06,
+      "loss": 0.7994,
+      "step": 4268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02076099346344e-06,
+      "loss": 0.7196,
+      "step": 4269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.994712825668438e-06,
+      "loss": 0.8159,
+      "step": 4270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.968705262347197e-06,
+      "loss": 0.7327,
+      "step": 4271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.942738314977628e-06,
+      "loss": 0.774,
+      "step": 4272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.916811995019602e-06,
+      "loss": 0.8101,
+      "step": 4273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89092631391517e-06,
+      "loss": 0.7437,
+      "step": 4274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.865081283088372e-06,
+      "loss": 0.8648,
+      "step": 4275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.839276913945325e-06,
+      "loss": 0.7547,
+      "step": 4276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.813513217874236e-06,
+      "loss": 0.8201,
+      "step": 4277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.787790206245316e-06,
+      "loss": 0.7609,
+      "step": 4278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.762107890410852e-06,
+      "loss": 0.744,
+      "step": 4279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.736466281705135e-06,
+      "loss": 0.7131,
+      "step": 4280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.710865391444556e-06,
+      "loss": 0.7584,
+      "step": 4281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685305230927464e-06,
+      "loss": 0.7761,
+      "step": 4282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.659785811434273e-06,
+      "loss": 0.7928,
+      "step": 4283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.634307144227403e-06,
+      "loss": 0.7893,
+      "step": 4284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.608869240551287e-06,
+      "loss": 0.7375,
+      "step": 4285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.583472111632406e-06,
+      "loss": 0.7955,
+      "step": 4286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.558115768679175e-06,
+      "loss": 0.6859,
+      "step": 4287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.532800222882064e-06,
+      "loss": 0.7405,
+      "step": 4288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.507525485413547e-06,
+      "loss": 0.804,
+      "step": 4289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4822915674280235e-06,
+      "loss": 0.772,
+      "step": 4290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4570984800619415e-06,
+      "loss": 0.7992,
+      "step": 4291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.431946234433718e-06,
+      "loss": 0.7736,
+      "step": 4292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406834841643684e-06,
+      "loss": 0.6788,
+      "step": 4293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.381764312774242e-06,
+      "loss": 0.7416,
+      "step": 4294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.356734658889675e-06,
+      "loss": 0.7673,
+      "step": 4295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3317458910362905e-06,
+      "loss": 0.7632,
+      "step": 4296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.30679802024229e-06,
+      "loss": 0.7314,
+      "step": 4297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281891057517887e-06,
+      "loss": 0.7469,
+      "step": 4298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25702501385519e-06,
+      "loss": 0.7303,
+      "step": 4299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.232199900228287e-06,
+      "loss": 0.7526,
+      "step": 4300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.207415727593169e-06,
+      "loss": 0.7898,
+      "step": 4301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182672506887789e-06,
+      "loss": 0.7419,
+      "step": 4302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.157970249032009e-06,
+      "loss": 0.7451,
+      "step": 4303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.133308964927621e-06,
+      "loss": 0.796,
+      "step": 4304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.108688665458319e-06,
+      "loss": 0.6899,
+      "step": 4305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0841093614897215e-06,
+      "loss": 0.7893,
+      "step": 4306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.059571063869386e-06,
+      "loss": 0.6995,
+      "step": 4307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.035073783426705e-06,
+      "loss": 0.714,
+      "step": 4308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010617530973007e-06,
+      "loss": 0.7863,
+      "step": 4309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986202317301538e-06,
+      "loss": 0.687,
+      "step": 4310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9618281531874e-06,
+      "loss": 0.7492,
+      "step": 4311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.937495049387543e-06,
+      "loss": 0.7693,
+      "step": 4312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.913203016640901e-06,
+      "loss": 0.8579,
+      "step": 4313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.888952065668153e-06,
+      "loss": 0.7327,
+      "step": 4314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8647422071719816e-06,
+      "loss": 0.697,
+      "step": 4315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840573451836796e-06,
+      "loss": 0.6641,
+      "step": 4316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.816445810328997e-06,
+      "loss": 0.8477,
+      "step": 4317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792359293296735e-06,
+      "loss": 0.7758,
+      "step": 4318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.768313911370094e-06,
+      "loss": 0.7988,
+      "step": 4319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.744309675160921e-06,
+      "loss": 0.7296,
+      "step": 4320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720346595262983e-06,
+      "loss": 0.7762,
+      "step": 4321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.696424682251823e-06,
+      "loss": 0.71,
+      "step": 4322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.672543946684873e-06,
+      "loss": 0.7831,
+      "step": 4323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6487043991013356e-06,
+      "loss": 0.7268,
+      "step": 4324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.624906050022272e-06,
+      "loss": 0.7546,
+      "step": 4325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.601148909950538e-06,
+      "loss": 0.7707,
+      "step": 4326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577432989370835e-06,
+      "loss": 0.788,
+      "step": 4327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553758298749635e-06,
+      "loss": 0.6746,
+      "step": 4328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530124848535246e-06,
+      "loss": 0.829,
+      "step": 4329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5065326491577705e-06,
+      "loss": 0.7047,
+      "step": 4330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.482981711029068e-06,
+      "loss": 0.7277,
+      "step": 4331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.459472044542847e-06,
+      "loss": 0.7162,
+      "step": 4332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436003660074541e-06,
+      "loss": 0.7895,
+      "step": 4333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412576567981443e-06,
+      "loss": 0.8083,
+      "step": 4334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3891907786025275e-06,
+      "loss": 0.7918,
+      "step": 4335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365846302258627e-06,
+      "loss": 0.7338,
+      "step": 4336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.342543149252267e-06,
+      "loss": 0.8148,
+      "step": 4337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.31928132986781e-06,
+      "loss": 0.8096,
+      "step": 4338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2960608543713e-06,
+      "loss": 0.7965,
+      "step": 4339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272881733010616e-06,
+      "loss": 0.6967,
+      "step": 4340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.24974397601531e-06,
+      "loss": 0.8204,
+      "step": 4341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226647593596735e-06,
+      "loss": 0.7213,
+      "step": 4342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203592595947938e-06,
+      "loss": 0.6775,
+      "step": 4343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1805789932437686e-06,
+      "loss": 0.7538,
+      "step": 4344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157606795640714e-06,
+      "loss": 0.7934,
+      "step": 4345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.134676013277074e-06,
+      "loss": 0.7042,
+      "step": 4346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111786656272822e-06,
+      "loss": 0.7684,
+      "step": 4347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.088938734729677e-06,
+      "loss": 0.7761,
+      "step": 4348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.066132258731028e-06,
+      "loss": 0.7621,
+      "step": 4349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.043367238342024e-06,
+      "loss": 0.7807,
+      "step": 4350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.020643683609517e-06,
+      "loss": 0.7561,
+      "step": 4351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.997961604561997e-06,
+      "loss": 0.7882,
+      "step": 4352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.975321011209722e-06,
+      "loss": 0.7784,
+      "step": 4353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.95272191354459e-06,
+      "loss": 0.7467,
+      "step": 4354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.930164321540244e-06,
+      "loss": 0.8835,
+      "step": 4355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90764824515192e-06,
+      "loss": 0.7789,
+      "step": 4356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.885173694316626e-06,
+      "loss": 0.7076,
+      "step": 4357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8627406789529715e-06,
+      "loss": 0.7242,
+      "step": 4358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840349208961304e-06,
+      "loss": 0.7479,
+      "step": 4359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8179992942235525e-06,
+      "loss": 0.785,
+      "step": 4360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.795690944603383e-06,
+      "loss": 0.6804,
+      "step": 4361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7734241699460735e-06,
+      "loss": 0.8298,
+      "step": 4362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.751198980078576e-06,
+      "loss": 0.7823,
+      "step": 4363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.729015384809456e-06,
+      "loss": 0.7042,
+      "step": 4364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.706873393928991e-06,
+      "loss": 0.7035,
+      "step": 4365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6847730172089995e-06,
+      "loss": 0.8159,
+      "step": 4366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.662714264403035e-06,
+      "loss": 0.7513,
+      "step": 4367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.640697145246221e-06,
+      "loss": 0.7556,
+      "step": 4368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618721669455284e-06,
+      "loss": 0.7602,
+      "step": 4369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.596787846728646e-06,
+      "loss": 0.8478,
+      "step": 4370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.574895686746318e-06,
+      "loss": 0.7811,
+      "step": 4371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553045199169882e-06,
+      "loss": 0.7423,
+      "step": 4372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.531236393642569e-06,
+      "loss": 0.8029,
+      "step": 4373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509469279789237e-06,
+      "loss": 0.8061,
+      "step": 4374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.487743867216266e-06,
+      "loss": 0.7829,
+      "step": 4375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.466060165511733e-06,
+      "loss": 0.7986,
+      "step": 4376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.444418184245215e-06,
+      "loss": 0.7067,
+      "step": 4377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422817932967939e-06,
+      "loss": 0.7386,
+      "step": 4378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4012594212126564e-06,
+      "loss": 0.8309,
+      "step": 4379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379742658493791e-06,
+      "loss": 0.6545,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358267654307225e-06,
+      "loss": 0.6566,
+      "step": 4381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.336834418130521e-06,
+      "loss": 0.7535,
+      "step": 4382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.315442959422723e-06,
+      "loss": 0.7379,
+      "step": 4383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.294093287624491e-06,
+      "loss": 0.7568,
+      "step": 4384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.27278541215801e-06,
+      "loss": 0.7957,
+      "step": 4385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2515193424270584e-06,
+      "loss": 0.7247,
+      "step": 4386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230295087816917e-06,
+      "loss": 0.7132,
+      "step": 4387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209112657694437e-06,
+      "loss": 0.7083,
+      "step": 4388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1879720614080195e-06,
+      "loss": 0.8414,
+      "step": 4389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1668733082875785e-06,
+      "loss": 0.7986,
+      "step": 4390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.145816407644588e-06,
+      "loss": 0.7999,
+      "step": 4391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.124801368772059e-06,
+      "loss": 0.8227,
+      "step": 4392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.10382820094446e-06,
+      "loss": 0.7712,
+      "step": 4393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.082896913417867e-06,
+      "loss": 0.8033,
+      "step": 4394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062007515429845e-06,
+      "loss": 0.814,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.04116001619942e-06,
+      "loss": 0.7851,
+      "step": 4396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020354424927198e-06,
+      "loss": 0.7755,
+      "step": 4397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.999590750795258e-06,
+      "loss": 0.8304,
+      "step": 4398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978869002967179e-06,
+      "loss": 0.7545,
+      "step": 4399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958189190588036e-06,
+      "loss": 0.7677,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937551322784406e-06,
+      "loss": 0.7534,
+      "step": 4401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916955408664348e-06,
+      "loss": 0.7437,
+      "step": 4402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896401457317423e-06,
+      "loss": 0.8221,
+      "step": 4403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875889477814632e-06,
+      "loss": 0.7559,
+      "step": 4404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.855419479208501e-06,
+      "loss": 0.7745,
+      "step": 4405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8349914705329905e-06,
+      "loss": 0.7914,
+      "step": 4406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814605460803534e-06,
+      "loss": 0.7464,
+      "step": 4407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794261459017069e-06,
+      "loss": 0.7374,
+      "step": 4408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7739594741519455e-06,
+      "loss": 0.7727,
+      "step": 4409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.753699515168009e-06,
+      "loss": 0.6855,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7334815910065055e-06,
+      "loss": 0.7802,
+      "step": 4411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713305710590199e-06,
+      "loss": 0.7789,
+      "step": 4412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.693171882823233e-06,
+      "loss": 0.8046,
+      "step": 4413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.673080116591244e-06,
+      "loss": 0.6414,
+      "step": 4414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6530304207612725e-06,
+      "loss": 0.7083,
+      "step": 4415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.633022804181808e-06,
+      "loss": 0.8136,
+      "step": 4416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.613057275682753e-06,
+      "loss": 0.8267,
+      "step": 4417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.59313384407547e-06,
+      "loss": 0.734,
+      "step": 4418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573252518152693e-06,
+      "loss": 0.8174,
+      "step": 4419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553413306688625e-06,
+      "loss": 0.8226,
+      "step": 4420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.533616218438829e-06,
+      "loss": 0.8183,
+      "step": 4421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.513861262140339e-06,
+      "loss": 0.7167,
+      "step": 4422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.494148446511525e-06,
+      "loss": 0.7043,
+      "step": 4423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.47447778025224e-06,
+      "loss": 0.7665,
+      "step": 4424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.454849272043671e-06,
+      "loss": 0.8124,
+      "step": 4425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4352629305484e-06,
+      "loss": 0.7723,
+      "step": 4426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415718764410459e-06,
+      "loss": 0.7625,
+      "step": 4427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.396216782255191e-06,
+      "loss": 0.8186,
+      "step": 4428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.37675699268939e-06,
+      "loss": 0.718,
+      "step": 4429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.357339404301175e-06,
+      "loss": 0.769,
+      "step": 4430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.33796402566009e-06,
+      "loss": 0.8085,
+      "step": 4431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318630865316997e-06,
+      "loss": 0.7772,
+      "step": 4432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2993399318041805e-06,
+      "loss": 0.7255,
+      "step": 4433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28009123363523e-06,
+      "loss": 0.7556,
+      "step": 4434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.260884779305152e-06,
+      "loss": 0.8335,
+      "step": 4435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.241720577290287e-06,
+      "loss": 0.811,
+      "step": 4436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222598636048303e-06,
+      "loss": 0.7148,
+      "step": 4437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203518964018261e-06,
+      "loss": 0.7621,
+      "step": 4438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184481569620546e-06,
+      "loss": 0.7748,
+      "step": 4439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165486461256873e-06,
+      "loss": 0.738,
+      "step": 4440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.146533647310336e-06,
+      "loss": 0.7732,
+      "step": 4441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.127623136145287e-06,
+      "loss": 0.7459,
+      "step": 4442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.108754936107518e-06,
+      "loss": 0.737,
+      "step": 4443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0899290555240335e-06,
+      "loss": 0.7583,
+      "step": 4444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.071145502703233e-06,
+      "loss": 0.8053,
+      "step": 4445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.052404285934841e-06,
+      "loss": 0.7457,
+      "step": 4446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.033705413489819e-06,
+      "loss": 0.7561,
+      "step": 4447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.015048893620554e-06,
+      "loss": 0.74,
+      "step": 4448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99643473456065e-06,
+      "loss": 0.7986,
+      "step": 4449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.977862944525057e-06,
+      "loss": 0.6942,
+      "step": 4450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.959333531710019e-06,
+      "loss": 0.7039,
+      "step": 4451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940846504293083e-06,
+      "loss": 0.6781,
+      "step": 4452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.922401870433068e-06,
+      "loss": 0.6925,
+      "step": 4453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.903999638270128e-06,
+      "loss": 0.8337,
+      "step": 4454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.885639815925635e-06,
+      "loss": 0.7473,
+      "step": 4455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.867322411502339e-06,
+      "loss": 0.7578,
+      "step": 4456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.849047433084163e-06,
+      "loss": 0.7569,
+      "step": 4457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.830814888736389e-06,
+      "loss": 0.7187,
+      "step": 4458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8126247865055364e-06,
+      "loss": 0.7774,
+      "step": 4459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7944771344194165e-06,
+      "loss": 0.8058,
+      "step": 4460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7763719404870666e-06,
+      "loss": 0.7921,
+      "step": 4461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7583092126988183e-06,
+      "loss": 0.7279,
+      "step": 4462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7402889590262503e-06,
+      "loss": 0.7477,
+      "step": 4463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7223111874221895e-06,
+      "loss": 0.793,
+      "step": 4464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.704375905820745e-06,
+      "loss": 0.7296,
+      "step": 4465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6864831221372187e-06,
+      "loss": 0.7272,
+      "step": 4466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.668632844268227e-06,
+      "loss": 0.7684,
+      "step": 4467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6508250800915577e-06,
+      "loss": 0.8251,
+      "step": 4468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6330598374662907e-06,
+      "loss": 0.7344,
+      "step": 4469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6153371242327095e-06,
+      "loss": 0.7987,
+      "step": 4470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.597656948212347e-06,
+      "loss": 0.79,
+      "step": 4471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5800193172079276e-06,
+      "loss": 0.7661,
+      "step": 4472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562424239003459e-06,
+      "loss": 0.7527,
+      "step": 4473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.544871721364107e-06,
+      "loss": 0.783,
+      "step": 4474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527361772036297e-06,
+      "loss": 0.7397,
+      "step": 4475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5098943987476483e-06,
+      "loss": 0.7479,
+      "step": 4476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4924696092070053e-06,
+      "loss": 0.7905,
+      "step": 4477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475087411104383e-06,
+      "loss": 0.7173,
+      "step": 4478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4577478121110453e-06,
+      "loss": 0.7019,
+      "step": 4479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4404508198794374e-06,
+      "loss": 0.7278,
+      "step": 4480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.423196442043197e-06,
+      "loss": 0.7021,
+      "step": 4481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.405984686217156e-06,
+      "loss": 0.6945,
+      "step": 4482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.388815559997327e-06,
+      "loss": 0.7639,
+      "step": 4483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3716890709609485e-06,
+      "loss": 0.6863,
+      "step": 4484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.354605226666374e-06,
+      "loss": 0.7415,
+      "step": 4485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3375640346532066e-06,
+      "loss": 0.7618,
+      "step": 4486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3205655024421745e-06,
+      "loss": 0.7477,
+      "step": 4487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303609637535221e-06,
+      "loss": 0.813,
+      "step": 4488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.286696447415416e-06,
+      "loss": 0.7423,
+      "step": 4489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269825939547044e-06,
+      "loss": 0.6507,
+      "step": 4490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2529981213754832e-06,
+      "loss": 0.6865,
+      "step": 4491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2362130003273594e-06,
+      "loss": 0.811,
+      "step": 4492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21947058381038e-06,
+      "loss": 0.7465,
+      "step": 4493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.202770879213457e-06,
+      "loss": 0.7512,
+      "step": 4494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.186113893906606e-06,
+      "loss": 0.7345,
+      "step": 4495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1694996352410465e-06,
+      "loss": 0.7473,
+      "step": 4496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1529281105490692e-06,
+      "loss": 0.7588,
+      "step": 4497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1363993271441904e-06,
+      "loss": 0.7972,
+      "step": 4498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.119913292320986e-06,
+      "loss": 0.6717,
+      "step": 4499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1034700133552032e-06,
+      "loss": 0.8336,
+      "step": 4500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0870694975037363e-06,
+      "loss": 0.7437,
+      "step": 4501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0707117520045515e-06,
+      "loss": 0.7657,
+      "step": 4502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0543967840768074e-06,
+      "loss": 0.6656,
+      "step": 4503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.038124600920722e-06,
+      "loss": 0.7157,
+      "step": 4504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0218952097176954e-06,
+      "loss": 0.7957,
+      "step": 4505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.005708617630154e-06,
+      "loss": 0.7895,
+      "step": 4506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.989564831801739e-06,
+      "loss": 0.7996,
+      "step": 4507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9734638593571174e-06,
+      "loss": 0.8129,
+      "step": 4508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9574057074021055e-06,
+      "loss": 0.7135,
+      "step": 4509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.94139038302359e-06,
+      "loss": 0.7001,
+      "step": 4510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.925417893289617e-06,
+      "loss": 0.7797,
+      "step": 4511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9094882452492368e-06,
+      "loss": 0.8096,
+      "step": 4512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8936014459326923e-06,
+      "loss": 0.7455,
+      "step": 4513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8777575023512192e-06,
+      "loss": 0.8337,
+      "step": 4514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.861956421497225e-06,
+      "loss": 0.806,
+      "step": 4515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.846198210344142e-06,
+      "loss": 0.771,
+      "step": 4516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8304828758465185e-06,
+      "loss": 0.7769,
+      "step": 4517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.81481042493994e-06,
+      "loss": 0.7729,
+      "step": 4518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.799180864541129e-06,
+      "loss": 0.722,
+      "step": 4519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.783594201547801e-06,
+      "loss": 0.7362,
+      "step": 4520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7680504428388208e-06,
+      "loss": 0.7437,
+      "step": 4521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7525495952740566e-06,
+      "loss": 0.7455,
+      "step": 4522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.737091665694447e-06,
+      "loss": 0.7926,
+      "step": 4523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7216766609220345e-06,
+      "loss": 0.6845,
+      "step": 4524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706304587759867e-06,
+      "loss": 0.7323,
+      "step": 4525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690975452992084e-06,
+      "loss": 0.6986,
+      "step": 4526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6756892633838183e-06,
+      "loss": 0.7452,
+      "step": 4527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.660446025681329e-06,
+      "loss": 0.7797,
+      "step": 4528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6452457466118464e-06,
+      "loss": 0.7925,
+      "step": 4529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.630088432883693e-06,
+      "loss": 0.723,
+      "step": 4530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6149740911861953e-06,
+      "loss": 0.8035,
+      "step": 4531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5999027281897515e-06,
+      "loss": 0.7593,
+      "step": 4532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5848743505457407e-06,
+      "loss": 0.7361,
+      "step": 4533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.569888964886624e-06,
+      "loss": 0.7247,
+      "step": 4534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5549465778258565e-06,
+      "loss": 0.8051,
+      "step": 4535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5400471959579397e-06,
+      "loss": 0.7593,
+      "step": 4536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.525190825858359e-06,
+      "loss": 0.6671,
+      "step": 4537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.510377474083647e-06,
+      "loss": 0.7428,
+      "step": 4538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4956071471713637e-06,
+      "loss": 0.7756,
+      "step": 4539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480879851640039e-06,
+      "loss": 0.7734,
+      "step": 4540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4661955939892513e-06,
+      "loss": 0.7703,
+      "step": 4541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.451554380699561e-06,
+      "loss": 0.7584,
+      "step": 4542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4369562182325333e-06,
+      "loss": 0.8406,
+      "step": 4543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4224011130307477e-06,
+      "loss": 0.7785,
+      "step": 4544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4078890715177884e-06,
+      "loss": 0.7141,
+      "step": 4545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.393420100098187e-06,
+      "loss": 0.7423,
+      "step": 4546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3789942051575363e-06,
+      "loss": 0.793,
+      "step": 4547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3646113930623547e-06,
+      "loss": 0.7383,
+      "step": 4548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3502716701601978e-06,
+      "loss": 0.7621,
+      "step": 4549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.335975042779559e-06,
+      "loss": 0.7455,
+      "step": 4550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3217215172299578e-06,
+      "loss": 0.7886,
+      "step": 4551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.30751109980184e-06,
+      "loss": 0.683,
+      "step": 4552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.293343796766678e-06,
+      "loss": 0.8318,
+      "step": 4553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2792196143768707e-06,
+      "loss": 0.7607,
+      "step": 4554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2651385588658426e-06,
+      "loss": 0.7521,
+      "step": 4555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2511006364479336e-06,
+      "loss": 0.8011,
+      "step": 4556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2371058533184553e-06,
+      "loss": 0.7827,
+      "step": 4557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223154215653722e-06,
+      "loss": 0.7879,
+      "step": 4558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2092457296109426e-06,
+      "loss": 0.7846,
+      "step": 4559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1953804013283398e-06,
+      "loss": 0.7827,
+      "step": 4560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181558236925063e-06,
+      "loss": 0.7998,
+      "step": 4561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1677792425012e-06,
+      "loss": 0.7876,
+      "step": 4562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1540434241378195e-06,
+      "loss": 0.7734,
+      "step": 4563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.140350787896905e-06,
+      "loss": 0.7202,
+      "step": 4564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126701339821402e-06,
+      "loss": 0.7885,
+      "step": 4565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1130950859351794e-06,
+      "loss": 0.7858,
+      "step": 4566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0995320322430677e-06,
+      "loss": 0.6911,
+      "step": 4567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.086012184730812e-06,
+      "loss": 0.8084,
+      "step": 4568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0725355493650845e-06,
+      "loss": 0.7928,
+      "step": 4569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0591021320935154e-06,
+      "loss": 0.712,
+      "step": 4570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.045711938844619e-06,
+      "loss": 0.7771,
+      "step": 4571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0323649755278785e-06,
+      "loss": 0.7939,
+      "step": 4572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.01906124803366e-06,
+      "loss": 0.7918,
+      "step": 4573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0058007622332897e-06,
+      "loss": 0.718,
+      "step": 4574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9925835239789637e-06,
+      "loss": 0.7728,
+      "step": 4575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9794095391038048e-06,
+      "loss": 0.7284,
+      "step": 4576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9662788134218955e-06,
+      "loss": 0.645,
+      "step": 4577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9531913527281453e-06,
+      "loss": 0.8192,
+      "step": 4578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9401471627984446e-06,
+      "loss": 0.7248,
+      "step": 4579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.927146249389533e-06,
+      "loss": 0.7687,
+      "step": 4580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.914188618239088e-06,
+      "loss": 0.7021,
+      "step": 4581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9012742750656566e-06,
+      "loss": 0.741,
+      "step": 4582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8884032255687023e-06,
+      "loss": 0.808,
+      "step": 4583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8755754754285705e-06,
+      "loss": 0.7773,
+      "step": 4584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.86279103030651e-06,
+      "loss": 0.8298,
+      "step": 4585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.850049895844652e-06,
+      "loss": 0.7285,
+      "step": 4586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.837352077665988e-06,
+      "loss": 0.8005,
+      "step": 4587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8246975813744348e-06,
+      "loss": 0.765,
+      "step": 4588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8120864125547698e-06,
+      "loss": 0.742,
+      "step": 4589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.799518576772652e-06,
+      "loss": 0.782,
+      "step": 4590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7869940795746109e-06,
+      "loss": 0.7145,
+      "step": 4591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7745129264880367e-06,
+      "loss": 0.7474,
+      "step": 4592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.762075123021234e-06,
+      "loss": 0.7404,
+      "step": 4593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7496806746633453e-06,
+      "loss": 0.7204,
+      "step": 4594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.737329586884362e-06,
+      "loss": 0.7797,
+      "step": 4595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.725021865135179e-06,
+      "loss": 0.7996,
+      "step": 4596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7127575148475183e-06,
+      "loss": 0.8241,
+      "step": 4597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7005365414339946e-06,
+      "loss": 0.7841,
+      "step": 4598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6883589502880492e-06,
+      "loss": 0.7649,
+      "step": 4599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6762247467839941e-06,
+      "loss": 0.7098,
+      "step": 4600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6641339362769793e-06,
+      "loss": 0.7756,
+      "step": 4601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6520865241030247e-06,
+      "loss": 0.7571,
+      "step": 4602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.640082515578978e-06,
+      "loss": 0.7877,
+      "step": 4603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6281219160025564e-06,
+      "loss": 0.7835,
+      "step": 4604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.616204730652271e-06,
+      "loss": 0.7852,
+      "step": 4605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604330964787537e-06,
+      "loss": 0.7778,
+      "step": 4606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592500623648563e-06,
+      "loss": 0.775,
+      "step": 4607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5807137124563942e-06,
+      "loss": 0.7372,
+      "step": 4608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5689702364129366e-06,
+      "loss": 0.6891,
+      "step": 4609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5572702007008889e-06,
+      "loss": 0.7634,
+      "step": 4610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5456136104838315e-06,
+      "loss": 0.6996,
+      "step": 4611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5340004709061163e-06,
+      "loss": 0.8035,
+      "step": 4612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5224307870929545e-06,
+      "loss": 0.808,
+      "step": 4613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5109045641503505e-06,
+      "loss": 0.7596,
+      "step": 4614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4994218071651689e-06,
+      "loss": 0.7508,
+      "step": 4615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4879825212050336e-06,
+      "loss": 0.7198,
+      "step": 4616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4765867113184617e-06,
+      "loss": 0.6948,
+      "step": 4617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4652343825346971e-06,
+      "loss": 0.7032,
+      "step": 4618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4539255398638651e-06,
+      "loss": 0.8073,
+      "step": 4619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4426601882968516e-06,
+      "loss": 0.7307,
+      "step": 4620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4314383328053794e-06,
+      "loss": 0.7796,
+      "step": 4621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4202599783419534e-06,
+      "loss": 0.7484,
+      "step": 4622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4091251298398943e-06,
+      "loss": 0.807,
+      "step": 4623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.398033792213327e-06,
+      "loss": 0.8107,
+      "step": 4624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3869859703571463e-06,
+      "loss": 0.7312,
+      "step": 4625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3759816691470751e-06,
+      "loss": 0.7228,
+      "step": 4626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365020893439617e-06,
+      "loss": 0.8403,
+      "step": 4627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3541036480720581e-06,
+      "loss": 0.7597,
+      "step": 4628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.343229937862478e-06,
+      "loss": 0.781,
+      "step": 4629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.33239976760976e-06,
+      "loss": 0.7372,
+      "step": 4630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3216131420935362e-06,
+      "loss": 0.7603,
+      "step": 4631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3108700660742546e-06,
+      "loss": 0.8071,
+      "step": 4632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3001705442931333e-06,
+      "loss": 0.8416,
+      "step": 4633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2895145814721622e-06,
+      "loss": 0.7705,
+      "step": 4634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2789021823141122e-06,
+      "loss": 0.759,
+      "step": 4635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2683333515025264e-06,
+      "loss": 0.751,
+      "step": 4636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2578080937017178e-06,
+      "loss": 0.7719,
+      "step": 4637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.247326413556793e-06,
+      "loss": 0.7716,
+      "step": 4638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2368883156935851e-06,
+      "loss": 0.7824,
+      "step": 4639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2264938047187314e-06,
+      "loss": 0.814,
+      "step": 4640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2161428852195955e-06,
+      "loss": 0.7708,
+      "step": 4641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2058355617643457e-06,
+      "loss": 0.7211,
+      "step": 4642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1955718389018766e-06,
+      "loss": 0.7626,
+      "step": 4643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1853517211618536e-06,
+      "loss": 0.7562,
+      "step": 4644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1751752130547022e-06,
+      "loss": 0.7986,
+      "step": 4645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1650423190715965e-06,
+      "loss": 0.8151,
+      "step": 4646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.154953043684448e-06,
+      "loss": 0.7991,
+      "step": 4647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1449073913459508e-06,
+      "loss": 0.7831,
+      "step": 4648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1349053664895138e-06,
+      "loss": 0.7789,
+      "step": 4649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1249469735293283e-06,
+      "loss": 0.8127,
+      "step": 4650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.115032216860279e-06,
+      "loss": 0.756,
+      "step": 4651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.105161100858043e-06,
+      "loss": 0.6765,
+      "step": 4652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.095333629879014e-06,
+      "loss": 0.7438,
+      "step": 4653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.085549808260322e-06,
+      "loss": 0.8027,
+      "step": 4654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0758096403198358e-06,
+      "loss": 0.765,
+      "step": 4655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.066113130356161e-06,
+      "loss": 0.7494,
+      "step": 4656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0564602826486414e-06,
+      "loss": 0.709,
+      "step": 4657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0468511014573356e-06,
+      "loss": 0.791,
+      "step": 4658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0372855910230517e-06,
+      "loss": 0.7966,
+      "step": 4659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0277637555673014e-06,
+      "loss": 0.7966,
+      "step": 4660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0182855992923456e-06,
+      "loss": 0.8012,
+      "step": 4661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0088511263811385e-06,
+      "loss": 0.7797,
+      "step": 4662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.994603409973934e-07,
+      "loss": 0.7219,
+      "step": 4663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.90113247285518e-07,
+      "loss": 0.7528,
+      "step": 4664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808098493706451e-07,
+      "loss": 0.8507,
+      "step": 4665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.715501513586023e-07,
+      "loss": 0.6594,
+      "step": 4666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.623341573359756e-07,
+      "loss": 0.7157,
+      "step": 4667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.53161871370023e-07,
+      "loss": 0.7479,
+      "step": 4668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.440332975087396e-07,
+      "loss": 0.7471,
+      "step": 4669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.349484397807917e-07,
+      "loss": 0.8331,
+      "step": 4670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.259073021956055e-07,
+      "loss": 0.7165,
+      "step": 4671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.16909888743267e-07,
+      "loss": 0.7695,
+      "step": 4672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.079562033945888e-07,
+      "loss": 0.6969,
+      "step": 4673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.990462501010877e-07,
+      "loss": 0.8401,
+      "step": 4674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.901800327949627e-07,
+      "loss": 0.7689,
+      "step": 4675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.813575553891173e-07,
+      "loss": 0.7161,
+      "step": 4676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.725788217771591e-07,
+      "loss": 0.8033,
+      "step": 4677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.638438358334e-07,
+      "loss": 0.7559,
+      "step": 4678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.551526014128009e-07,
+      "loss": 0.7843,
+      "step": 4679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.465051223510711e-07,
+      "loss": 0.6614,
+      "step": 4680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.379014024645581e-07,
+      "loss": 0.7631,
+      "step": 4681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.293414455503468e-07,
+      "loss": 0.7815,
+      "step": 4682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.208252553861595e-07,
+      "loss": 0.7692,
+      "step": 4683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.123528357304344e-07,
+      "loss": 0.7714,
+      "step": 4684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.039241903222916e-07,
+      "loss": 0.7733,
+      "step": 4685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.955393228815111e-07,
+      "loss": 0.8513,
+      "step": 4686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.871982371085774e-07,
+      "loss": 0.7753,
+      "step": 4687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.789009366846456e-07,
+      "loss": 0.6792,
+      "step": 4688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.706474252715312e-07,
+      "loss": 0.7373,
+      "step": 4689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.624377065117428e-07,
+      "loss": 0.7747,
+      "step": 4690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.542717840284486e-07,
+      "loss": 0.6865,
+      "step": 4691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.461496614255104e-07,
+      "loss": 0.7272,
+      "step": 4692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.380713422874275e-07,
+      "loss": 0.7665,
+      "step": 4693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.300368301794036e-07,
+      "loss": 0.7261,
+      "step": 4694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.220461286472802e-07,
+      "loss": 0.6916,
+      "step": 4695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.140992412175696e-07,
+      "loss": 0.8023,
+      "step": 4696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.061961713974663e-07,
+      "loss": 0.816,
+      "step": 4697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.983369226748026e-07,
+      "loss": 0.78,
+      "step": 4698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.90521498518093e-07,
+      "loss": 0.7638,
+      "step": 4699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.827499023764783e-07,
+      "loss": 0.7324,
+      "step": 4700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.750221376798149e-07,
+      "loss": 0.7899,
+      "step": 4701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.673382078385304e-07,
+      "loss": 0.7974,
+      "step": 4702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.596981162438009e-07,
+      "loss": 0.7748,
+      "step": 4703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.52101866267385e-07,
+      "loss": 0.7567,
+      "step": 4704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.445494612617231e-07,
+      "loss": 0.7723,
+      "step": 4705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.370409045598935e-07,
+      "loss": 0.7757,
+      "step": 4706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.295761994756455e-07,
+      "loss": 0.7773,
+      "step": 4707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.221553493033439e-07,
+      "loss": 0.8467,
+      "step": 4708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.147783573180243e-07,
+      "loss": 0.7421,
+      "step": 4709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.074452267753494e-07,
+      "loss": 0.7654,
+      "step": 4710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.001559609116413e-07,
+      "loss": 0.7517,
+      "step": 4711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.929105629438492e-07,
+      "loss": 0.7231,
+      "step": 4712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.857090360695594e-07,
+      "loss": 0.7635,
+      "step": 4713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.785513834670187e-07,
+      "loss": 0.8347,
+      "step": 4714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.71437608295089e-07,
+      "loss": 0.7462,
+      "step": 4715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.643677136932813e-07,
+      "loss": 0.7055,
+      "step": 4716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.573417027817219e-07,
+      "loss": 0.7061,
+      "step": 4717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.503595786612082e-07,
+      "loss": 0.697,
+      "step": 4718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.434213444131198e-07,
+      "loss": 0.7612,
+      "step": 4719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.365270030994962e-07,
+      "loss": 0.7446,
+      "step": 4720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.296765577630148e-07,
+      "loss": 0.7646,
+      "step": 4721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.228700114269569e-07,
+      "loss": 0.8508,
+      "step": 4722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.16107367095231e-07,
+      "loss": 0.7528,
+      "step": 4723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.093886277523829e-07,
+      "loss": 0.7444,
+      "step": 4724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.027137963635963e-07,
+      "loss": 0.7324,
+      "step": 4725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.960828758746371e-07,
+      "loss": 0.6992,
+      "step": 4726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8949586921192e-07,
+      "loss": 0.6657,
+      "step": 4727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.82952779282464e-07,
+      "loss": 0.7723,
+      "step": 4728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7645360897393687e-07,
+      "loss": 0.7111,
+      "step": 4729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6999836115458884e-07,
+      "loss": 0.8158,
+      "step": 4730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.635870386732966e-07,
+      "loss": 0.8896,
+      "step": 4731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5721964435955224e-07,
+      "loss": 0.7162,
+      "step": 4732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5089618102347463e-07,
+      "loss": 0.7988,
+      "step": 4733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4461665145577593e-07,
+      "loss": 0.8582,
+      "step": 4734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.383810584277837e-07,
+      "loss": 0.7361,
+      "step": 4735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.321894046914299e-07,
+      "loss": 0.8074,
+      "step": 4736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.260416929792843e-07,
+      "loss": 0.7325,
+      "step": 4737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1993792600447666e-07,
+      "loss": 0.7339,
+      "step": 4738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.138781064607966e-07,
+      "loss": 0.696,
+      "step": 4739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0786223702258266e-07,
+      "loss": 0.8201,
+      "step": 4740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.018903203448221e-07,
+      "loss": 0.7804,
+      "step": 4741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9596235906307345e-07,
+      "loss": 0.7426,
+      "step": 4742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.900783557935217e-07,
+      "loss": 0.7839,
+      "step": 4743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.84238313132923e-07,
+      "loss": 0.6991,
+      "step": 4744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.784422336586713e-07,
+      "loss": 0.6526,
+      "step": 4745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7269011992872027e-07,
+      "loss": 0.7099,
+      "step": 4746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.669819744816505e-07,
+      "loss": 0.745,
+      "step": 4747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.613177998366135e-07,
+      "loss": 0.8041,
+      "step": 4748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5569759849335414e-07,
+      "loss": 0.8096,
+      "step": 4749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.50121372932255e-07,
+      "loss": 0.7904,
+      "step": 4750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.445891256142253e-07,
+      "loss": 0.7814,
+      "step": 4751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391008589808231e-07,
+      "loss": 0.7481,
+      "step": 4752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.336565754541554e-07,
+      "loss": 0.7431,
+      "step": 4753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2825627743695576e-07,
+      "loss": 0.777,
+      "step": 4754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.228999673124955e-07,
+      "loss": 0.7401,
+      "step": 4755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1758764744468374e-07,
+      "loss": 0.7548,
+      "step": 4756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.123193201779784e-07,
+      "loss": 0.7679,
+      "step": 4757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.070949878374529e-07,
+      "loss": 0.7245,
+      "step": 4758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0191465272875197e-07,
+      "loss": 0.7374,
+      "step": 4759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9677831713808e-07,
+      "loss": 0.771,
+      "step": 4760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9168598333225717e-07,
+      "loss": 0.7621,
+      "step": 4761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8663765355867454e-07,
+      "loss": 0.7859,
+      "step": 4762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.816333300452945e-07,
+      "loss": 0.7855,
+      "step": 4763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7667301500066135e-07,
+      "loss": 0.8452,
+      "step": 4764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7175671061390187e-07,
+      "loss": 0.7817,
+      "step": 4765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6688441905471373e-07,
+      "loss": 0.7023,
+      "step": 4766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6205614247337695e-07,
+      "loss": 0.7322,
+      "step": 4767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5727188300074256e-07,
+      "loss": 0.6681,
+      "step": 4768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.525316427482327e-07,
+      "loss": 0.7692,
+      "step": 4769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4783542380785174e-07,
+      "loss": 0.7796,
+      "step": 4770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.43183228252164e-07,
+      "loss": 0.8426,
+      "step": 4771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.385750581343271e-07,
+      "loss": 0.7217,
+      "step": 4772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3401091548803656e-07,
+      "loss": 0.7547,
+      "step": 4773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2949080232758102e-07,
+      "loss": 0.8017,
+      "step": 4774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2501472064780926e-07,
+      "loss": 0.78,
+      "step": 4775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.205826724241522e-07,
+      "loss": 0.7617,
+      "step": 4776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1619465961257856e-07,
+      "loss": 0.7471,
+      "step": 4777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1185068414963928e-07,
+      "loss": 0.8301,
+      "step": 4778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0755074795246744e-07,
+      "loss": 0.7884,
+      "step": 4779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0329485291872287e-07,
+      "loss": 0.8182,
+      "step": 4780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9908300092666975e-07,
+      "loss": 0.7787,
+      "step": 4781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9491519383509905e-07,
+      "loss": 0.7357,
+      "step": 4782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.907914334833838e-07,
+      "loss": 0.8401,
+      "step": 4783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8671172169145711e-07,
+      "loss": 0.7592,
+      "step": 4784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.826760602598121e-07,
+      "loss": 0.6907,
+      "step": 4785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.786844509694907e-07,
+      "loss": 0.7849,
+      "step": 4786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7473689558209494e-07,
+      "loss": 0.7636,
+      "step": 4787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7083339583979784e-07,
+      "loss": 0.812,
+      "step": 4788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.669739534653103e-07,
+      "loss": 0.7443,
+      "step": 4789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.631585701619365e-07,
+      "loss": 0.7901,
+      "step": 4790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.59387247613485e-07,
+      "loss": 0.7803,
+      "step": 4791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5565998748435783e-07,
+      "loss": 0.7448,
+      "step": 4792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.519767914194836e-07,
+      "loss": 0.713,
+      "step": 4793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.483376610443843e-07,
+      "loss": 0.8053,
+      "step": 4794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4474259796508627e-07,
+      "loss": 0.7597,
+      "step": 4795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.411916037681982e-07,
+      "loss": 0.7948,
+      "step": 4796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3768468002086643e-07,
+      "loss": 0.778,
+      "step": 4797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3422182827079745e-07,
+      "loss": 0.7921,
+      "step": 4798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.308030500462465e-07,
+      "loss": 0.7313,
+      "step": 4799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2742834685600668e-07,
+      "loss": 0.762,
+      "step": 4800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2409772018944222e-07,
+      "loss": 0.765,
+      "step": 4801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2081117151643283e-07,
+      "loss": 0.8031,
+      "step": 4802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1756870228742945e-07,
+      "loss": 0.7891,
+      "step": 4803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1437031393343179e-07,
+      "loss": 0.7892,
+      "step": 4804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1121600786596631e-07,
+      "loss": 0.7953,
+      "step": 4805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0810578547713058e-07,
+      "loss": 0.7716,
+      "step": 4806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.050396481395266e-07,
+      "loss": 0.7531,
+      "step": 4807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0201759720634973e-07,
+      "loss": 0.779,
+      "step": 4808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.903963401128869e-08,
+      "loss": 0.7342,
+      "step": 4809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.610575986862547e-08,
+      "loss": 0.7686,
+      "step": 4810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.321597607315769e-08,
+      "loss": 0.8331,
+      "step": 4811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.037028390020962e-08,
+      "loss": 0.7234,
+      "step": 4812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.756868460567669e-08,
+      "loss": 0.7824,
+      "step": 4813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.481117942598093e-08,
+      "loss": 0.7447,
+      "step": 4814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.209776957808225e-08,
+      "loss": 0.7212,
+      "step": 4815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.942845625948937e-08,
+      "loss": 0.8147,
+      "step": 4816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.680324064823774e-08,
+      "loss": 0.7565,
+      "step": 4817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.422212390290062e-08,
+      "loss": 0.7175,
+      "step": 4818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.168510716261123e-08,
+      "loss": 0.8219,
+      "step": 4819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.91921915470184e-08,
+      "loss": 0.7481,
+      "step": 4820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.674337815631981e-08,
+      "loss": 0.8138,
+      "step": 4821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.433866807122879e-08,
+      "loss": 0.7567,
+      "step": 4822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.197806235302971e-08,
+      "loss": 0.7668,
+      "step": 4823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.966156204351148e-08,
+      "loss": 0.7577,
+      "step": 4824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.738916816501183e-08,
+      "loss": 0.8082,
+      "step": 4825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.516088172039524e-08,
+      "loss": 0.7996,
+      "step": 4826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.297670369308616e-08,
+      "loss": 0.7633,
+      "step": 4827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.083663504699132e-08,
+      "loss": 0.7325,
+      "step": 4828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.874067672661076e-08,
+      "loss": 0.6743,
+      "step": 4829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.668882965693788e-08,
+      "loss": 0.6975,
+      "step": 4830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.468109474350391e-08,
+      "loss": 0.7957,
+      "step": 4831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.271747287238892e-08,
+      "loss": 0.8085,
+      "step": 4832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.079796491018861e-08,
+      "loss": 0.7394,
+      "step": 4833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.892257170402536e-08,
+      "loss": 0.7126,
+      "step": 4834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.709129408158152e-08,
+      "loss": 0.7612,
+      "step": 4835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.530413285103285e-08,
+      "loss": 0.6444,
+      "step": 4836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356108880110398e-08,
+      "loss": 0.7523,
+      "step": 4837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.186216270105735e-08,
+      "loss": 0.7708,
+      "step": 4838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.020735530068208e-08,
+      "loss": 0.6743,
+      "step": 4839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8596667330271776e-08,
+      "loss": 0.714,
+      "step": 4840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.703009950068003e-08,
+      "loss": 0.7473,
+      "step": 4841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5507652503264924e-08,
+      "loss": 0.7995,
+      "step": 4842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4029327009944535e-08,
+      "loss": 0.7267,
+      "step": 4843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2595123673119224e-08,
+      "loss": 0.7042,
+      "step": 4844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1205043125760438e-08,
+      "loss": 0.7043,
+      "step": 4845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.985908598135522e-08,
+      "loss": 0.7257,
+      "step": 4846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855725283389509e-08,
+      "loss": 0.7314,
+      "step": 4847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7299544257909363e-08,
+      "loss": 0.7592,
+      "step": 4848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6085960808487343e-08,
+      "loss": 0.7453,
+      "step": 4849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.491650302118952e-08,
+      "loss": 0.7534,
+      "step": 4850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3791171412147474e-08,
+      "loss": 0.8043,
+      "step": 4851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2709966477986168e-08,
+      "loss": 0.7595,
+      "step": 4852
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.167288869589056e-08,
+      "loss": 0.7726,
+      "step": 4853
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0679938523527888e-08,
+      "loss": 0.8262,
+      "step": 4854
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.731116399136487e-09,
+      "loss": 0.764,
+      "step": 4855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.82642274144807e-09,
+      "loss": 0.7236,
+      "step": 4856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.965857949732148e-09,
+      "loss": 0.7069,
+      "step": 4857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.149422403773809e-09,
+      "loss": 0.7919,
+      "step": 4858
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.377116463884836e-09,
+      "loss": 0.787,
+      "step": 4859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.648940470925901e-09,
+      "loss": 0.7453,
+      "step": 4860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.964894746239957e-09,
+      "loss": 0.7632,
+      "step": 4861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3249795917188475e-09,
+      "loss": 0.7717,
+      "step": 4862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.729195289781107e-09,
+      "loss": 0.6566,
+      "step": 4863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.177542103360853e-09,
+      "loss": 0.7779,
+      "step": 4864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.670020275929996e-09,
+      "loss": 0.7812,
+      "step": 4865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2066300314427248e-09,
+      "loss": 0.8143,
+      "step": 4866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7873715744243235e-09,
+      "loss": 0.7894,
+      "step": 4867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4122450899156648e-09,
+      "loss": 0.7501,
+      "step": 4868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.081250743451001e-09,
+      "loss": 0.7121,
+      "step": 4869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.943886811134782e-10,
+      "loss": 0.7029,
+      "step": 4870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.516590295018275e-10,
+      "loss": 0.6706,
+      "step": 4871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5306189574146886e-10,
+      "loss": 0.7325,
+      "step": 4872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9859736748450985e-10,
+      "loss": 0.7562,
+      "step": 4873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.826551288754204e-11,
+      "loss": 0.7811,
+      "step": 4874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.206638065604949e-11,
+      "loss": 0.7692,
+      "step": 4875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7813,
+      "step": 4876
+    },
+    {
+      "epoch": 1.0,
+      "step": 4876,
+      "total_flos": 6627309380829184.0,
+      "train_loss": 0.8137266478448544,
+      "train_runtime": 40315.1287,
+      "train_samples_per_second": 15.484,
+      "train_steps_per_second": 0.121
+    }
+  ],
+  "max_steps": 4876,
+  "num_train_epochs": 1,
+  "total_flos": 6627309380829184.0,
+  "trial_name": null,
+  "trial_params": null
+}