diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17419 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9531568228105907,
+  "eval_steps": 500,
+  "global_step": 2900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.247191011235955e-06,
+      "loss": 0.8254,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49438202247191e-06,
+      "loss": 1.046,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.741573033707865e-06,
+      "loss": 0.9197,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.98876404494382e-06,
+      "loss": 0.9961,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1235955056179776e-05,
+      "loss": 1.0181,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.348314606741573e-05,
+      "loss": 1.052,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5730337078651687e-05,
+      "loss": 1.0831,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.797752808988764e-05,
+      "loss": 1.0871,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0224719101123596e-05,
+      "loss": 1.102,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2471910112359552e-05,
+      "loss": 1.0938,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4719101123595505e-05,
+      "loss": 1.0667,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.696629213483146e-05,
+      "loss": 1.0763,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9213483146067417e-05,
+      "loss": 1.0574,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1460674157303374e-05,
+      "loss": 1.1122,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.370786516853933e-05,
+      "loss": 1.1003,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.595505617977528e-05,
+      "loss": 1.1272,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8202247191011236e-05,
+      "loss": 1.1923,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.044943820224719e-05,
+      "loss": 1.1731,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.269662921348315e-05,
+      "loss": 1.0674,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4943820224719104e-05,
+      "loss": 1.0595,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.719101123595506e-05,
+      "loss": 0.9749,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.943820224719101e-05,
+      "loss": 1.1203,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.168539325842697e-05,
+      "loss": 1.1086,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.393258426966292e-05,
+      "loss": 1.071,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6179775280898885e-05,
+      "loss": 1.0467,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8426966292134835e-05,
+      "loss": 1.0194,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.067415730337079e-05,
+      "loss": 0.9577,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.292134831460675e-05,
+      "loss": 0.9345,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.51685393258427e-05,
+      "loss": 0.8376,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.741573033707866e-05,
+      "loss": 0.9179,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.966292134831462e-05,
+      "loss": 0.9753,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.191011235955056e-05,
+      "loss": 0.8388,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.415730337078653e-05,
+      "loss": 0.7649,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.640449438202247e-05,
+      "loss": 0.7842,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.865168539325843e-05,
+      "loss": 0.8453,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.089887640449438e-05,
+      "loss": 0.8328,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.314606741573034e-05,
+      "loss": 0.7431,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.53932584269663e-05,
+      "loss": 0.7872,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.764044943820225e-05,
+      "loss": 0.7584,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.988764044943821e-05,
+      "loss": 0.7106,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.213483146067416e-05,
+      "loss": 0.9839,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.438202247191012e-05,
+      "loss": 0.6744,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.662921348314608e-05,
+      "loss": 0.7247,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.887640449438202e-05,
+      "loss": 0.7546,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010112359550561799,
+      "loss": 0.7806,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010337078651685395,
+      "loss": 0.5714,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010561797752808989,
+      "loss": 0.8032,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010786516853932584,
+      "loss": 0.7078,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001101123595505618,
+      "loss": 0.6966,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011235955056179777,
+      "loss": 0.6984,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001146067415730337,
+      "loss": 0.6214,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011685393258426967,
+      "loss": 0.5486,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011910112359550563,
+      "loss": 0.6808,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012134831460674158,
+      "loss": 0.5786,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012359550561797752,
+      "loss": 0.586,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001258426966292135,
+      "loss": 0.5482,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012808988764044944,
+      "loss": 0.5106,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001303370786516854,
+      "loss": 0.5754,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013258426966292135,
+      "loss": 0.5048,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013483146067415732,
+      "loss": 0.5145,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013707865168539326,
+      "loss": 0.5832,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013932584269662923,
+      "loss": 0.5094,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014157303370786517,
+      "loss": 0.5035,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014382022471910112,
+      "loss": 0.4312,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001460674157303371,
+      "loss": 0.6612,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014831460674157306,
+      "loss": 0.5597,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000150561797752809,
+      "loss": 0.5859,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015280898876404494,
+      "loss": 0.6278,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001550561797752809,
+      "loss": 0.5873,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015730337078651685,
+      "loss": 0.4841,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001595505617977528,
+      "loss": 0.6553,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016179775280898877,
+      "loss": 0.613,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016404494382022474,
+      "loss": 0.5434,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016629213483146068,
+      "loss": 0.5261,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016853932584269662,
+      "loss": 0.5512,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001707865168539326,
+      "loss": 0.5666,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017303370786516853,
+      "loss": 0.4988,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001752808988764045,
+      "loss": 0.5,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017752808988764045,
+      "loss": 0.5185,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017977528089887642,
+      "loss": 0.6003,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018202247191011236,
+      "loss": 0.5247,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018426966292134833,
+      "loss": 0.4914,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018651685393258427,
+      "loss": 0.5098,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018876404494382024,
+      "loss": 0.5598,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019101123595505618,
+      "loss": 0.5973,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019325842696629215,
+      "loss": 0.4494,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001955056179775281,
+      "loss": 0.5404,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019775280898876404,
+      "loss": 0.4969,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4632,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999993954263357,
+      "loss": 0.494,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999975817060732,
+      "loss": 0.5633,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999945588414057,
+      "loss": 0.5379,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999903268359882,
+      "loss": 0.6052,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999848856949384,
+      "loss": 0.5884,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999782354248346,
+      "loss": 0.5024,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999703760337185,
+      "loss": 0.4861,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999961307531093,
+      "loss": 0.6242,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999510299279237,
+      "loss": 0.4253,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999939543236637,
+      "loss": 0.5883,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999268474711228,
+      "loss": 0.7311,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999129426467315,
+      "loss": 0.65,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998978287802765,
+      "loss": 0.6306,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998815058900325,
+      "loss": 0.6467,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998639739957362,
+      "loss": 0.5509,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998452331185866,
+      "loss": 0.5411,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998252832812438,
+      "loss": 0.566,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998041245078302,
+      "loss": 0.4258,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199978175682393,
+      "loss": 0.503,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997581802565887,
+      "loss": 0.5369,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997333948343146,
+      "loss": 0.6057,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999707400587076,
+      "loss": 0.576,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999680197546304,
+      "loss": 0.5973,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019996517857448915,
+      "loss": 0.4609,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019996221652171924,
+      "loss": 0.451,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999591335999022,
+      "loss": 0.4483,
+      "step": 115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999559298127658,
+      "loss": 0.4452,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999526051641838,
+      "loss": 0.5005,
+      "step": 117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994915965817627,
+      "loss": 0.48,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994559329890928,
+      "loss": 0.5123,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994190609069512,
+      "loss": 0.4618,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019993809803799212,
+      "loss": 0.4386,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019993416914540485,
+      "loss": 0.4597,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019993011941768388,
+      "loss": 0.5077,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001999259488597259,
+      "loss": 0.5641,
+      "step": 124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019992165747657374,
+      "loss": 0.4633,
+      "step": 125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019991724527341638,
+      "loss": 0.6939,
+      "step": 126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019991271225558874,
+      "loss": 0.4553,
+      "step": 127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019990805842857196,
+      "loss": 0.548,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001999032837979932,
+      "loss": 0.4714,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019989838836962563,
+      "loss": 0.5626,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019989337214938862,
+      "loss": 0.4359,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019988823514334748,
+      "loss": 0.4714,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001998829773577136,
+      "loss": 0.4725,
+      "step": 133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019987759879884446,
+      "loss": 0.5201,
+      "step": 134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019987209947324353,
+      "loss": 0.6127,
+      "step": 135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019986647938756023,
+      "loss": 0.442,
+      "step": 136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019986073854859012,
+      "loss": 0.4271,
+      "step": 137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019985487696327475,
+      "loss": 0.5441,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019984889463870158,
+      "loss": 0.5935,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019984279158210416,
+      "loss": 0.4738,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000199836567800862,
+      "loss": 0.4323,
+      "step": 141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019983022330250053,
+      "loss": 0.4152,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019982375809469118,
+      "loss": 0.5317,
+      "step": 143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019981717218525138,
+      "loss": 0.4043,
+      "step": 144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019981046558214445,
+      "loss": 0.4099,
+      "step": 145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019980363829347965,
+      "loss": 0.3937,
+      "step": 146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019979669032751214,
+      "loss": 0.5095,
+      "step": 147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001997896216926431,
+      "loss": 0.5522,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019978243239741954,
+      "loss": 0.5025,
+      "step": 149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019977512245053435,
+      "loss": 0.5271,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019976769186082636,
+      "loss": 0.5682,
+      "step": 151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019976014063728022,
+      "loss": 0.5727,
+      "step": 152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019975246878902649,
+      "loss": 0.4575,
+      "step": 153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019974467632534156,
+      "loss": 0.5744,
+      "step": 154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019973676325564768,
+      "loss": 0.5024,
+      "step": 155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001997287295895129,
+      "loss": 0.4662,
+      "step": 156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001997205753366511,
+      "loss": 0.4365,
+      "step": 157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019971230050692197,
+      "loss": 0.5171,
+      "step": 158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019970390511033105,
+      "loss": 0.3913,
+      "step": 159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019969538915702955,
+      "loss": 0.4408,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019968675265731454,
+      "loss": 0.4441,
+      "step": 161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001996779956216288,
+      "loss": 0.6463,
+      "step": 162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001996691180605609,
+      "loss": 0.523,
+      "step": 163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019966011998484513,
+      "loss": 0.4613,
+      "step": 164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019965100140536147,
+      "loss": 0.4109,
+      "step": 165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001996417623331356,
+      "loss": 0.347,
+      "step": 166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019963240277933898,
+      "loss": 0.409,
+      "step": 167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019962292275528862,
+      "loss": 0.3978,
+      "step": 168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019961332227244732,
+      "loss": 0.3026,
+      "step": 169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019960360134242344,
+      "loss": 0.4776,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019959375997697105,
+      "loss": 0.4078,
+      "step": 171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019958379818798983,
+      "loss": 0.4576,
+      "step": 172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019957371598752497,
+      "loss": 0.4501,
+      "step": 173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001995635133877674,
+      "loss": 0.7466,
+      "step": 174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019955319040105356,
+      "loss": 0.5378,
+      "step": 175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019954274703986541,
+      "loss": 0.4596,
+      "step": 176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019953218331683061,
+      "loss": 0.4381,
+      "step": 177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019952149924472217,
+      "loss": 0.493,
+      "step": 178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019951069483645873,
+      "loss": 0.5756,
+      "step": 179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019949977010510445,
+      "loss": 0.537,
+      "step": 180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001994887250638689,
+      "loss": 0.5364,
+      "step": 181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019947755972610717,
+      "loss": 0.5098,
+      "step": 182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001994662741053198,
+      "loss": 0.4544,
+      "step": 183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019945486821515274,
+      "loss": 0.5581,
+      "step": 184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019944334206939746,
+      "loss": 0.391,
+      "step": 185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001994316956819907,
+      "loss": 0.4224,
+      "step": 186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001994199290670147,
+      "loss": 0.4177,
+      "step": 187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019940804223869704,
+      "loss": 0.4837,
+      "step": 188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019939603521141057,
+      "loss": 0.4524,
+      "step": 189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019938390799967363,
+      "loss": 0.3787,
+      "step": 190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001993716606181498,
+      "loss": 0.4168,
+      "step": 191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019935929308164793,
+      "loss": 0.4402,
+      "step": 192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019934680540512222,
+      "loss": 0.396,
+      "step": 193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019933419760367213,
+      "loss": 0.4407,
+      "step": 194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019932146969254232,
+      "loss": 0.4512,
+      "step": 195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019930862168712272,
+      "loss": 0.4237,
+      "step": 196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019929565360294845,
+      "loss": 0.3339,
+      "step": 197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019928256545569985,
+      "loss": 0.4368,
+      "step": 198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001992693572612024,
+      "loss": 0.5597,
+      "step": 199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019925602903542676,
+      "loss": 0.5745,
+      "step": 200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019924258079448873,
+      "loss": 0.4734,
+      "step": 201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019922901255464923,
+      "loss": 0.4999,
+      "step": 202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019921532433231422,
+      "loss": 0.4051,
+      "step": 203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001992015161440348,
+      "loss": 0.4044,
+      "step": 204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001991875880065071,
+      "loss": 0.4351,
+      "step": 205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019917353993657226,
+      "loss": 0.3806,
+      "step": 206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019915937195121654,
+      "loss": 0.4407,
+      "step": 207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019914508406757104,
+      "loss": 0.3591,
+      "step": 208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019913067630291197,
+      "loss": 0.415,
+      "step": 209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019911614867466044,
+      "loss": 0.4753,
+      "step": 210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019910150120038247,
+      "loss": 0.4452,
+      "step": 211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019908673389778901,
+      "loss": 0.5559,
+      "step": 212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019907184678473594,
+      "loss": 0.4722,
+      "step": 213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019905683987922392,
+      "loss": 0.3546,
+      "step": 214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019904171319939855,
+      "loss": 0.4172,
+      "step": 215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019902646676355022,
+      "loss": 0.4165,
+      "step": 216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001990111005901141,
+      "loss": 0.5215,
+      "step": 217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001989956146976701,
+      "loss": 0.3929,
+      "step": 218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019898000910494306,
+      "loss": 0.4312,
+      "step": 219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019896428383080238,
+      "loss": 0.3423,
+      "step": 220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019894843889426222,
+      "loss": 0.4239,
+      "step": 221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001989324743144815,
+      "loss": 0.3056,
+      "step": 222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019891639011076367,
+      "loss": 0.5578,
+      "step": 223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019890018630255694,
+      "loss": 0.3934,
+      "step": 224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001988838629094541,
+      "loss": 0.3376,
+      "step": 225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019886741995119256,
+      "loss": 0.3588,
+      "step": 226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019885085744765423,
+      "loss": 0.4654,
+      "step": 227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019883417541886569,
+      "loss": 0.4056,
+      "step": 228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019881737388499792,
+      "loss": 0.4981,
+      "step": 229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001988004528663664,
+      "loss": 0.5102,
+      "step": 230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019878341238343126,
+      "loss": 0.5584,
+      "step": 231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019876625245679686,
+      "loss": 0.4471,
+      "step": 232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001987489731072121,
+      "loss": 0.6286,
+      "step": 233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019873157435557027,
+      "loss": 0.5207,
+      "step": 234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019871405622290901,
+      "loss": 0.525,
+      "step": 235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019869641873041034,
+      "loss": 0.3923,
+      "step": 236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019867866189940056,
+      "loss": 0.3726,
+      "step": 237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019866078575135032,
+      "loss": 0.4186,
+      "step": 238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019864279030787452,
+      "loss": 0.4912,
+      "step": 239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001986246755907323,
+      "loss": 0.4511,
+      "step": 240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000198606441621827,
+      "loss": 0.5478,
+      "step": 241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019858808842320618,
+      "loss": 0.4366,
+      "step": 242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019856961601706158,
+      "loss": 0.4102,
+      "step": 243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019855102442572906,
+      "loss": 0.3902,
+      "step": 244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019853231367168857,
+      "loss": 0.4998,
+      "step": 245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019851348377756417,
+      "loss": 0.2967,
+      "step": 246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019849453476612403,
+      "loss": 0.3987,
+      "step": 247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019847546666028023,
+      "loss": 0.3314,
+      "step": 248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019845627948308895,
+      "loss": 0.4172,
+      "step": 249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001984369732577503,
+      "loss": 0.6093,
+      "step": 250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019841754800760833,
+      "loss": 0.4567,
+      "step": 251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019839800375615112,
+      "loss": 0.5522,
+      "step": 252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019837834052701046,
+      "loss": 0.4533,
+      "step": 253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001983585583439621,
+      "loss": 0.4319,
+      "step": 254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019833865723092563,
+      "loss": 0.5046,
+      "step": 255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019831863721196446,
+      "loss": 0.4786,
+      "step": 256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001982984983112857,
+      "loss": 0.4659,
+      "step": 257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019827824055324028,
+      "loss": 0.3657,
+      "step": 258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019825786396232274,
+      "loss": 0.4467,
+      "step": 259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019823736856317147,
+      "loss": 0.4263,
+      "step": 260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001982167543805684,
+      "loss": 0.4808,
+      "step": 261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001981960214394391,
+      "loss": 0.5833,
+      "step": 262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019817516976485276,
+      "loss": 0.4079,
+      "step": 263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001981541993820221,
+      "loss": 0.4657,
+      "step": 264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019813311031630345,
+      "loss": 0.3537,
+      "step": 265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019811190259319655,
+      "loss": 0.5174,
+      "step": 266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019809057623834472,
+      "loss": 0.4179,
+      "step": 267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001980691312775346,
+      "loss": 0.3337,
+      "step": 268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019804756773669633,
+      "loss": 0.3871,
+      "step": 269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019802588564190343,
+      "loss": 0.4062,
+      "step": 270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019800408501937272,
+      "loss": 0.409,
+      "step": 271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001979821658954644,
+      "loss": 0.3633,
+      "step": 272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019796012829668188,
+      "loss": 0.353,
+      "step": 273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019793797224967188,
+      "loss": 0.5171,
+      "step": 274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019791569778122433,
+      "loss": 0.4185,
+      "step": 275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001978933049182723,
+      "loss": 0.5375,
+      "step": 276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019787079368789216,
+      "loss": 0.417,
+      "step": 277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001978481641173032,
+      "loss": 0.3537,
+      "step": 278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019782541623386795,
+      "loss": 0.4069,
+      "step": 279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197802550065092,
+      "loss": 0.309,
+      "step": 280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019777956563862384,
+      "loss": 0.5332,
+      "step": 281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019775646298225504,
+      "loss": 0.5683,
+      "step": 282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019773324212392013,
+      "loss": 0.519,
+      "step": 283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019770990309169655,
+      "loss": 0.3606,
+      "step": 284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019768644591380464,
+      "loss": 0.5679,
+      "step": 285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019766287061860757,
+      "loss": 0.4989,
+      "step": 286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019763917723461134,
+      "loss": 0.3552,
+      "step": 287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019761536579046476,
+      "loss": 0.2874,
+      "step": 288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019759143631495937,
+      "loss": 0.4719,
+      "step": 289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001975673888370294,
+      "loss": 0.3916,
+      "step": 290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019754322338575185,
+      "loss": 0.4228,
+      "step": 291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019751893999034624,
+      "loss": 0.3805,
+      "step": 292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019749453868017484,
+      "loss": 0.4021,
+      "step": 293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001974700194847424,
+      "loss": 0.3616,
+      "step": 294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001974453824336962,
+      "loss": 0.4452,
+      "step": 295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019742062755682616,
+      "loss": 0.373,
+      "step": 296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973957548840645,
+      "loss": 0.418,
+      "step": 297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019737076444548593,
+      "loss": 0.3941,
+      "step": 298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019734565627130764,
+      "loss": 0.3521,
+      "step": 299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001973204303918891,
+      "loss": 0.5013,
+      "step": 300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019729508683773205,
+      "loss": 0.5282,
+      "step": 301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019726962563948065,
+      "loss": 0.4893,
+      "step": 302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019724404682792122,
+      "loss": 0.4586,
+      "step": 303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019721835043398227,
+      "loss": 0.454,
+      "step": 304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971925364887346,
+      "loss": 0.4473,
+      "step": 305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019716660502339098,
+      "loss": 0.5645,
+      "step": 306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019714055606930649,
+      "loss": 0.3188,
+      "step": 307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019711438965797803,
+      "loss": 0.3629,
+      "step": 308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019708810582104476,
+      "loss": 0.4478,
+      "step": 309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019706170459028762,
+      "loss": 0.4505,
+      "step": 310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019703518599762965,
+      "loss": 0.3777,
+      "step": 311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001970085500751357,
+      "loss": 0.3458,
+      "step": 312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019698179685501252,
+      "loss": 0.3768,
+      "step": 313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969549263696087,
+      "loss": 0.4302,
+      "step": 314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019692793865141469,
+      "loss": 0.4514,
+      "step": 315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969008337330625,
+      "loss": 0.4264,
+      "step": 316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019687361164732602,
+      "loss": 0.52,
+      "step": 317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019684627242712077,
+      "loss": 0.3915,
+      "step": 318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019681881610550387,
+      "loss": 0.4538,
+      "step": 319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019679124271567412,
+      "loss": 0.4411,
+      "step": 320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019676355229097174,
+      "loss": 0.4181,
+      "step": 321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001967357448648786,
+      "loss": 0.3414,
+      "step": 322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001967078204710179,
+      "loss": 0.4784,
+      "step": 323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019667977914315435,
+      "loss": 0.4811,
+      "step": 324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019665162091519412,
+      "loss": 0.3714,
+      "step": 325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966233458211846,
+      "loss": 0.4852,
+      "step": 326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001965949538953146,
+      "loss": 0.5778,
+      "step": 327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019656644517191408,
+      "loss": 0.4532,
+      "step": 328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019653781968545427,
+      "loss": 0.4616,
+      "step": 329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019650907747054768,
+      "loss": 0.5085,
+      "step": 330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019648021856194783,
+      "loss": 0.5269,
+      "step": 331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964512429945494,
+      "loss": 0.4043,
+      "step": 332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019642215080338814,
+      "loss": 0.3773,
+      "step": 333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019639294202364075,
+      "loss": 0.4352,
+      "step": 334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000196363616690625,
+      "loss": 0.4284,
+      "step": 335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001963341748397995,
+      "loss": 0.3828,
+      "step": 336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019630461650676378,
+      "loss": 0.3517,
+      "step": 337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019627494172725827,
+      "loss": 0.4708,
+      "step": 338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962451505371641,
+      "loss": 0.3365,
+      "step": 339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019621524297250322,
+      "loss": 0.4408,
+      "step": 340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019618521906943832,
+      "loss": 0.4585,
+      "step": 341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019615507886427266,
+      "loss": 0.3419,
+      "step": 342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019612482239345024,
+      "loss": 0.4281,
+      "step": 343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019609444969355557,
+      "loss": 0.4062,
+      "step": 344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019606396080131374,
+      "loss": 0.3166,
+      "step": 345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019603335575359024,
+      "loss": 0.3492,
+      "step": 346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019600263458739118,
+      "loss": 0.4103,
+      "step": 347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001959717973398629,
+      "loss": 0.4705,
+      "step": 348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001959408440482922,
+      "loss": 0.343,
+      "step": 349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001959097747501062,
+      "loss": 0.5649,
+      "step": 350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001958785894828722,
+      "loss": 0.5357,
+      "step": 351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019584728828429785,
+      "loss": 0.5167,
+      "step": 352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019581587119223089,
+      "loss": 0.4339,
+      "step": 353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019578433824465915,
+      "loss": 0.4218,
+      "step": 354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001957526894797107,
+      "loss": 0.4531,
+      "step": 355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019572092493565352,
+      "loss": 0.5455,
+      "step": 356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019568904465089561,
+      "loss": 0.521,
+      "step": 357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019565704866398494,
+      "loss": 0.5418,
+      "step": 358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019562493701360937,
+      "loss": 0.3689,
+      "step": 359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019559270973859664,
+      "loss": 0.4334,
+      "step": 360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019556036687791423,
+      "loss": 0.4064,
+      "step": 361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001955279084706695,
+      "loss": 0.5414,
+      "step": 362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019549533455610934,
+      "loss": 0.4393,
+      "step": 363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001954626451736205,
+      "loss": 0.4307,
+      "step": 364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019542984036272922,
+      "loss": 0.5857,
+      "step": 365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019539692016310137,
+      "loss": 0.4936,
+      "step": 366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019536388461454228,
+      "loss": 0.3318,
+      "step": 367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019533073375699683,
+      "loss": 0.4419,
+      "step": 368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019529746763054928,
+      "loss": 0.4549,
+      "step": 369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001952640862754233,
+      "loss": 0.3202,
+      "step": 370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019523058973198185,
+      "loss": 0.466,
+      "step": 371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019519697804072718,
+      "loss": 0.2524,
+      "step": 372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019516325124230077,
+      "loss": 0.4931,
+      "step": 373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019512940937748332,
+      "loss": 0.3946,
+      "step": 374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019509545248719458,
+      "loss": 0.4788,
+      "step": 375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001950613806124935,
+      "loss": 0.499,
+      "step": 376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019502719379457792,
+      "loss": 0.3752,
+      "step": 377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019499289207478482,
+      "loss": 0.4115,
+      "step": 378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019495847549458995,
+      "loss": 0.4688,
+      "step": 379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001949239440956081,
+      "loss": 0.3994,
+      "step": 380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001948892979195928,
+      "loss": 0.4946,
+      "step": 381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001948545370084363,
+      "loss": 0.3617,
+      "step": 382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001948196614041698,
+      "loss": 0.5197,
+      "step": 383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019478467114896294,
+      "loss": 0.5431,
+      "step": 384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019474956628512413,
+      "loss": 0.3986,
+      "step": 385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019471434685510033,
+      "loss": 0.4114,
+      "step": 386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019467901290147702,
+      "loss": 0.4039,
+      "step": 387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019464356446697814,
+      "loss": 0.3551,
+      "step": 388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019460800159446607,
+      "loss": 0.5216,
+      "step": 389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019457232432694156,
+      "loss": 0.4784,
+      "step": 390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001945365327075437,
+      "loss": 0.3702,
+      "step": 391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019450062677954982,
+      "loss": 0.5012,
+      "step": 392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019446460658637546,
+      "loss": 0.4606,
+      "step": 393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019442847217157437,
+      "loss": 0.2874,
+      "step": 394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001943922235788384,
+      "loss": 0.2891,
+      "step": 395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019435586085199738,
+      "loss": 0.3277,
+      "step": 396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019431938403501926,
+      "loss": 0.361,
+      "step": 397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019428279317200984,
+      "loss": 0.3659,
+      "step": 398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019424608830721288,
+      "loss": 0.2845,
+      "step": 399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019420926948501,
+      "loss": 0.7692,
+      "step": 400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019417233674992053,
+      "loss": 0.4799,
+      "step": 401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019413529014660158,
+      "loss": 0.4417,
+      "step": 402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019409812971984803,
+      "loss": 0.4223,
+      "step": 403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019406085551459227,
+      "loss": 0.4718,
+      "step": 404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019402346757590427,
+      "loss": 0.4381,
+      "step": 405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001939859659489916,
+      "loss": 0.3665,
+      "step": 406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019394835067919924,
+      "loss": 0.4075,
+      "step": 407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019391062181200956,
+      "loss": 0.3827,
+      "step": 408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001938727793930424,
+      "loss": 0.4478,
+      "step": 409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019383482346805472,
+      "loss": 0.3498,
+      "step": 410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019379675408294088,
+      "loss": 0.4023,
+      "step": 411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001937585712837324,
+      "loss": 0.4118,
+      "step": 412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019372027511659786,
+      "loss": 0.3455,
+      "step": 413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000193681865627843,
+      "loss": 0.5404,
+      "step": 414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019364334286391054,
+      "loss": 0.4631,
+      "step": 415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019360470687138017,
+      "loss": 0.4348,
+      "step": 416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019356595769696848,
+      "loss": 0.4193,
+      "step": 417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000193527095387529,
+      "loss": 0.4555,
+      "step": 418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019348811999005193,
+      "loss": 0.4824,
+      "step": 419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019344903155166425,
+      "loss": 0.498,
+      "step": 420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019340983011962968,
+      "loss": 0.5353,
+      "step": 421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019337051574134853,
+      "loss": 0.4089,
+      "step": 422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019333108846435765,
+      "loss": 0.2933,
+      "step": 423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019329154833633042,
+      "loss": 0.4465,
+      "step": 424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019325189540507674,
+      "loss": 0.4347,
+      "step": 425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001932121297185428,
+      "loss": 0.435,
+      "step": 426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019317225132481114,
+      "loss": 0.4328,
+      "step": 427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019313226027210069,
+      "loss": 0.361,
+      "step": 428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019309215660876644,
+      "loss": 0.4814,
+      "step": 429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019305194038329973,
+      "loss": 0.4327,
+      "step": 430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001930116116443278,
+      "loss": 0.4087,
+      "step": 431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019297117044061413,
+      "loss": 0.4899,
+      "step": 432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019293061682105803,
+      "loss": 0.5218,
+      "step": 433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001928899508346948,
+      "loss": 0.3464,
+      "step": 434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001928491725306956,
+      "loss": 0.4049,
+      "step": 435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019280828195836748,
+      "loss": 0.4296,
+      "step": 436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019276727916715312,
+      "loss": 0.4275,
+      "step": 437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019272616420663093,
+      "loss": 0.385,
+      "step": 438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001926849371265149,
+      "loss": 0.272,
+      "step": 439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019264359797665478,
+      "loss": 0.4677,
+      "step": 440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001926021468070356,
+      "loss": 0.3223,
+      "step": 441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001925605836677779,
+      "loss": 0.3061,
+      "step": 442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019251890860913775,
+      "loss": 0.374,
+      "step": 443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019247712168150635,
+      "loss": 0.4109,
+      "step": 444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019243522293541025,
+      "loss": 0.3597,
+      "step": 445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019239321242151125,
+      "loss": 0.3735,
+      "step": 446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019235109019060623,
+      "loss": 0.4416,
+      "step": 447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019230885629362718,
+      "loss": 0.2686,
+      "step": 448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001922665107816411,
+      "loss": 0.2857,
+      "step": 449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019222405370584992,
+      "loss": 0.9318,
+      "step": 450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019218148511759053,
+      "loss": 0.395,
+      "step": 451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019213880506833465,
+      "loss": 0.3878,
+      "step": 452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001920960136096887,
+      "loss": 0.385,
+      "step": 453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001920531107933939,
+      "loss": 0.4308,
+      "step": 454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019201009667132608,
+      "loss": 0.3984,
+      "step": 455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001919669712954956,
+      "loss": 0.4595,
+      "step": 456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001919237347180474,
+      "loss": 0.4033,
+      "step": 457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001918803869912609,
+      "loss": 0.3281,
+      "step": 458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019183692816754992,
+      "loss": 0.407,
+      "step": 459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019179335829946253,
+      "loss": 0.434,
+      "step": 460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001917496774396811,
+      "loss": 0.3835,
+      "step": 461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019170588564102227,
+      "loss": 0.4008,
+      "step": 462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001916619829564368,
+      "loss": 0.4059,
+      "step": 463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001916179694390094,
+      "loss": 0.3444,
+      "step": 464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000191573845141959,
+      "loss": 0.3398,
+      "step": 465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001915296101186383,
+      "loss": 0.4739,
+      "step": 466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019148526442253405,
+      "loss": 0.3859,
+      "step": 467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019144080810726665,
+      "loss": 0.3754,
+      "step": 468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019139624122659036,
+      "loss": 0.3341,
+      "step": 469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019135156383439313,
+      "loss": 0.5296,
+      "step": 470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019130677598469648,
+      "loss": 0.375,
+      "step": 471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001912618777316555,
+      "loss": 0.3603,
+      "step": 472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019121686912955887,
+      "loss": 0.3545,
+      "step": 473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019117175023282851,
+      "loss": 0.4268,
+      "step": 474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019112652109601994,
+      "loss": 0.405,
+      "step": 475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019108118177382175,
+      "loss": 0.426,
+      "step": 476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019103573232105586,
+      "loss": 0.4785,
+      "step": 477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019099017279267745,
+      "loss": 0.5296,
+      "step": 478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001909445032437746,
+      "loss": 0.5151,
+      "step": 479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001908987237295686,
+      "loss": 0.2981,
+      "step": 480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019085283430541355,
+      "loss": 0.32,
+      "step": 481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019080683502679659,
+      "loss": 0.3759,
+      "step": 482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001907607259493376,
+      "loss": 0.4676,
+      "step": 483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019071450712878924,
+      "loss": 0.4307,
+      "step": 484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001906681786210369,
+      "loss": 0.5253,
+      "step": 485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019062174048209855,
+      "loss": 0.3549,
+      "step": 486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019057519276812472,
+      "loss": 0.4329,
+      "step": 487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001905285355353985,
+      "loss": 0.389,
+      "step": 488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019048176884033533,
+      "loss": 0.5373,
+      "step": 489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019043489273948305,
+      "loss": 0.3532,
+      "step": 490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019038790728952173,
+      "loss": 0.242,
+      "step": 491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001903408125472638,
+      "loss": 0.3831,
+      "step": 492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019029360856965362,
+      "loss": 0.5659,
+      "step": 493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001902462954137678,
+      "loss": 0.3011,
+      "step": 494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000190198873136815,
+      "loss": 0.3155,
+      "step": 495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019015134179613561,
+      "loss": 0.3373,
+      "step": 496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019010370144920211,
+      "loss": 0.2781,
+      "step": 497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019005595215361863,
+      "loss": 0.2951,
+      "step": 498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019000809396712117,
+      "loss": 0.3324,
+      "step": 499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001899601269475773,
+      "loss": 0.3,
+      "step": 500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001899120511529862,
+      "loss": 0.4367,
+      "step": 501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898638666414786,
+      "loss": 0.545,
+      "step": 502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018981557347131668,
+      "loss": 0.4514,
+      "step": 503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018976717170089398,
+      "loss": 0.432,
+      "step": 504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001897186613887354,
+      "loss": 0.4117,
+      "step": 505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018967004259349705,
+      "loss": 0.4019,
+      "step": 506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018962131537396618,
+      "loss": 0.3003,
+      "step": 507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018957247978906123,
+      "loss": 0.3648,
+      "step": 508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018952353589783157,
+      "loss": 0.3498,
+      "step": 509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001894744837594576,
+      "loss": 0.3418,
+      "step": 510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018942532343325057,
+      "loss": 0.3531,
+      "step": 511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018937605497865257,
+      "loss": 0.3959,
+      "step": 512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001893266784552364,
+      "loss": 0.3913,
+      "step": 513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001892771939227056,
+      "loss": 0.4228,
+      "step": 514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001892276014408942,
+      "loss": 0.318,
+      "step": 515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018917790106976682,
+      "loss": 0.4663,
+      "step": 516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001891280928694186,
+      "loss": 0.3263,
+      "step": 517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018907817690007493,
+      "loss": 0.3645,
+      "step": 518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018902815322209158,
+      "loss": 0.3845,
+      "step": 519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018897802189595458,
+      "loss": 0.535,
+      "step": 520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018892778298228003,
+      "loss": 0.5016,
+      "step": 521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018887743654181423,
+      "loss": 0.4374,
+      "step": 522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001888269826354334,
+      "loss": 0.4395,
+      "step": 523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018877642132414382,
+      "loss": 0.3356,
+      "step": 524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018872575266908149,
+      "loss": 0.5864,
+      "step": 525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001886749767315123,
+      "loss": 0.4755,
+      "step": 526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018862409357283183,
+      "loss": 0.4566,
+      "step": 527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018857310325456536,
+      "loss": 0.3683,
+      "step": 528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018852200583836765,
+      "loss": 0.4402,
+      "step": 529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018847080138602302,
+      "loss": 0.4491,
+      "step": 530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001884194899594452,
+      "loss": 0.3004,
+      "step": 531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018836807162067725,
+      "loss": 0.3436,
+      "step": 532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018831654643189157,
+      "loss": 0.3971,
+      "step": 533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018826491445538964,
+      "loss": 0.3707,
+      "step": 534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018821317575360214,
+      "loss": 0.2896,
+      "step": 535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018816133038908878,
+      "loss": 0.3704,
+      "step": 536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018810937842453828,
+      "loss": 0.3768,
+      "step": 537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001880573199227682,
+      "loss": 0.2844,
+      "step": 538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018800515494672495,
+      "loss": 0.245,
+      "step": 539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018795288355948363,
+      "loss": 0.3722,
+      "step": 540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001879005058242481,
+      "loss": 0.5425,
+      "step": 541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018784802180435068,
+      "loss": 0.3674,
+      "step": 542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018779543156325238,
+      "loss": 0.375,
+      "step": 543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001877427351645425,
+      "loss": 0.228,
+      "step": 544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018768993267193876,
+      "loss": 0.4299,
+      "step": 545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001876370241492871,
+      "loss": 0.382,
+      "step": 546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001875840096605618,
+      "loss": 0.2571,
+      "step": 547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018753088926986517,
+      "loss": 0.3124,
+      "step": 548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018747766304142758,
+      "loss": 0.2887,
+      "step": 549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018742433103960732,
+      "loss": 0.5482,
+      "step": 550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018737089332889075,
+      "loss": 0.3812,
+      "step": 551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018731734997389185,
+      "loss": 0.5179,
+      "step": 552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018726370103935245,
+      "loss": 0.4855,
+      "step": 553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018720994659014205,
+      "loss": 0.4474,
+      "step": 554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018715608669125762,
+      "loss": 0.5144,
+      "step": 555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018710212140782376,
+      "loss": 0.4499,
+      "step": 556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001870480508050925,
+      "loss": 0.4387,
+      "step": 557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018699387494844306,
+      "loss": 0.4309,
+      "step": 558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018693959390338213,
+      "loss": 0.4505,
+      "step": 559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018688520773554342,
+      "loss": 0.4847,
+      "step": 560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018683071651068783,
+      "loss": 0.4062,
+      "step": 561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018677612029470333,
+      "loss": 0.4036,
+      "step": 562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018672141915360472,
+      "loss": 0.413,
+      "step": 563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018666661315353377,
+      "loss": 0.4223,
+      "step": 564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018661170236075904,
+      "loss": 0.4855,
+      "step": 565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018655668684167573,
+      "loss": 0.4075,
+      "step": 566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001865015666628057,
+      "loss": 0.3258,
+      "step": 567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018644634189079738,
+      "loss": 0.4118,
+      "step": 568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018639101259242572,
+      "loss": 0.3621,
+      "step": 569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018633557883459187,
+      "loss": 0.2893,
+      "step": 570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018628004068432346,
+      "loss": 0.3627,
+      "step": 571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018622439820877436,
+      "loss": 0.4321,
+      "step": 572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018616865147522447,
+      "loss": 0.333,
+      "step": 573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018611280055107977,
+      "loss": 0.3765,
+      "step": 574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001860568455038723,
+      "loss": 0.3856,
+      "step": 575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018600078640125994,
+      "loss": 0.3122,
+      "step": 576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018594462331102642,
+      "loss": 0.3178,
+      "step": 577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001858883563010812,
+      "loss": 0.353,
+      "step": 578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018583198543945933,
+      "loss": 0.4896,
+      "step": 579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018577551079432153,
+      "loss": 0.3842,
+      "step": 580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018571893243395394,
+      "loss": 0.417,
+      "step": 581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018566225042676817,
+      "loss": 0.401,
+      "step": 582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001856054648413011,
+      "loss": 0.379,
+      "step": 583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001855485757462149,
+      "loss": 0.3958,
+      "step": 584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018549158321029676,
+      "loss": 0.3901,
+      "step": 585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001854344873024592,
+      "loss": 0.4435,
+      "step": 586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018537728809173945,
+      "loss": 0.3426,
+      "step": 587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001853199856472999,
+      "loss": 0.424,
+      "step": 588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018526258003842756,
+      "loss": 0.4306,
+      "step": 589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001852050713345343,
+      "loss": 0.2639,
+      "step": 590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018514745960515665,
+      "loss": 0.3712,
+      "step": 591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001850897449199556,
+      "loss": 0.3388,
+      "step": 592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001850319273487168,
+      "loss": 0.4338,
+      "step": 593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018497400696135014,
+      "loss": 0.4764,
+      "step": 594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018491598382788994,
+      "loss": 0.4954,
+      "step": 595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018485785801849466,
+      "loss": 0.3732,
+      "step": 596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018479962960344705,
+      "loss": 0.278,
+      "step": 597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018474129865315381,
+      "loss": 0.2712,
+      "step": 598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018468286523814565,
+      "loss": 0.2711,
+      "step": 599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018462432942907713,
+      "loss": 0.5263,
+      "step": 600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018456569129672674,
+      "loss": 0.3766,
+      "step": 601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018450695091199658,
+      "loss": 0.3868,
+      "step": 602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018444810834591246,
+      "loss": 0.3897,
+      "step": 603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001843891636696237,
+      "loss": 0.4674,
+      "step": 604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018433011695440302,
+      "loss": 0.3578,
+      "step": 605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018427096827164675,
+      "loss": 0.4128,
+      "step": 606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018421171769287425,
+      "loss": 0.3767,
+      "step": 607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018415236528972822,
+      "loss": 0.345,
+      "step": 608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001840929111339745,
+      "loss": 0.4297,
+      "step": 609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018403335529750185,
+      "loss": 0.4425,
+      "step": 610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018397369785232213,
+      "loss": 0.316,
+      "step": 611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018391393887056992,
+      "loss": 0.2888,
+      "step": 612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018385407842450268,
+      "loss": 0.3326,
+      "step": 613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018379411658650047,
+      "loss": 0.3513,
+      "step": 614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000183734053429066,
+      "loss": 0.3943,
+      "step": 615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001836738890248245,
+      "loss": 0.4231,
+      "step": 616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018361362344652355,
+      "loss": 0.3825,
+      "step": 617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018355325676703315,
+      "loss": 0.3588,
+      "step": 618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001834927890593455,
+      "loss": 0.3022,
+      "step": 619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018343222039657496,
+      "loss": 0.3088,
+      "step": 620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018337155085195798,
+      "loss": 0.3337,
+      "step": 621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018331078049885297,
+      "loss": 0.3034,
+      "step": 622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018324990941074026,
+      "loss": 0.3297,
+      "step": 623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018318893766122192,
+      "loss": 0.2957,
+      "step": 624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018312786532402183,
+      "loss": 0.4076,
+      "step": 625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018306669247298535,
+      "loss": 0.3678,
+      "step": 626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018300541918207962,
+      "loss": 0.3714,
+      "step": 627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018294404552539293,
+      "loss": 0.4179,
+      "step": 628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018288257157713517,
+      "loss": 0.357,
+      "step": 629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018282099741163737,
+      "loss": 0.2943,
+      "step": 630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018275932310335172,
+      "loss": 0.3234,
+      "step": 631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018269754872685164,
+      "loss": 0.2383,
+      "step": 632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018263567435683136,
+      "loss": 0.4156,
+      "step": 633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018257370006810616,
+      "loss": 0.3682,
+      "step": 634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001825116259356121,
+      "loss": 0.4156,
+      "step": 635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018244945203440592,
+      "loss": 0.4449,
+      "step": 636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000182387178439665,
+      "loss": 0.3236,
+      "step": 637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018232480522668737,
+      "loss": 0.5582,
+      "step": 638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018226233247089132,
+      "loss": 0.3678,
+      "step": 639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018219976024781576,
+      "loss": 0.3889,
+      "step": 640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001821370886331196,
+      "loss": 0.3749,
+      "step": 641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018207431770258213,
+      "loss": 0.4096,
+      "step": 642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018201144753210262,
+      "loss": 0.269,
+      "step": 643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018194847819770038,
+      "loss": 0.2762,
+      "step": 644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018188540977551464,
+      "loss": 0.3453,
+      "step": 645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018182224234180433,
+      "loss": 0.3538,
+      "step": 646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018175897597294825,
+      "loss": 0.3154,
+      "step": 647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018169561074544478,
+      "loss": 0.3902,
+      "step": 648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018163214673591176,
+      "loss": 0.283,
+      "step": 649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018156858402108657,
+      "loss": 0.3136,
+      "step": 650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018150492267782586,
+      "loss": 0.4009,
+      "step": 651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018144116278310558,
+      "loss": 0.4286,
+      "step": 652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018137730441402088,
+      "loss": 0.3668,
+      "step": 653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018131334764778592,
+      "loss": 0.3464,
+      "step": 654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001812492925617338,
+      "loss": 0.5536,
+      "step": 655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001811851392333166,
+      "loss": 0.4047,
+      "step": 656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018112088774010517,
+      "loss": 0.3952,
+      "step": 657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018105653815978894,
+      "loss": 0.3189,
+      "step": 658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018099209057017613,
+      "loss": 0.3845,
+      "step": 659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018092754504919336,
+      "loss": 0.3866,
+      "step": 660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001808629016748856,
+      "loss": 0.3441,
+      "step": 661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001807981605254163,
+      "loss": 0.4593,
+      "step": 662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018073332167906702,
+      "loss": 0.402,
+      "step": 663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018066838521423745,
+      "loss": 0.4783,
+      "step": 664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018060335120944533,
+      "loss": 0.3166,
+      "step": 665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018053821974332643,
+      "loss": 0.4376,
+      "step": 666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018047299089463424,
+      "loss": 0.375,
+      "step": 667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018040766474224004,
+      "loss": 0.4071,
+      "step": 668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018034224136513277,
+      "loss": 0.3457,
+      "step": 669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018027672084241898,
+      "loss": 0.3633,
+      "step": 670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018021110325332259,
+      "loss": 0.3447,
+      "step": 671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018014538867718494,
+      "loss": 0.3182,
+      "step": 672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001800795771934646,
+      "loss": 0.4514,
+      "step": 673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018001366888173746,
+      "loss": 0.3331,
+      "step": 674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017994766382169625,
+      "loss": 0.3349,
+      "step": 675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017988156209315087,
+      "loss": 0.3116,
+      "step": 676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017981536377602807,
+      "loss": 0.4662,
+      "step": 677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017974906895037136,
+      "loss": 0.3421,
+      "step": 678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001796826776963409,
+      "loss": 0.4823,
+      "step": 679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017961619009421357,
+      "loss": 0.3948,
+      "step": 680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001795496062243826,
+      "loss": 0.2809,
+      "step": 681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017948292616735777,
+      "loss": 0.3925,
+      "step": 682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017941615000376506,
+      "loss": 0.4587,
+      "step": 683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017934927781434668,
+      "loss": 0.4064,
+      "step": 684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000179282309679961,
+      "loss": 0.3433,
+      "step": 685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017921524568158227,
+      "loss": 0.3937,
+      "step": 686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017914808590030088,
+      "loss": 0.4832,
+      "step": 687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017908083041732277,
+      "loss": 0.3243,
+      "step": 688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001790134793139698,
+      "loss": 0.5036,
+      "step": 689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017894603267167939,
+      "loss": 0.2923,
+      "step": 690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001788784905720044,
+      "loss": 0.1905,
+      "step": 691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017881085309661324,
+      "loss": 0.2943,
+      "step": 692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017874312032728956,
+      "loss": 0.3385,
+      "step": 693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017867529234593223,
+      "loss": 0.2446,
+      "step": 694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017860736923455534,
+      "loss": 0.3791,
+      "step": 695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017853935107528787,
+      "loss": 0.28,
+      "step": 696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017847123795037385,
+      "loss": 0.3019,
+      "step": 697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017840302994217205,
+      "loss": 0.315,
+      "step": 698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017833472713315598,
+      "loss": 0.2653,
+      "step": 699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017826632960591386,
+      "loss": 0.3417,
+      "step": 700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017819783744314833,
+      "loss": 0.4163,
+      "step": 701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001781292507276765,
+      "loss": 0.4507,
+      "step": 702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017806056954242986,
+      "loss": 0.3128,
+      "step": 703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017799179397045406,
+      "loss": 0.3533,
+      "step": 704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001779229240949089,
+      "loss": 0.3978,
+      "step": 705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001778539599990682,
+      "loss": 0.4482,
+      "step": 706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001777849017663197,
+      "loss": 0.3429,
+      "step": 707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017771574948016497,
+      "loss": 0.3317,
+      "step": 708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017764650322421937,
+      "loss": 0.2927,
+      "step": 709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001775771630822118,
+      "loss": 0.3654,
+      "step": 710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017750772913798467,
+      "loss": 0.3889,
+      "step": 711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017743820147549387,
+      "loss": 0.356,
+      "step": 712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017736858017880857,
+      "loss": 0.4042,
+      "step": 713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017729886533211127,
+      "loss": 0.5013,
+      "step": 714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017722905701969736,
+      "loss": 0.346,
+      "step": 715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017715915532597545,
+      "loss": 0.3422,
+      "step": 716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017708916033546697,
+      "loss": 0.3275,
+      "step": 717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017701907213280615,
+      "loss": 0.42,
+      "step": 718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017694889080274,
+      "loss": 0.3876,
+      "step": 719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017687861643012808,
+      "loss": 0.412,
+      "step": 720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001768082490999424,
+      "loss": 0.3224,
+      "step": 721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001767377888972675,
+      "loss": 0.3564,
+      "step": 722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017666723590730013,
+      "loss": 0.2865,
+      "step": 723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017659659021534922,
+      "loss": 0.4786,
+      "step": 724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017652585190683584,
+      "loss": 0.3935,
+      "step": 725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017645502106729307,
+      "loss": 0.4762,
+      "step": 726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017638409778236572,
+      "loss": 0.3299,
+      "step": 727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001763130821378106,
+      "loss": 0.5845,
+      "step": 728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017624197421949603,
+      "loss": 0.3045,
+      "step": 729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017617077411340198,
+      "loss": 0.2617,
+      "step": 730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017609948190561986,
+      "loss": 0.387,
+      "step": 731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017602809768235242,
+      "loss": 0.4581,
+      "step": 732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017595662152991378,
+      "loss": 0.3329,
+      "step": 733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001758850535347291,
+      "loss": 0.4211,
+      "step": 734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017581339378333459,
+      "loss": 0.332,
+      "step": 735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001757416423623775,
+      "loss": 0.3718,
+      "step": 736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017566979935861585,
+      "loss": 0.2627,
+      "step": 737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017559786485891841,
+      "loss": 0.3988,
+      "step": 738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017552583895026462,
+      "loss": 0.3335,
+      "step": 739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017545372171974438,
+      "loss": 0.3251,
+      "step": 740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017538151325455808,
+      "loss": 0.3775,
+      "step": 741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017530921364201633,
+      "loss": 0.3454,
+      "step": 742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017523682296954006,
+      "loss": 0.3718,
+      "step": 743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017516434132466027,
+      "loss": 0.2185,
+      "step": 744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017509176879501793,
+      "loss": 0.3144,
+      "step": 745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001750191054683639,
+      "loss": 0.2662,
+      "step": 746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017494635143255888,
+      "loss": 0.2622,
+      "step": 747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017487350677557322,
+      "loss": 0.2455,
+      "step": 748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001748005715854868,
+      "loss": 0.4092,
+      "step": 749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001747275459504891,
+      "loss": 0.8742,
+      "step": 750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017465442995887872,
+      "loss": 0.4034,
+      "step": 751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017458122369906382,
+      "loss": 0.4462,
+      "step": 752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017450792725956147,
+      "loss": 0.3813,
+      "step": 753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017443454072899788,
+      "loss": 0.3834,
+      "step": 754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001743610641961082,
+      "loss": 0.337,
+      "step": 755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001742874977497363,
+      "loss": 0.3174,
+      "step": 756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017421384147883498,
+      "loss": 0.2568,
+      "step": 757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017414009547246545,
+      "loss": 0.464,
+      "step": 758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017406625981979752,
+      "loss": 0.3422,
+      "step": 759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017399233461010933,
+      "loss": 0.3478,
+      "step": 760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017391831993278739,
+      "loss": 0.47,
+      "step": 761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001738442158773263,
+      "loss": 0.4094,
+      "step": 762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001737700225333289,
+      "loss": 0.4129,
+      "step": 763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017369573999050572,
+      "loss": 0.3324,
+      "step": 764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001736213683386754,
+      "loss": 0.3435,
+      "step": 765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017354690766776417,
+      "loss": 0.3641,
+      "step": 766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017347235806780602,
+      "loss": 0.4173,
+      "step": 767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017339771962894234,
+      "loss": 0.3912,
+      "step": 768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000173322992441422,
+      "loss": 0.5109,
+      "step": 769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017324817659560118,
+      "loss": 0.352,
+      "step": 770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017317327218194332,
+      "loss": 0.5482,
+      "step": 771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017309827929101878,
+      "loss": 0.3461,
+      "step": 772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017302319801350513,
+      "loss": 0.2916,
+      "step": 773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017294802844018662,
+      "loss": 0.2628,
+      "step": 774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017287277066195439,
+      "loss": 0.3449,
+      "step": 775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017279742476980614,
+      "loss": 0.227,
+      "step": 776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017272199085484617,
+      "loss": 0.3323,
+      "step": 777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017264646900828524,
+      "loss": 0.3537,
+      "step": 778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017257085932144027,
+      "loss": 0.413,
+      "step": 779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017249516188573463,
+      "loss": 0.3727,
+      "step": 780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017241937679269765,
+      "loss": 0.2764,
+      "step": 781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001723435041339646,
+      "loss": 0.5259,
+      "step": 782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001722675440012768,
+      "loss": 0.5365,
+      "step": 783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017219149648648116,
+      "loss": 0.3736,
+      "step": 784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017211536168153038,
+      "loss": 0.3713,
+      "step": 785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017203913967848264,
+      "loss": 0.2164,
+      "step": 786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017196283056950158,
+      "loss": 0.3771,
+      "step": 787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017188643444685612,
+      "loss": 0.3066,
+      "step": 788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001718099514029205,
+      "loss": 0.2483,
+      "step": 789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001717333815301739,
+      "loss": 0.3877,
+      "step": 790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017165672492120066,
+      "loss": 0.3163,
+      "step": 791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017157998166868988,
+      "loss": 0.302,
+      "step": 792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017150315186543542,
+      "loss": 0.3446,
+      "step": 793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001714262356043359,
+      "loss": 0.343,
+      "step": 794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017134923297839434,
+      "loss": 0.2408,
+      "step": 795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001712721440807183,
+      "loss": 0.2744,
+      "step": 796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017119496900451964,
+      "loss": 0.2849,
+      "step": 797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017111770784311437,
+      "loss": 0.2548,
+      "step": 798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017104036068992258,
+      "loss": 0.2818,
+      "step": 799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001709629276384684,
+      "loss": 0.5474,
+      "step": 800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017088540878237987,
+      "loss": 0.5003,
+      "step": 801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017080780421538857,
+      "loss": 0.3725,
+      "step": 802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017073011403133,
+      "loss": 0.4113,
+      "step": 803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001706523383241429,
+      "loss": 0.3415,
+      "step": 804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001705744771878697,
+      "loss": 0.4306,
+      "step": 805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017049653071665586,
+      "loss": 0.2534,
+      "step": 806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017041849900475023,
+      "loss": 0.3514,
+      "step": 807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001703403821465046,
+      "loss": 0.5084,
+      "step": 808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017026218023637375,
+      "loss": 0.294,
+      "step": 809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017018389336891535,
+      "loss": 0.4401,
+      "step": 810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017010552163878977,
+      "loss": 0.3045,
+      "step": 811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017002706514075992,
+      "loss": 0.2121,
+      "step": 812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016994852396969132,
+      "loss": 0.2838,
+      "step": 813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016986989822055177,
+      "loss": 0.2697,
+      "step": 814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016979118798841143,
+      "loss": 0.3614,
+      "step": 815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016971239336844253,
+      "loss": 0.4153,
+      "step": 816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001696335144559194,
+      "loss": 0.4897,
+      "step": 817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016955455134621826,
+      "loss": 0.2936,
+      "step": 818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001694755041348171,
+      "loss": 0.3607,
+      "step": 819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016939637291729572,
+      "loss": 0.3622,
+      "step": 820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016931715778933535,
+      "loss": 0.2779,
+      "step": 821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016923785884671885,
+      "loss": 0.384,
+      "step": 822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016915847618533022,
+      "loss": 0.4317,
+      "step": 823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001690790099011548,
+      "loss": 0.3363,
+      "step": 824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001689994600902791,
+      "loss": 0.4145,
+      "step": 825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001689198268488905,
+      "loss": 0.2636,
+      "step": 826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001688401102732774,
+      "loss": 0.2694,
+      "step": 827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001687603104598288,
+      "loss": 0.271,
+      "step": 828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016868042750503448,
+      "loss": 0.488,
+      "step": 829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016860046150548469,
+      "loss": 0.3742,
+      "step": 830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016852041255787005,
+      "loss": 0.3677,
+      "step": 831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016844028075898164,
+      "loss": 0.3087,
+      "step": 832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016836006620571053,
+      "loss": 0.4976,
+      "step": 833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016827976899504797,
+      "loss": 0.3101,
+      "step": 834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001681993892240851,
+      "loss": 0.3284,
+      "step": 835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001681189269900129,
+      "loss": 0.4635,
+      "step": 836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016803838239012206,
+      "loss": 0.3172,
+      "step": 837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001679577555218029,
+      "loss": 0.5652,
+      "step": 838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016787704648254517,
+      "loss": 0.2554,
+      "step": 839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016779625536993798,
+      "loss": 0.224,
+      "step": 840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016771538228166967,
+      "loss": 0.3557,
+      "step": 841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016763442731552777,
+      "loss": 0.322,
+      "step": 842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001675533905693987,
+      "loss": 0.3063,
+      "step": 843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001674722721412679,
+      "loss": 0.2217,
+      "step": 844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016739107212921942,
+      "loss": 0.1963,
+      "step": 845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016730979063143611,
+      "loss": 0.2483,
+      "step": 846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001672284277461992,
+      "loss": 0.2986,
+      "step": 847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001671469835718885,
+      "loss": 0.3508,
+      "step": 848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001670654582069819,
+      "loss": 0.2871,
+      "step": 849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016698385175005566,
+      "loss": 0.2691,
+      "step": 850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016690216429978403,
+      "loss": 0.4691,
+      "step": 851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001668203959549391,
+      "loss": 0.3946,
+      "step": 852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016673854681439085,
+      "loss": 0.3967,
+      "step": 853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016665661697710702,
+      "loss": 0.3944,
+      "step": 854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016657460654215277,
+      "loss": 0.3377,
+      "step": 855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001664925156086909,
+      "loss": 0.353,
+      "step": 856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001664103442759813,
+      "loss": 0.3681,
+      "step": 857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001663280926433814,
+      "loss": 0.3169,
+      "step": 858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016624576081034537,
+      "loss": 0.4421,
+      "step": 859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016616334887642462,
+      "loss": 0.2646,
+      "step": 860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001660808569412673,
+      "loss": 0.3663,
+      "step": 861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001659982851046183,
+      "loss": 0.3365,
+      "step": 862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016591563346631913,
+      "loss": 0.4046,
+      "step": 863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016583290212630783,
+      "loss": 0.38,
+      "step": 864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016575009118461877,
+      "loss": 0.3058,
+      "step": 865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016566720074138257,
+      "loss": 0.4012,
+      "step": 866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016558423089682597,
+      "loss": 0.4099,
+      "step": 867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016550118175127175,
+      "loss": 0.2995,
+      "step": 868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001654180534051386,
+      "loss": 0.4689,
+      "step": 869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016533484595894092,
+      "loss": 0.2603,
+      "step": 870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016525155951328872,
+      "loss": 0.3172,
+      "step": 871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016516819416888763,
+      "loss": 0.3197,
+      "step": 872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016508475002653862,
+      "loss": 0.261,
+      "step": 873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016500122718713796,
+      "loss": 0.2614,
+      "step": 874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016491762575167706,
+      "loss": 0.276,
+      "step": 875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016483394582124237,
+      "loss": 0.5533,
+      "step": 876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016475018749701525,
+      "loss": 0.2574,
+      "step": 877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016466635088027188,
+      "loss": 0.3455,
+      "step": 878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001645824360723831,
+      "loss": 0.3398,
+      "step": 879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001644984431748142,
+      "loss": 0.3479,
+      "step": 880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016441437228912503,
+      "loss": 0.353,
+      "step": 881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016433022351696966,
+      "loss": 0.4094,
+      "step": 882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016424599696009633,
+      "loss": 0.3477,
+      "step": 883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016416169272034735,
+      "loss": 0.3717,
+      "step": 884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016407731089965904,
+      "loss": 0.3978,
+      "step": 885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001639928516000614,
+      "loss": 0.4035,
+      "step": 886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016390831492367816,
+      "loss": 0.4,
+      "step": 887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016382370097272662,
+      "loss": 0.3569,
+      "step": 888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001637390098495175,
+      "loss": 0.3343,
+      "step": 889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001636542416564549,
+      "loss": 0.3347,
+      "step": 890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016356939649603606,
+      "loss": 0.3228,
+      "step": 891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016348447447085115,
+      "loss": 0.4968,
+      "step": 892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016339947568358356,
+      "loss": 0.2302,
+      "step": 893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016331440023700927,
+      "loss": 0.2172,
+      "step": 894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016322924823399703,
+      "loss": 0.3064,
+      "step": 895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016314401977750819,
+      "loss": 0.2793,
+      "step": 896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016305871497059648,
+      "loss": 0.439,
+      "step": 897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016297333391640802,
+      "loss": 0.1945,
+      "step": 898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016288787671818103,
+      "loss": 0.1786,
+      "step": 899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001628023434792459,
+      "loss": 0.3723,
+      "step": 900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016271673430302488,
+      "loss": 0.3178,
+      "step": 901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016263104929303208,
+      "loss": 0.436,
+      "step": 902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016254528855287332,
+      "loss": 0.3773,
+      "step": 903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016245945218624596,
+      "loss": 0.2968,
+      "step": 904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016237354029693886,
+      "loss": 0.347,
+      "step": 905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016228755298883205,
+      "loss": 0.2934,
+      "step": 906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016220149036589695,
+      "loss": 0.3579,
+      "step": 907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001621153525321959,
+      "loss": 0.3406,
+      "step": 908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016202913959188225,
+      "loss": 0.3203,
+      "step": 909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016194285164920013,
+      "loss": 0.3382,
+      "step": 910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001618564888084844,
+      "loss": 0.4345,
+      "step": 911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016177005117416044,
+      "loss": 0.3312,
+      "step": 912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016168353885074409,
+      "loss": 0.2821,
+      "step": 913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001615969519428415,
+      "loss": 0.3469,
+      "step": 914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016151029055514897,
+      "loss": 0.2199,
+      "step": 915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016142355479245292,
+      "loss": 0.3339,
+      "step": 916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016133674475962966,
+      "loss": 0.3786,
+      "step": 917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001612498605616453,
+      "loss": 0.4012,
+      "step": 918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016116290230355564,
+      "loss": 0.3426,
+      "step": 919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016107587009050602,
+      "loss": 0.361,
+      "step": 920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016098876402773123,
+      "loss": 0.3583,
+      "step": 921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016090158422055528,
+      "loss": 0.6244,
+      "step": 922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016081433077439145,
+      "loss": 0.4215,
+      "step": 923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016072700379474202,
+      "loss": 0.449,
+      "step": 924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016063960338719813,
+      "loss": 0.3901,
+      "step": 925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001605521296574398,
+      "loss": 0.3178,
+      "step": 926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001604645827112356,
+      "loss": 0.3942,
+      "step": 927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016037696265444273,
+      "loss": 0.3032,
+      "step": 928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016028926959300673,
+      "loss": 0.2975,
+      "step": 929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016020150363296143,
+      "loss": 0.4525,
+      "step": 930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001601136648804288,
+      "loss": 0.3292,
+      "step": 931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016002575344161883,
+      "loss": 0.4181,
+      "step": 932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015993776942282946,
+      "loss": 0.3562,
+      "step": 933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015984971293044623,
+      "loss": 0.3384,
+      "step": 934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001597615840709425,
+      "loss": 0.3758,
+      "step": 935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000159673382950879,
+      "loss": 0.496,
+      "step": 936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001595851096769039,
+      "loss": 0.2833,
+      "step": 937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001594967643557526,
+      "loss": 0.3192,
+      "step": 938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015940834709424757,
+      "loss": 0.4199,
+      "step": 939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015931985799929837,
+      "loss": 0.3662,
+      "step": 940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015923129717790126,
+      "loss": 0.3041,
+      "step": 941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015914266473713939,
+      "loss": 0.2303,
+      "step": 942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015905396078418242,
+      "loss": 0.3952,
+      "step": 943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001589651854262865,
+      "loss": 0.3358,
+      "step": 944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015887633877079412,
+      "loss": 0.1999,
+      "step": 945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000158787420925134,
+      "loss": 0.326,
+      "step": 946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015869843199682086,
+      "loss": 0.2179,
+      "step": 947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015860937209345543,
+      "loss": 0.2987,
+      "step": 948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015852024132272433,
+      "loss": 0.3239,
+      "step": 949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001584310397923997,
+      "loss": 0.4334,
+      "step": 950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001583417676103394,
+      "loss": 0.4502,
+      "step": 951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015825242488448662,
+      "loss": 0.4434,
+      "step": 952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015816301172286984,
+      "loss": 0.3429,
+      "step": 953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015807352823360283,
+      "loss": 0.3877,
+      "step": 954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015798397452488426,
+      "loss": 0.2669,
+      "step": 955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001578943507049978,
+      "loss": 0.3828,
+      "step": 956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015780465688231175,
+      "loss": 0.3633,
+      "step": 957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001577148931652793,
+      "loss": 0.3958,
+      "step": 958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015762505966243788,
+      "loss": 0.3181,
+      "step": 959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001575351564824095,
+      "loss": 0.4268,
+      "step": 960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015744518373390034,
+      "loss": 0.3189,
+      "step": 961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015735514152570072,
+      "loss": 0.2125,
+      "step": 962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015726502996668492,
+      "loss": 0.3343,
+      "step": 963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015717484916581101,
+      "loss": 0.3944,
+      "step": 964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015708459923212105,
+      "loss": 0.2677,
+      "step": 965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015699428027474035,
+      "loss": 0.3244,
+      "step": 966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015690389240287787,
+      "loss": 0.4407,
+      "step": 967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015681343572582592,
+      "loss": 0.476,
+      "step": 968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015672291035295988,
+      "loss": 0.4228,
+      "step": 969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001566323163937383,
+      "loss": 0.2798,
+      "step": 970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001565416539577026,
+      "loss": 0.245,
+      "step": 971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015645092315447708,
+      "loss": 0.3049,
+      "step": 972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015636012409376857,
+      "loss": 0.308,
+      "step": 973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015626925688536658,
+      "loss": 0.3938,
+      "step": 974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001561783216391429,
+      "loss": 0.2798,
+      "step": 975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015608731846505168,
+      "loss": 0.2698,
+      "step": 976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015599624747312914,
+      "loss": 0.3301,
+      "step": 977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015590510877349351,
+      "loss": 0.2341,
+      "step": 978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015581390247634494,
+      "loss": 0.2256,
+      "step": 979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015572262869196528,
+      "loss": 0.2214,
+      "step": 980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015563128753071797,
+      "loss": 0.3589,
+      "step": 981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015553987910304793,
+      "loss": 0.5487,
+      "step": 982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015544840351948143,
+      "loss": 0.4218,
+      "step": 983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015535686089062587,
+      "loss": 0.4048,
+      "step": 984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015526525132716988,
+      "loss": 0.2943,
+      "step": 985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015517357493988283,
+      "loss": 0.3018,
+      "step": 986
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015508183183961499,
+      "loss": 0.3672,
+      "step": 987
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015499002213729734,
+      "loss": 0.3051,
+      "step": 988
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015489814594394128,
+      "loss": 0.2852,
+      "step": 989
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001548062033706387,
+      "loss": 0.3243,
+      "step": 990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015471419452856166,
+      "loss": 0.258,
+      "step": 991
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015462211952896248,
+      "loss": 0.2953,
+      "step": 992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015452997848317338,
+      "loss": 0.3376,
+      "step": 993
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001544377715026064,
+      "loss": 0.3166,
+      "step": 994
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015434549869875345,
+      "loss": 0.2413,
+      "step": 995
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001542531601831859,
+      "loss": 0.2238,
+      "step": 996
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001541607560675546,
+      "loss": 0.2896,
+      "step": 997
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001540682864635898,
+      "loss": 0.377,
+      "step": 998
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001539757514831008,
+      "loss": 0.3031,
+      "step": 999
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015388315123797606,
+      "loss": 0.3646,
+      "step": 1000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015379048584018297,
+      "loss": 0.3462,
+      "step": 1001
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015369775540176756,
+      "loss": 0.2546,
+      "step": 1002
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015360496003485462,
+      "loss": 0.2315,
+      "step": 1003
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015351209985164745,
+      "loss": 0.1718,
+      "step": 1004
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015341917496442767,
+      "loss": 0.3028,
+      "step": 1005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001533261854855552,
+      "loss": 0.3941,
+      "step": 1006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015323313152746794,
+      "loss": 0.235,
+      "step": 1007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001531400132026819,
+      "loss": 0.3164,
+      "step": 1008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001530468306237908,
+      "loss": 0.2467,
+      "step": 1009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015295358390346618,
+      "loss": 0.4383,
+      "step": 1010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000152860273154457,
+      "loss": 0.3811,
+      "step": 1011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001527668984895897,
+      "loss": 0.2852,
+      "step": 1012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015267346002176808,
+      "loss": 0.2771,
+      "step": 1013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015257995786397297,
+      "loss": 0.2941,
+      "step": 1014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001524863921292622,
+      "loss": 0.2501,
+      "step": 1015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015239276293077062,
+      "loss": 0.4293,
+      "step": 1016
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015229907038170963,
+      "loss": 0.3188,
+      "step": 1017
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015220531459536743,
+      "loss": 0.367,
+      "step": 1018
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015211149568510849,
+      "loss": 0.3596,
+      "step": 1019
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015201761376437375,
+      "loss": 0.274,
+      "step": 1020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015192366894668026,
+      "loss": 0.2304,
+      "step": 1021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015182966134562114,
+      "loss": 0.3986,
+      "step": 1022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015173559107486547,
+      "loss": 0.1625,
+      "step": 1023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000151641458248158,
+      "loss": 0.2093,
+      "step": 1024
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015154726297931926,
+      "loss": 0.3936,
+      "step": 1025
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015145300538224512,
+      "loss": 0.3256,
+      "step": 1026
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000151358685570907,
+      "loss": 0.2723,
+      "step": 1027
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015126430365935137,
+      "loss": 0.2058,
+      "step": 1028
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001511698597616999,
+      "loss": 0.1646,
+      "step": 1029
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015107535399214918,
+      "loss": 0.2776,
+      "step": 1030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015098078646497064,
+      "loss": 0.1912,
+      "step": 1031
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015088615729451027,
+      "loss": 0.2544,
+      "step": 1032
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015079146659518873,
+      "loss": 0.2779,
+      "step": 1033
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015069671448150103,
+      "loss": 0.342,
+      "step": 1034
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001506019010680164,
+      "loss": 0.3176,
+      "step": 1035
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015050702646937829,
+      "loss": 0.304,
+      "step": 1036
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015041209080030398,
+      "loss": 0.3159,
+      "step": 1037
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015031709417558475,
+      "loss": 0.3771,
+      "step": 1038
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001502220367100855,
+      "loss": 0.409,
+      "step": 1039
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001501269185187447,
+      "loss": 0.3335,
+      "step": 1040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015003173971657425,
+      "loss": 0.1905,
+      "step": 1041
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014993650041865937,
+      "loss": 0.2972,
+      "step": 1042
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014984120074015836,
+      "loss": 0.272,
+      "step": 1043
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014974584079630265,
+      "loss": 0.2637,
+      "step": 1044
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001496504207023964,
+      "loss": 0.2386,
+      "step": 1045
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014955494057381654,
+      "loss": 0.2746,
+      "step": 1046
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014945940052601267,
+      "loss": 0.2663,
+      "step": 1047
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014936380067450674,
+      "loss": 0.3119,
+      "step": 1048
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014926814113489311,
+      "loss": 0.3166,
+      "step": 1049
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014917242202283819,
+      "loss": 0.2629,
+      "step": 1050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001490766434540805,
+      "loss": 0.2861,
+      "step": 1051
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001489808055444305,
+      "loss": 0.2702,
+      "step": 1052
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014888490840977032,
+      "loss": 0.2919,
+      "step": 1053
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014878895216605367,
+      "loss": 0.2776,
+      "step": 1054
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001486929369293058,
+      "loss": 0.2922,
+      "step": 1055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014859686281562336,
+      "loss": 0.3448,
+      "step": 1056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014850072994117402,
+      "loss": 0.2802,
+      "step": 1057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014840453842219662,
+      "loss": 0.2323,
+      "step": 1058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014830828837500087,
+      "loss": 0.3107,
+      "step": 1059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014821197991596725,
+      "loss": 0.3564,
+      "step": 1060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001481156131615469,
+      "loss": 0.3183,
+      "step": 1061
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014801918822826142,
+      "loss": 0.2107,
+      "step": 1062
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014792270523270273,
+      "loss": 0.3255,
+      "step": 1063
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014782616429153303,
+      "loss": 0.27,
+      "step": 1064
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001477295655214845,
+      "loss": 0.3509,
+      "step": 1065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001476329090393593,
+      "loss": 0.3234,
+      "step": 1066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014753619496202938,
+      "loss": 0.246,
+      "step": 1067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001474394234064363,
+      "loss": 0.3111,
+      "step": 1068
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014734259448959109,
+      "loss": 0.3908,
+      "step": 1069
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014724570832857424,
+      "loss": 0.3454,
+      "step": 1070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001471487650405353,
+      "loss": 0.2298,
+      "step": 1071
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014705176474269312,
+      "loss": 0.2831,
+      "step": 1072
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001469547075523352,
+      "loss": 0.2708,
+      "step": 1073
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014685759358681812,
+      "loss": 0.2564,
+      "step": 1074
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014676042296356688,
+      "loss": 0.2616,
+      "step": 1075
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001466631958000751,
+      "loss": 0.2121,
+      "step": 1076
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014656591221390479,
+      "loss": 0.2806,
+      "step": 1077
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014646857232268606,
+      "loss": 0.2145,
+      "step": 1078
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014637117624411727,
+      "loss": 0.2632,
+      "step": 1079
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014627372409596454,
+      "loss": 0.2172,
+      "step": 1080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014617621599606192,
+      "loss": 0.3451,
+      "step": 1081
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014607865206231108,
+      "loss": 0.3716,
+      "step": 1082
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014598103241268118,
+      "loss": 0.3626,
+      "step": 1083
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014588335716520873,
+      "loss": 0.3376,
+      "step": 1084
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014578562643799753,
+      "loss": 0.3198,
+      "step": 1085
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014568784034921844,
+      "loss": 0.2669,
+      "step": 1086
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001455899990171092,
+      "loss": 0.3007,
+      "step": 1087
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001454921025599744,
+      "loss": 0.2322,
+      "step": 1088
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001453941510961853,
+      "loss": 0.3664,
+      "step": 1089
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014529614474417967,
+      "loss": 0.1788,
+      "step": 1090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001451980836224616,
+      "loss": 0.404,
+      "step": 1091
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014509996784960145,
+      "loss": 0.289,
+      "step": 1092
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001450017975442356,
+      "loss": 0.2834,
+      "step": 1093
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001449035728250665,
+      "loss": 0.2349,
+      "step": 1094
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014480529381086224,
+      "loss": 0.3564,
+      "step": 1095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014470696062045664,
+      "loss": 0.354,
+      "step": 1096
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000144608573372749,
+      "loss": 0.3037,
+      "step": 1097
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014451013218670403,
+      "loss": 0.3131,
+      "step": 1098
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001444116371813516,
+      "loss": 0.2596,
+      "step": 1099
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001443130884757867,
+      "loss": 0.2615,
+      "step": 1100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014421448618916923,
+      "loss": 0.3319,
+      "step": 1101
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001441158304407239,
+      "loss": 0.3047,
+      "step": 1102
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014401712134974,
+      "loss": 0.3492,
+      "step": 1103
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014391835903557141,
+      "loss": 0.3736,
+      "step": 1104
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014381954361763628,
+      "loss": 0.2177,
+      "step": 1105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000143720675215417,
+      "loss": 0.2144,
+      "step": 1106
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014362175394846011,
+      "loss": 0.4535,
+      "step": 1107
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014352277993637594,
+      "loss": 0.303,
+      "step": 1108
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014342375329883864,
+      "loss": 0.2972,
+      "step": 1109
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014332467415558602,
+      "loss": 0.2932,
+      "step": 1110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014322554262641935,
+      "loss": 0.381,
+      "step": 1111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001431263588312033,
+      "loss": 0.3265,
+      "step": 1112
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014302712288986565,
+      "loss": 0.2053,
+      "step": 1113
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001429278349223973,
+      "loss": 0.2208,
+      "step": 1114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014282849504885196,
+      "loss": 0.2854,
+      "step": 1115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014272910338934626,
+      "loss": 0.3107,
+      "step": 1116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014262966006405937,
+      "loss": 0.3146,
+      "step": 1117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001425301651932328,
+      "loss": 0.3713,
+      "step": 1118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014243061889717063,
+      "loss": 0.4032,
+      "step": 1119
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014233102129623893,
+      "loss": 0.3204,
+      "step": 1120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001422313725108659,
+      "loss": 0.3946,
+      "step": 1121
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001421316726615416,
+      "loss": 0.3372,
+      "step": 1122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001420319218688178,
+      "loss": 0.2649,
+      "step": 1123
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014193212025330799,
+      "loss": 0.2179,
+      "step": 1124
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014183226793568695,
+      "loss": 0.1419,
+      "step": 1125
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001417323650366909,
+      "loss": 0.2216,
+      "step": 1126
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014163241167711708,
+      "loss": 0.2339,
+      "step": 1127
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014153240797782393,
+      "loss": 0.2459,
+      "step": 1128
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014143235405973057,
+      "loss": 0.2358,
+      "step": 1129
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000141332250043817,
+      "loss": 0.1866,
+      "step": 1130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014123209605112366,
+      "loss": 0.281,
+      "step": 1131
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014113189220275153,
+      "loss": 0.2354,
+      "step": 1132
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014103163861986178,
+      "loss": 0.2749,
+      "step": 1133
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001409313354236758,
+      "loss": 0.2604,
+      "step": 1134
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014083098273547494,
+      "loss": 0.3619,
+      "step": 1135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014073058067660033,
+      "loss": 0.2758,
+      "step": 1136
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014063012936845287,
+      "loss": 0.3353,
+      "step": 1137
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014052962893249305,
+      "loss": 0.3259,
+      "step": 1138
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014042907949024064,
+      "loss": 0.2693,
+      "step": 1139
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014032848116327475,
+      "loss": 0.2505,
+      "step": 1140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001402278340732336,
+      "loss": 0.2127,
+      "step": 1141
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014012713834181428,
+      "loss": 0.3312,
+      "step": 1142
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014002639409077286,
+      "loss": 0.2834,
+      "step": 1143
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013992560144192393,
+      "loss": 0.2541,
+      "step": 1144
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013982476051714065,
+      "loss": 0.3344,
+      "step": 1145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013972387143835452,
+      "loss": 0.3414,
+      "step": 1146
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001396229343275554,
+      "loss": 0.2457,
+      "step": 1147
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013952194930679106,
+      "loss": 0.2954,
+      "step": 1148
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013942091649816725,
+      "loss": 0.3632,
+      "step": 1149
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013931983602384756,
+      "loss": 0.4058,
+      "step": 1150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013921870800605318,
+      "loss": 0.2837,
+      "step": 1151
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013911753256706276,
+      "loss": 0.4771,
+      "step": 1152
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013901630982921233,
+      "loss": 0.2516,
+      "step": 1153
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013891503991489502,
+      "loss": 0.3101,
+      "step": 1154
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013881372294656118,
+      "loss": 0.3063,
+      "step": 1155
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013871235904671796,
+      "loss": 0.2504,
+      "step": 1156
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013861094833792913,
+      "loss": 0.2768,
+      "step": 1157
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013850949094281528,
+      "loss": 0.3006,
+      "step": 1158
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013840798698405327,
+      "loss": 0.4186,
+      "step": 1159
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013830643658437647,
+      "loss": 0.332,
+      "step": 1160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013820483986657413,
+      "loss": 0.2622,
+      "step": 1161
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001381031969534917,
+      "loss": 0.3624,
+      "step": 1162
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013800150796803047,
+      "loss": 0.3305,
+      "step": 1163
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001378997730331474,
+      "loss": 0.2542,
+      "step": 1164
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013779799227185498,
+      "loss": 0.4199,
+      "step": 1165
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013769616580722112,
+      "loss": 0.2073,
+      "step": 1166
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001375942937623691,
+      "loss": 0.2626,
+      "step": 1167
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013749237626047722,
+      "loss": 0.2913,
+      "step": 1168
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013739041342477866,
+      "loss": 0.2681,
+      "step": 1169
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013728840537856161,
+      "loss": 0.214,
+      "step": 1170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001371863522451688,
+      "loss": 0.3178,
+      "step": 1171
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013708425414799748,
+      "loss": 0.2909,
+      "step": 1172
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013698211121049934,
+      "loss": 0.2024,
+      "step": 1173
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013687992355618017,
+      "loss": 0.2274,
+      "step": 1174
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013677769130859997,
+      "loss": 0.2523,
+      "step": 1175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013667541459137252,
+      "loss": 0.2365,
+      "step": 1176
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001365730935281655,
+      "loss": 0.2047,
+      "step": 1177
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013647072824270014,
+      "loss": 0.222,
+      "step": 1178
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001363683188587511,
+      "loss": 0.2044,
+      "step": 1179
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013626586550014648,
+      "loss": 0.1688,
+      "step": 1180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013616336829076745,
+      "loss": 0.19,
+      "step": 1181
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013606082735454826,
+      "loss": 0.2595,
+      "step": 1182
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013595824281547597,
+      "loss": 0.3329,
+      "step": 1183
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001358556147975904,
+      "loss": 0.2564,
+      "step": 1184
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013575294342498402,
+      "loss": 0.364,
+      "step": 1185
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013565022882180156,
+      "loss": 0.2095,
+      "step": 1186
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013554747111224014,
+      "loss": 0.3261,
+      "step": 1187
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000135444670420549,
+      "loss": 0.3935,
+      "step": 1188
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013534182687102925,
+      "loss": 0.4297,
+      "step": 1189
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013523894058803398,
+      "loss": 0.3956,
+      "step": 1190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001351360116959678,
+      "loss": 0.3616,
+      "step": 1191
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001350330403192869,
+      "loss": 0.287,
+      "step": 1192
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013493002658249888,
+      "loss": 0.2958,
+      "step": 1193
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013482697061016256,
+      "loss": 0.2872,
+      "step": 1194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001347238725268877,
+      "loss": 0.3383,
+      "step": 1195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013462073245733516,
+      "loss": 0.2598,
+      "step": 1196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001345175505262164,
+      "loss": 0.2826,
+      "step": 1197
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013441432685829368,
+      "loss": 0.2331,
+      "step": 1198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001343110615783795,
+      "loss": 0.2265,
+      "step": 1199
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013420775481133687,
+      "loss": 0.3167,
+      "step": 1200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013410440668207887,
+      "loss": 0.3183,
+      "step": 1201
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013400101731556862,
+      "loss": 0.3493,
+      "step": 1202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001338975868368191,
+      "loss": 0.2703,
+      "step": 1203
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013379411537089303,
+      "loss": 0.236,
+      "step": 1204
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001336906030429026,
+      "loss": 0.2325,
+      "step": 1205
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001335870499780095,
+      "loss": 0.3018,
+      "step": 1206
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013348345630142462,
+      "loss": 0.3618,
+      "step": 1207
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013337982213840798,
+      "loss": 0.3939,
+      "step": 1208
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013327614761426856,
+      "loss": 0.2557,
+      "step": 1209
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013317243285436415,
+      "loss": 0.2921,
+      "step": 1210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013306867798410115,
+      "loss": 0.3765,
+      "step": 1211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013296488312893447,
+      "loss": 0.2575,
+      "step": 1212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013286104841436742,
+      "loss": 0.2905,
+      "step": 1213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013275717396595143,
+      "loss": 0.2835,
+      "step": 1214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001326532599092861,
+      "loss": 0.3016,
+      "step": 1215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001325493063700187,
+      "loss": 0.2802,
+      "step": 1216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013244531347384445,
+      "loss": 0.2814,
+      "step": 1217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013234128134650608,
+      "loss": 0.2323,
+      "step": 1218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013223721011379374,
+      "loss": 0.2478,
+      "step": 1219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013213309990154495,
+      "loss": 0.2846,
+      "step": 1220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013202895083564418,
+      "loss": 0.2625,
+      "step": 1221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013192476304202307,
+      "loss": 0.2354,
+      "step": 1222
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013182053664666002,
+      "loss": 0.3749,
+      "step": 1223
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013171627177558008,
+      "loss": 0.3098,
+      "step": 1224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001316119685548548,
+      "loss": 0.2125,
+      "step": 1225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013150762711060222,
+      "loss": 0.1695,
+      "step": 1226
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001314032475689864,
+      "loss": 0.2861,
+      "step": 1227
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013129883005621776,
+      "loss": 0.2047,
+      "step": 1228
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001311943746985523,
+      "loss": 0.2007,
+      "step": 1229
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000131089881622292,
+      "loss": 0.2067,
+      "step": 1230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013098535095378434,
+      "loss": 0.2602,
+      "step": 1231
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013088078281942237,
+      "loss": 0.2615,
+      "step": 1232
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013077617734564436,
+      "loss": 0.2407,
+      "step": 1233
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001306715346589337,
+      "loss": 0.4403,
+      "step": 1234
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001305668548858188,
+      "loss": 0.3273,
+      "step": 1235
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000130462138152873,
+      "loss": 0.3372,
+      "step": 1236
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001303573845867142,
+      "loss": 0.265,
+      "step": 1237
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001302525943140049,
+      "loss": 0.2654,
+      "step": 1238
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000130147767461452,
+      "loss": 0.2799,
+      "step": 1239
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001300429041558066,
+      "loss": 0.2889,
+      "step": 1240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012993800452386388,
+      "loss": 0.2297,
+      "step": 1241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012983306869246297,
+      "loss": 0.2293,
+      "step": 1242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001297280967884867,
+      "loss": 0.1927,
+      "step": 1243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001296230889388616,
+      "loss": 0.2317,
+      "step": 1244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001295180452705577,
+      "loss": 0.2948,
+      "step": 1245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012941296591058814,
+      "loss": 0.3027,
+      "step": 1246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012930785098600942,
+      "loss": 0.327,
+      "step": 1247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012920270062392097,
+      "loss": 0.2402,
+      "step": 1248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012909751495146504,
+      "loss": 0.3102,
+      "step": 1249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012899229409582666,
+      "loss": 0.2854,
+      "step": 1250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012888703818423328,
+      "loss": 0.1684,
+      "step": 1251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012878174734395485,
+      "loss": 0.2498,
+      "step": 1252
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001286764217023035,
+      "loss": 0.2419,
+      "step": 1253
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012857106138663347,
+      "loss": 0.3143,
+      "step": 1254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012846566652434082,
+      "loss": 0.3312,
+      "step": 1255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012836023724286354,
+      "loss": 0.3387,
+      "step": 1256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012825477366968114,
+      "loss": 0.2104,
+      "step": 1257
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012814927593231466,
+      "loss": 0.3595,
+      "step": 1258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012804374415832637,
+      "loss": 0.2045,
+      "step": 1259
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001279381784753197,
+      "loss": 0.2671,
+      "step": 1260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012783257901093916,
+      "loss": 0.3571,
+      "step": 1261
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001277269458928701,
+      "loss": 0.3194,
+      "step": 1262
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001276212792488384,
+      "loss": 0.3156,
+      "step": 1263
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001275155792066107,
+      "loss": 0.29,
+      "step": 1264
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001274098458939939,
+      "loss": 0.3587,
+      "step": 1265
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012730407943883513,
+      "loss": 0.2373,
+      "step": 1266
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012719827996902168,
+      "loss": 0.2468,
+      "step": 1267
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001270924476124806,
+      "loss": 0.3437,
+      "step": 1268
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012698658249717888,
+      "loss": 0.3474,
+      "step": 1269
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000126880684751123,
+      "loss": 0.2405,
+      "step": 1270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000126774754502359,
+      "loss": 0.2571,
+      "step": 1271
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012666879187897204,
+      "loss": 0.3867,
+      "step": 1272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012656279700908664,
+      "loss": 0.2699,
+      "step": 1273
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012645677002086619,
+      "loss": 0.2065,
+      "step": 1274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012635071104251294,
+      "loss": 0.2716,
+      "step": 1275
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001262446202022678,
+      "loss": 0.1577,
+      "step": 1276
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012613849762841024,
+      "loss": 0.3179,
+      "step": 1277
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012603234344925806,
+      "loss": 0.2042,
+      "step": 1278
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001259261577931674,
+      "loss": 0.2356,
+      "step": 1279
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012581994078853225,
+      "loss": 0.1975,
+      "step": 1280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012571369256378468,
+      "loss": 0.2024,
+      "step": 1281
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001256074132473944,
+      "loss": 0.3007,
+      "step": 1282
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001255011029678688,
+      "loss": 0.3444,
+      "step": 1283
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012539476185375266,
+      "loss": 0.3943,
+      "step": 1284
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012528839003362807,
+      "loss": 0.4688,
+      "step": 1285
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001251819876361142,
+      "loss": 0.2258,
+      "step": 1286
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012507555478986722,
+      "loss": 0.2619,
+      "step": 1287
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001249690916235802,
+      "loss": 0.2901,
+      "step": 1288
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012486259826598268,
+      "loss": 0.3898,
+      "step": 1289
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012475607484584088,
+      "loss": 0.2327,
+      "step": 1290
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012464952149195732,
+      "loss": 0.3566,
+      "step": 1291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001245429383331707,
+      "loss": 0.2461,
+      "step": 1292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012443632549835574,
+      "loss": 0.2602,
+      "step": 1293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012432968311642306,
+      "loss": 0.2015,
+      "step": 1294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012422301131631902,
+      "loss": 0.308,
+      "step": 1295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012411631022702553,
+      "loss": 0.2147,
+      "step": 1296
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012400957997756,
+      "loss": 0.2947,
+      "step": 1297
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012390282069697493,
+      "loss": 0.2757,
+      "step": 1298
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001237960325143581,
+      "loss": 0.2201,
+      "step": 1299
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012368921555883203,
+      "loss": 0.3232,
+      "step": 1300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012358236995955432,
+      "loss": 0.2211,
+      "step": 1301
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012347549584571692,
+      "loss": 0.1755,
+      "step": 1302
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012336859334654643,
+      "loss": 0.2226,
+      "step": 1303
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001232616625913037,
+      "loss": 0.2656,
+      "step": 1304
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012315470370928382,
+      "loss": 0.2854,
+      "step": 1305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012304771682981576,
+      "loss": 0.2496,
+      "step": 1306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012294070208226246,
+      "loss": 0.2411,
+      "step": 1307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012283365959602052,
+      "loss": 0.1909,
+      "step": 1308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012272658950052004,
+      "loss": 0.3022,
+      "step": 1309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012261949192522462,
+      "loss": 0.3115,
+      "step": 1310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001225123669996309,
+      "loss": 0.2939,
+      "step": 1311
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001224052148532688,
+      "loss": 0.2994,
+      "step": 1312
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012229803561570097,
+      "loss": 0.3378,
+      "step": 1313
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012219082941652295,
+      "loss": 0.2577,
+      "step": 1314
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001220835963853628,
+      "loss": 0.3639,
+      "step": 1315
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012197633665188107,
+      "loss": 0.3768,
+      "step": 1316
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012186905034577058,
+      "loss": 0.2183,
+      "step": 1317
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012176173759675628,
+      "loss": 0.1987,
+      "step": 1318
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012165439853459508,
+      "loss": 0.3789,
+      "step": 1319
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012154703328907573,
+      "loss": 0.2292,
+      "step": 1320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012143964199001866,
+      "loss": 0.2557,
+      "step": 1321
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001213322247672757,
+      "loss": 0.267,
+      "step": 1322
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012122478175073015,
+      "loss": 0.2323,
+      "step": 1323
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012111731307029645,
+      "loss": 0.1887,
+      "step": 1324
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012100981885592005,
+      "loss": 0.2016,
+      "step": 1325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001209022992375773,
+      "loss": 0.2584,
+      "step": 1326
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012079475434527527,
+      "loss": 0.2378,
+      "step": 1327
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012068718430905155,
+      "loss": 0.3152,
+      "step": 1328
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012057958925897417,
+      "loss": 0.199,
+      "step": 1329
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001204719693251414,
+      "loss": 0.2105,
+      "step": 1330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012036432463768163,
+      "loss": 0.1801,
+      "step": 1331
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001202566553267531,
+      "loss": 0.2073,
+      "step": 1332
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012014896152254389,
+      "loss": 0.2941,
+      "step": 1333
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012004124335527166,
+      "loss": 0.3877,
+      "step": 1334
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011993350095518359,
+      "loss": 0.3008,
+      "step": 1335
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011982573445255605,
+      "loss": 0.3013,
+      "step": 1336
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011971794397769464,
+      "loss": 0.299,
+      "step": 1337
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011961012966093397,
+      "loss": 0.2857,
+      "step": 1338
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001195022916326374,
+      "loss": 0.4252,
+      "step": 1339
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011939443002319699,
+      "loss": 0.2337,
+      "step": 1340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011928654496303329,
+      "loss": 0.272,
+      "step": 1341
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011917863658259527,
+      "loss": 0.2336,
+      "step": 1342
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011907070501236007,
+      "loss": 0.3068,
+      "step": 1343
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011896275038283282,
+      "loss": 0.4444,
+      "step": 1344
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011885477282454662,
+      "loss": 0.1946,
+      "step": 1345
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001187467724680622,
+      "loss": 0.2776,
+      "step": 1346
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011863874944396793,
+      "loss": 0.2549,
+      "step": 1347
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011853070388287955,
+      "loss": 0.3697,
+      "step": 1348
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011842263591544007,
+      "loss": 0.3064,
+      "step": 1349
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011831454567231959,
+      "loss": 0.2885,
+      "step": 1350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001182064332842151,
+      "loss": 0.2207,
+      "step": 1351
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011809829888185044,
+      "loss": 0.2928,
+      "step": 1352
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011799014259597606,
+      "loss": 0.426,
+      "step": 1353
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011788196455736877,
+      "loss": 0.2995,
+      "step": 1354
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011777376489683183,
+      "loss": 0.3822,
+      "step": 1355
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011766554374519452,
+      "loss": 0.2699,
+      "step": 1356
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011755730123331217,
+      "loss": 0.3018,
+      "step": 1357
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011744903749206595,
+      "loss": 0.3885,
+      "step": 1358
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011734075265236262,
+      "loss": 0.2829,
+      "step": 1359
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011723244684513455,
+      "loss": 0.4065,
+      "step": 1360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001171241202013394,
+      "loss": 0.1958,
+      "step": 1361
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011701577285196009,
+      "loss": 0.3184,
+      "step": 1362
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011690740492800445,
+      "loss": 0.4248,
+      "step": 1363
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011679901656050529,
+      "loss": 0.281,
+      "step": 1364
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011669060788052013,
+      "loss": 0.2821,
+      "step": 1365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011658217901913106,
+      "loss": 0.2464,
+      "step": 1366
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011647373010744451,
+      "loss": 0.3472,
+      "step": 1367
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011636526127659117,
+      "loss": 0.2648,
+      "step": 1368
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011625677265772589,
+      "loss": 0.1573,
+      "step": 1369
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011614826438202739,
+      "loss": 0.2207,
+      "step": 1370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011603973658069813,
+      "loss": 0.1932,
+      "step": 1371
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011593118938496421,
+      "loss": 0.3681,
+      "step": 1372
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011582262292607521,
+      "loss": 0.3885,
+      "step": 1373
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011571403733530395,
+      "loss": 0.1771,
+      "step": 1374
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011560543274394645,
+      "loss": 0.2155,
+      "step": 1375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011549680928332161,
+      "loss": 0.2225,
+      "step": 1376
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001153881670847712,
+      "loss": 0.1589,
+      "step": 1377
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011527950627965967,
+      "loss": 0.2087,
+      "step": 1378
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011517082699937395,
+      "loss": 0.1476,
+      "step": 1379
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011506212937532328,
+      "loss": 0.2183,
+      "step": 1380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011495341353893909,
+      "loss": 0.2081,
+      "step": 1381
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011484467962167487,
+      "loss": 0.2793,
+      "step": 1382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011473592775500595,
+      "loss": 0.1758,
+      "step": 1383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011462715807042932,
+      "loss": 0.2375,
+      "step": 1384
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011451837069946359,
+      "loss": 0.3086,
+      "step": 1385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001144095657736487,
+      "loss": 0.2526,
+      "step": 1386
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011430074342454587,
+      "loss": 0.1895,
+      "step": 1387
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011419190378373728,
+      "loss": 0.237,
+      "step": 1388
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011408304698282614,
+      "loss": 0.2246,
+      "step": 1389
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011397417315343639,
+      "loss": 0.3295,
+      "step": 1390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011386528242721248,
+      "loss": 0.3867,
+      "step": 1391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011375637493581937,
+      "loss": 0.3453,
+      "step": 1392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011364745081094224,
+      "loss": 0.2659,
+      "step": 1393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011353851018428639,
+      "loss": 0.2823,
+      "step": 1394
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011342955318757713,
+      "loss": 0.3307,
+      "step": 1395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011332057995255953,
+      "loss": 0.2451,
+      "step": 1396
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011321159061099822,
+      "loss": 0.3339,
+      "step": 1397
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011310258529467743,
+      "loss": 0.2835,
+      "step": 1398
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001129935641354006,
+      "loss": 0.2795,
+      "step": 1399
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011288452726499046,
+      "loss": 0.2598,
+      "step": 1400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011277547481528854,
+      "loss": 0.3706,
+      "step": 1401
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001126664069181554,
+      "loss": 0.2239,
+      "step": 1402
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011255732370547014,
+      "loss": 0.2524,
+      "step": 1403
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011244822530913049,
+      "loss": 0.2974,
+      "step": 1404
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011233911186105244,
+      "loss": 0.3812,
+      "step": 1405
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011222998349317024,
+      "loss": 0.325,
+      "step": 1406
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011212084033743617,
+      "loss": 0.3324,
+      "step": 1407
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011201168252582038,
+      "loss": 0.278,
+      "step": 1408
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011190251019031075,
+      "loss": 0.2783,
+      "step": 1409
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011179332346291271,
+      "loss": 0.3431,
+      "step": 1410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001116841224756491,
+      "loss": 0.3072,
+      "step": 1411
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011157490736056,
+      "loss": 0.2684,
+      "step": 1412
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011146567824970262,
+      "loss": 0.2193,
+      "step": 1413
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011135643527515096,
+      "loss": 0.306,
+      "step": 1414
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011124717856899593,
+      "loss": 0.3885,
+      "step": 1415
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000111137908263345,
+      "loss": 0.238,
+      "step": 1416
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011102862449032202,
+      "loss": 0.2897,
+      "step": 1417
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011091932738206723,
+      "loss": 0.3356,
+      "step": 1418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011081001707073686,
+      "loss": 0.3723,
+      "step": 1419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011070069368850324,
+      "loss": 0.2069,
+      "step": 1420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011059135736755441,
+      "loss": 0.2724,
+      "step": 1421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011048200824009413,
+      "loss": 0.1794,
+      "step": 1422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011037264643834157,
+      "loss": 0.2465,
+      "step": 1423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011026327209453128,
+      "loss": 0.305,
+      "step": 1424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011015388534091295,
+      "loss": 0.2412,
+      "step": 1425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001100444863097513,
+      "loss": 0.2001,
+      "step": 1426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010993507513332584,
+      "loss": 0.2144,
+      "step": 1427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010982565194393082,
+      "loss": 0.2138,
+      "step": 1428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010971621687387498,
+      "loss": 0.1872,
+      "step": 1429
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001096067700554815,
+      "loss": 0.2144,
+      "step": 1430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010949731162108763,
+      "loss": 0.1674,
+      "step": 1431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010938784170304479,
+      "loss": 0.3296,
+      "step": 1432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010927836043371822,
+      "loss": 0.2473,
+      "step": 1433
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010916886794548691,
+      "loss": 0.2724,
+      "step": 1434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010905936437074344,
+      "loss": 0.2449,
+      "step": 1435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010894984984189373,
+      "loss": 0.2252,
+      "step": 1436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010884032449135697,
+      "loss": 0.1889,
+      "step": 1437
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010873078845156548,
+      "loss": 0.2063,
+      "step": 1438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010862124185496447,
+      "loss": 0.2773,
+      "step": 1439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010851168483401188,
+      "loss": 0.329,
+      "step": 1440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001084021175211783,
+      "loss": 0.297,
+      "step": 1441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010829254004894675,
+      "loss": 0.3092,
+      "step": 1442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010818295254981259,
+      "loss": 0.4018,
+      "step": 1443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010807335515628319,
+      "loss": 0.2313,
+      "step": 1444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010796374800087796,
+      "loss": 0.2233,
+      "step": 1445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001078541312161281,
+      "loss": 0.2417,
+      "step": 1446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010774450493457648,
+      "loss": 0.344,
+      "step": 1447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001076348692887774,
+      "loss": 0.3929,
+      "step": 1448
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010752522441129648,
+      "loss": 0.3923,
+      "step": 1449
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001074155704347106,
+      "loss": 0.285,
+      "step": 1450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010730590749160751,
+      "loss": 0.2192,
+      "step": 1451
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010719623571458592,
+      "loss": 0.2883,
+      "step": 1452
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010708655523625512,
+      "loss": 0.3108,
+      "step": 1453
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010697686618923499,
+      "loss": 0.1752,
+      "step": 1454
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010686716870615571,
+      "loss": 0.2499,
+      "step": 1455
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010675746291965777,
+      "loss": 0.2161,
+      "step": 1456
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001066477489623916,
+      "loss": 0.3386,
+      "step": 1457
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010653802696701749,
+      "loss": 0.2979,
+      "step": 1458
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010642829706620556,
+      "loss": 0.2888,
+      "step": 1459
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001063185593926354,
+      "loss": 0.3359,
+      "step": 1460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010620881407899601,
+      "loss": 0.2766,
+      "step": 1461
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010609906125798568,
+      "loss": 0.2701,
+      "step": 1462
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010598930106231171,
+      "loss": 0.2771,
+      "step": 1463
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010587953362469034,
+      "loss": 0.3489,
+      "step": 1464
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010576975907784664,
+      "loss": 0.2076,
+      "step": 1465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010565997755451412,
+      "loss": 0.4236,
+      "step": 1466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010555018918743487,
+      "loss": 0.3316,
+      "step": 1467
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010544039410935918,
+      "loss": 0.2117,
+      "step": 1468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010533059245304551,
+      "loss": 0.2797,
+      "step": 1469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001052207843512602,
+      "loss": 0.2239,
+      "step": 1470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010511096993677742,
+      "loss": 0.2286,
+      "step": 1471
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010500114934237899,
+      "loss": 0.2675,
+      "step": 1472
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010489132270085422,
+      "loss": 0.1947,
+      "step": 1473
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010478149014499967,
+      "loss": 0.1554,
+      "step": 1474
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010467165180761907,
+      "loss": 0.2383,
+      "step": 1475
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010456180782152314,
+      "loss": 0.2755,
+      "step": 1476
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010445195831952952,
+      "loss": 0.1411,
+      "step": 1477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010434210343446236,
+      "loss": 0.2566,
+      "step": 1478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010423224329915242,
+      "loss": 0.1645,
+      "step": 1479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010412237804643678,
+      "loss": 0.1714,
+      "step": 1480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010401250780915877,
+      "loss": 0.2342,
+      "step": 1481
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010390263272016766,
+      "loss": 0.3312,
+      "step": 1482
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010379275291231863,
+      "loss": 0.3537,
+      "step": 1483
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010368286851847253,
+      "loss": 0.3664,
+      "step": 1484
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001035729796714958,
+      "loss": 0.2438,
+      "step": 1485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010346308650426024,
+      "loss": 0.3057,
+      "step": 1486
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010335318914964289,
+      "loss": 0.3175,
+      "step": 1487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010324328774052582,
+      "loss": 0.4564,
+      "step": 1488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010313338240979604,
+      "loss": 0.2201,
+      "step": 1489
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010302347329034532,
+      "loss": 0.2145,
+      "step": 1490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010291356051506993,
+      "loss": 0.3514,
+      "step": 1491
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010280364421687062,
+      "loss": 0.2899,
+      "step": 1492
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001026937245286524,
+      "loss": 0.1967,
+      "step": 1493
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010258380158332437,
+      "loss": 0.2484,
+      "step": 1494
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010247387551379954,
+      "loss": 0.3051,
+      "step": 1495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010236394645299473,
+      "loss": 0.2189,
+      "step": 1496
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010225401453383038,
+      "loss": 0.3204,
+      "step": 1497
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010214407988923041,
+      "loss": 0.2572,
+      "step": 1498
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010203414265212197,
+      "loss": 0.1508,
+      "step": 1499
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010192420295543534,
+      "loss": 0.3096,
+      "step": 1500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010181426093210387,
+      "loss": 0.2454,
+      "step": 1501
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010170431671506361,
+      "loss": 0.2624,
+      "step": 1502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001015943704372534,
+      "loss": 0.2646,
+      "step": 1503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010148442223161438,
+      "loss": 0.2577,
+      "step": 1504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001013744722310902,
+      "loss": 0.2262,
+      "step": 1505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010126452056862661,
+      "loss": 0.2563,
+      "step": 1506
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010115456737717136,
+      "loss": 0.1909,
+      "step": 1507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010104461278967401,
+      "loss": 0.1264,
+      "step": 1508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010093465693908593,
+      "loss": 0.273,
+      "step": 1509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010082469995835988,
+      "loss": 0.26,
+      "step": 1510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010071474198045007,
+      "loss": 0.2724,
+      "step": 1511
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010060478313831196,
+      "loss": 0.4683,
+      "step": 1512
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001004948235649019,
+      "loss": 0.3056,
+      "step": 1513
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010038486339317723,
+      "loss": 0.2585,
+      "step": 1514
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010027490275609603,
+      "loss": 0.3112,
+      "step": 1515
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010016494178661693,
+      "loss": 0.2795,
+      "step": 1516
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010005498061769887,
+      "loss": 0.2092,
+      "step": 1517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994501938230115e-05,
+      "loss": 0.2538,
+      "step": 1518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98350582133831e-05,
+      "loss": 0.2668,
+      "step": 1519
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972509724390398e-05,
+      "loss": 0.2774,
+      "step": 1520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961513660682278e-05,
+      "loss": 0.2449,
+      "step": 1521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950517643509813e-05,
+      "loss": 0.2128,
+      "step": 1522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939521686168805e-05,
+      "loss": 0.2593,
+      "step": 1523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928525801954994e-05,
+      "loss": 0.2478,
+      "step": 1524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917530004164015e-05,
+      "loss": 0.2382,
+      "step": 1525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906534306091411e-05,
+      "loss": 0.2918,
+      "step": 1526
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.8955387210326e-05,
+      "loss": 0.2802,
+      "step": 1527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884543262282867e-05,
+      "loss": 0.2032,
+      "step": 1528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873547943137341e-05,
+      "loss": 0.2266,
+      "step": 1529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86255277689098e-05,
+      "loss": 0.1545,
+      "step": 1530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851557776838564e-05,
+      "loss": 0.1712,
+      "step": 1531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.840562956274662e-05,
+      "loss": 0.3316,
+      "step": 1532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82956832849364e-05,
+      "loss": 0.3091,
+      "step": 1533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818573906789617e-05,
+      "loss": 0.2476,
+      "step": 1534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807579704456468e-05,
+      "loss": 0.3211,
+      "step": 1535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796585734787806e-05,
+      "loss": 0.2825,
+      "step": 1536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.785592011076962e-05,
+      "loss": 0.2381,
+      "step": 1537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774598546616963e-05,
+      "loss": 0.245,
+      "step": 1538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763605354700529e-05,
+      "loss": 0.2866,
+      "step": 1539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752612448620048e-05,
+      "loss": 0.294,
+      "step": 1540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.741619841667566e-05,
+      "loss": 0.3472,
+      "step": 1541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.730627547134763e-05,
+      "loss": 0.3466,
+      "step": 1542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71963557831294e-05,
+      "loss": 0.2985,
+      "step": 1543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70864394849301e-05,
+      "loss": 0.2574,
+      "step": 1544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697652670965469e-05,
+      "loss": 0.3082,
+      "step": 1545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686661759020397e-05,
+      "loss": 0.2118,
+      "step": 1546
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67567122594742e-05,
+      "loss": 0.3278,
+      "step": 1547
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664681085035714e-05,
+      "loss": 0.3661,
+      "step": 1548
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653691349573978e-05,
+      "loss": 0.1966,
+      "step": 1549
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642702032850424e-05,
+      "loss": 0.3741,
+      "step": 1550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63171314815275e-05,
+      "loss": 0.237,
+      "step": 1551
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620724708768139e-05,
+      "loss": 0.284,
+      "step": 1552
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.609736727983235e-05,
+      "loss": 0.2808,
+      "step": 1553
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.598749219084124e-05,
+      "loss": 0.3342,
+      "step": 1554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587762195356323e-05,
+      "loss": 0.3788,
+      "step": 1555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57677567008476e-05,
+      "loss": 0.2422,
+      "step": 1556
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.565789656553767e-05,
+      "loss": 0.2278,
+      "step": 1557
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554804168047052e-05,
+      "loss": 0.2701,
+      "step": 1558
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.543819217847688e-05,
+      "loss": 0.1653,
+      "step": 1559
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532834819238097e-05,
+      "loss": 0.3099,
+      "step": 1560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521850985500036e-05,
+      "loss": 0.3174,
+      "step": 1561
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510867729914579e-05,
+      "loss": 0.2088,
+      "step": 1562
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499885065762102e-05,
+      "loss": 0.2654,
+      "step": 1563
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48890300632226e-05,
+      "loss": 0.3778,
+      "step": 1564
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477921564873983e-05,
+      "loss": 0.3043,
+      "step": 1565
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.466940754695452e-05,
+      "loss": 0.1696,
+      "step": 1566
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.455960589064084e-05,
+      "loss": 0.2608,
+      "step": 1567
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444981081256515e-05,
+      "loss": 0.1909,
+      "step": 1568
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43400224454859e-05,
+      "loss": 0.2396,
+      "step": 1569
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423024092215339e-05,
+      "loss": 0.2645,
+      "step": 1570
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412046637530968e-05,
+      "loss": 0.2308,
+      "step": 1571
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401069893768833e-05,
+      "loss": 0.2812,
+      "step": 1572
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390093874201434e-05,
+      "loss": 0.2725,
+      "step": 1573
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3791185921004e-05,
+      "loss": 0.2097,
+      "step": 1574
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368144060736463e-05,
+      "loss": 0.1821,
+      "step": 1575
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357170293379447e-05,
+      "loss": 0.2349,
+      "step": 1576
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346197303298253e-05,
+      "loss": 0.1466,
+      "step": 1577
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335225103760844e-05,
+      "loss": 0.3125,
+      "step": 1578
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324253708034224e-05,
+      "loss": 0.131,
+      "step": 1579
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31328312938443e-05,
+      "loss": 0.376,
+      "step": 1580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302313381076503e-05,
+      "loss": 0.2482,
+      "step": 1581
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29134447637449e-05,
+      "loss": 0.3789,
+      "step": 1582
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28037642854141e-05,
+      "loss": 0.2495,
+      "step": 1583
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269409250839251e-05,
+      "loss": 0.3608,
+      "step": 1584
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258442956528942e-05,
+      "loss": 0.2969,
+      "step": 1585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.247477558870353e-05,
+      "loss": 0.2482,
+      "step": 1586
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236513071122262e-05,
+      "loss": 0.2368,
+      "step": 1587
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225549506542354e-05,
+      "loss": 0.2557,
+      "step": 1588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214586878387192e-05,
+      "loss": 0.2941,
+      "step": 1589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203625199912207e-05,
+      "loss": 0.2013,
+      "step": 1590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192664484371684e-05,
+      "loss": 0.2075,
+      "step": 1591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181704745018742e-05,
+      "loss": 0.2902,
+      "step": 1592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170745995105326e-05,
+      "loss": 0.3112,
+      "step": 1593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159788247882173e-05,
+      "loss": 0.4513,
+      "step": 1594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148831516598814e-05,
+      "loss": 0.2187,
+      "step": 1595
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137875814503555e-05,
+      "loss": 0.2631,
+      "step": 1596
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126921154843454e-05,
+      "loss": 0.2702,
+      "step": 1597
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115967550864304e-05,
+      "loss": 0.3195,
+      "step": 1598
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105015015810629e-05,
+      "loss": 0.3912,
+      "step": 1599
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094063562925658e-05,
+      "loss": 0.1876,
+      "step": 1600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08311320545131e-05,
+      "loss": 0.2063,
+      "step": 1601
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07216395662818e-05,
+      "loss": 0.2859,
+      "step": 1602
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061215829695523e-05,
+      "loss": 0.2867,
+      "step": 1603
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05026883789124e-05,
+      "loss": 0.1736,
+      "step": 1604
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039322994451853e-05,
+      "loss": 0.2758,
+      "step": 1605
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028378312612504e-05,
+      "loss": 0.2616,
+      "step": 1606
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017434805606922e-05,
+      "loss": 0.2517,
+      "step": 1607
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.006492486667418e-05,
+      "loss": 0.4025,
+      "step": 1608
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995551369024872e-05,
+      "loss": 0.419,
+      "step": 1609
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984611465908707e-05,
+      "loss": 0.185,
+      "step": 1610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.973672790546873e-05,
+      "loss": 0.2636,
+      "step": 1611
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962735356165844e-05,
+      "loss": 0.2968,
+      "step": 1612
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951799175990588e-05,
+      "loss": 0.1987,
+      "step": 1613
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.940864263244561e-05,
+      "loss": 0.3423,
+      "step": 1614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929930631149678e-05,
+      "loss": 0.1749,
+      "step": 1615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918998292926316e-05,
+      "loss": 0.3332,
+      "step": 1616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.908067261793281e-05,
+      "loss": 0.2582,
+      "step": 1617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897137550967798e-05,
+      "loss": 0.2065,
+      "step": 1618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886209173665503e-05,
+      "loss": 0.3832,
+      "step": 1619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875282143100407e-05,
+      "loss": 0.2018,
+      "step": 1620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.864356472484905e-05,
+      "loss": 0.1364,
+      "step": 1621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853432175029742e-05,
+      "loss": 0.1769,
+      "step": 1622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842509263944002e-05,
+      "loss": 0.1813,
+      "step": 1623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831587752435094e-05,
+      "loss": 0.1382,
+      "step": 1624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820667653708731e-05,
+      "loss": 0.1969,
+      "step": 1625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809748980968928e-05,
+      "loss": 0.1491,
+      "step": 1626
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798831747417966e-05,
+      "loss": 0.194,
+      "step": 1627
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787915966256385e-05,
+      "loss": 0.1616,
+      "step": 1628
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777001650682978e-05,
+      "loss": 0.1471,
+      "step": 1629
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766088813894758e-05,
+      "loss": 0.1886,
+      "step": 1630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755177469086955e-05,
+      "loss": 0.1993,
+      "step": 1631
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744267629452988e-05,
+      "loss": 0.3163,
+      "step": 1632
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733359308184463e-05,
+      "loss": 0.3069,
+      "step": 1633
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722452518471146e-05,
+      "loss": 0.2736,
+      "step": 1634
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.711547273500957e-05,
+      "loss": 0.3043,
+      "step": 1635
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70064358645994e-05,
+      "loss": 0.2564,
+      "step": 1636
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68974147053226e-05,
+      "loss": 0.1933,
+      "step": 1637
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67884093890018e-05,
+      "loss": 0.3559,
+      "step": 1638
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667942004744049e-05,
+      "loss": 0.3728,
+      "step": 1639
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65704468124229e-05,
+      "loss": 0.2729,
+      "step": 1640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646148981571362e-05,
+      "loss": 0.2471,
+      "step": 1641
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63525491890578e-05,
+      "loss": 0.2317,
+      "step": 1642
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624362506418065e-05,
+      "loss": 0.2492,
+      "step": 1643
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613471757278755e-05,
+      "loss": 0.3943,
+      "step": 1644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.602582684656363e-05,
+      "loss": 0.2376,
+      "step": 1645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591695301717387e-05,
+      "loss": 0.1971,
+      "step": 1646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580809621626273e-05,
+      "loss": 0.3763,
+      "step": 1647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569925657545417e-05,
+      "loss": 0.2053,
+      "step": 1648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559043422635134e-05,
+      "loss": 0.2098,
+      "step": 1649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.548162930053644e-05,
+      "loss": 0.3262,
+      "step": 1650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53728419295707e-05,
+      "loss": 0.2215,
+      "step": 1651
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526407224499407e-05,
+      "loss": 0.2171,
+      "step": 1652
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515532037832515e-05,
+      "loss": 0.3824,
+      "step": 1653
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504658646106093e-05,
+      "loss": 0.2965,
+      "step": 1654
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.493787062467675e-05,
+      "loss": 0.2754,
+      "step": 1655
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482917300062606e-05,
+      "loss": 0.231,
+      "step": 1656
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472049372034034e-05,
+      "loss": 0.2335,
+      "step": 1657
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461183291522882e-05,
+      "loss": 0.1763,
+      "step": 1658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450319071667843e-05,
+      "loss": 0.2657,
+      "step": 1659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439456725605356e-05,
+      "loss": 0.1894,
+      "step": 1660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428596266469607e-05,
+      "loss": 0.2252,
+      "step": 1661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417737707392481e-05,
+      "loss": 0.2761,
+      "step": 1662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406881061503581e-05,
+      "loss": 0.3488,
+      "step": 1663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39602634193019e-05,
+      "loss": 0.2514,
+      "step": 1664
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.385173561797263e-05,
+      "loss": 0.4422,
+      "step": 1665
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374322734227413e-05,
+      "loss": 0.3872,
+      "step": 1666
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363473872340884e-05,
+      "loss": 0.4086,
+      "step": 1667
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352626989255552e-05,
+      "loss": 0.2954,
+      "step": 1668
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341782098086896e-05,
+      "loss": 0.3022,
+      "step": 1669
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330939211947988e-05,
+      "loss": 0.3345,
+      "step": 1670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.320098343949473e-05,
+      "loss": 0.1668,
+      "step": 1671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309259507199557e-05,
+      "loss": 0.2946,
+      "step": 1672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298422714803992e-05,
+      "loss": 0.3248,
+      "step": 1673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287587979866061e-05,
+      "loss": 0.1918,
+      "step": 1674
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276755315486546e-05,
+      "loss": 0.2295,
+      "step": 1675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.265924734763739e-05,
+      "loss": 0.1889,
+      "step": 1676
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.255096250793409e-05,
+      "loss": 0.168,
+      "step": 1677
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244269876668787e-05,
+      "loss": 0.2174,
+      "step": 1678
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.233445625480552e-05,
+      "loss": 0.1841,
+      "step": 1679
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22262351031682e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.211803544263124e-05,
+      "loss": 0.2245,
+      "step": 1681
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200985740402395e-05,
+      "loss": 0.3818,
+      "step": 1682
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190170111814958e-05,
+      "loss": 0.2978,
+      "step": 1683
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.179356671578492e-05,
+      "loss": 0.2657,
+      "step": 1684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.168545432768045e-05,
+      "loss": 0.3445,
+      "step": 1685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.157736408455994e-05,
+      "loss": 0.3144,
+      "step": 1686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146929611712048e-05,
+      "loss": 0.239,
+      "step": 1687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13612505560321e-05,
+      "loss": 0.184,
+      "step": 1688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125322753193782e-05,
+      "loss": 0.2778,
+      "step": 1689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11452271754534e-05,
+      "loss": 0.2737,
+      "step": 1690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10372496171672e-05,
+      "loss": 0.2756,
+      "step": 1691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.092929498763996e-05,
+      "loss": 0.3082,
+      "step": 1692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082136341740474e-05,
+      "loss": 0.2254,
+      "step": 1693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.071345503696672e-05,
+      "loss": 0.3594,
+      "step": 1694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060556997680304e-05,
+      "loss": 0.2773,
+      "step": 1695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.049770836736264e-05,
+      "loss": 0.3917,
+      "step": 1696
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038987033906604e-05,
+      "loss": 0.2478,
+      "step": 1697
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028205602230535e-05,
+      "loss": 0.2261,
+      "step": 1698
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017426554744396e-05,
+      "loss": 0.2247,
+      "step": 1699
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006649904481645e-05,
+      "loss": 0.2594,
+      "step": 1700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995875664472834e-05,
+      "loss": 0.2522,
+      "step": 1701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985103847745612e-05,
+      "loss": 0.2779,
+      "step": 1702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97433446732469e-05,
+      "loss": 0.1772,
+      "step": 1703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.963567536231839e-05,
+      "loss": 0.3167,
+      "step": 1704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95280306748586e-05,
+      "loss": 0.1823,
+      "step": 1705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942041074102584e-05,
+      "loss": 0.1613,
+      "step": 1706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.931281569094846e-05,
+      "loss": 0.1892,
+      "step": 1707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.920524565472476e-05,
+      "loss": 0.1552,
+      "step": 1708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90977007624227e-05,
+      "loss": 0.235,
+      "step": 1709
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899018114407995e-05,
+      "loss": 0.2506,
+      "step": 1710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888268692970354e-05,
+      "loss": 0.409,
+      "step": 1711
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.877521824926983e-05,
+      "loss": 0.2692,
+      "step": 1712
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866777523272431e-05,
+      "loss": 0.2682,
+      "step": 1713
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.856035800998135e-05,
+      "loss": 0.2656,
+      "step": 1714
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.845296671092426e-05,
+      "loss": 0.2759,
+      "step": 1715
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834560146540492e-05,
+      "loss": 0.2053,
+      "step": 1716
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823826240324374e-05,
+      "loss": 0.1695,
+      "step": 1717
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813094965422943e-05,
+      "loss": 0.2247,
+      "step": 1718
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802366334811893e-05,
+      "loss": 0.2905,
+      "step": 1719
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791640361463719e-05,
+      "loss": 0.3217,
+      "step": 1720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780917058347707e-05,
+      "loss": 0.1929,
+      "step": 1721
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770196438429904e-05,
+      "loss": 0.2488,
+      "step": 1722
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759478514673122e-05,
+      "loss": 0.1395,
+      "step": 1723
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748763300036909e-05,
+      "loss": 0.2587,
+      "step": 1724
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738050807477539e-05,
+      "loss": 0.2547,
+      "step": 1725
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.727341049947997e-05,
+      "loss": 0.1918,
+      "step": 1726
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71663404039795e-05,
+      "loss": 0.1665,
+      "step": 1727
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705929791773754e-05,
+      "loss": 0.1451,
+      "step": 1728
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695228317018425e-05,
+      "loss": 0.1765,
+      "step": 1729
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68452962907162e-05,
+      "loss": 0.1855,
+      "step": 1730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673833740869629e-05,
+      "loss": 0.2319,
+      "step": 1731
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663140665345358e-05,
+      "loss": 0.2279,
+      "step": 1732
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.652450415428309e-05,
+      "loss": 0.191,
+      "step": 1733
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.641763004044572e-05,
+      "loss": 0.2538,
+      "step": 1734
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631078444116797e-05,
+      "loss": 0.3367,
+      "step": 1735
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620396748564194e-05,
+      "loss": 0.3265,
+      "step": 1736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609717930302506e-05,
+      "loss": 0.2343,
+      "step": 1737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599042002244003e-05,
+      "loss": 0.1864,
+      "step": 1738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588368977297445e-05,
+      "loss": 0.2081,
+      "step": 1739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577698868368099e-05,
+      "loss": 0.2466,
+      "step": 1740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.567031688357695e-05,
+      "loss": 0.2181,
+      "step": 1741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.556367450164427e-05,
+      "loss": 0.2866,
+      "step": 1742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545706166682931e-05,
+      "loss": 0.2826,
+      "step": 1743
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535047850804267e-05,
+      "loss": 0.2658,
+      "step": 1744
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52439251541591e-05,
+      "loss": 0.2884,
+      "step": 1745
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51374017340173e-05,
+      "loss": 0.2882,
+      "step": 1746
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503090837641982e-05,
+      "loss": 0.2339,
+      "step": 1747
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.492444521013277e-05,
+      "loss": 0.3448,
+      "step": 1748
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481801236388581e-05,
+      "loss": 0.3239,
+      "step": 1749
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471160996637194e-05,
+      "loss": 0.3641,
+      "step": 1750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460523814624735e-05,
+      "loss": 0.2339,
+      "step": 1751
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.44988970321312e-05,
+      "loss": 0.3098,
+      "step": 1752
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.439258675260561e-05,
+      "loss": 0.3046,
+      "step": 1753
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428630743621533e-05,
+      "loss": 0.2522,
+      "step": 1754
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418005921146777e-05,
+      "loss": 0.2199,
+      "step": 1755
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.407384220683263e-05,
+      "loss": 0.1959,
+      "step": 1756
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.396765655074192e-05,
+      "loss": 0.2375,
+      "step": 1757
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386150237158977e-05,
+      "loss": 0.269,
+      "step": 1758
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.37553797977322e-05,
+      "loss": 0.2756,
+      "step": 1759
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364928895748709e-05,
+      "loss": 0.2792,
+      "step": 1760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354322997913382e-05,
+      "loss": 0.2774,
+      "step": 1761
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343720299091337e-05,
+      "loss": 0.1876,
+      "step": 1762
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.333120812102796e-05,
+      "loss": 0.2435,
+      "step": 1763
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.322524549764105e-05,
+      "loss": 0.3581,
+      "step": 1764
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3119315248877e-05,
+      "loss": 0.3903,
+      "step": 1765
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301341750282112e-05,
+      "loss": 0.2474,
+      "step": 1766
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290755238751939e-05,
+      "loss": 0.3138,
+      "step": 1767
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280172003097835e-05,
+      "loss": 0.2632,
+      "step": 1768
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269592056116486e-05,
+      "loss": 0.2306,
+      "step": 1769
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259015410600611e-05,
+      "loss": 0.2099,
+      "step": 1770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24844207933893e-05,
+      "loss": 0.2095,
+      "step": 1771
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.237872075116158e-05,
+      "loss": 0.2362,
+      "step": 1772
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227305410712994e-05,
+      "loss": 0.2881,
+      "step": 1773
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216742098906085e-05,
+      "loss": 0.1758,
+      "step": 1774
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20618215246803e-05,
+      "loss": 0.21,
+      "step": 1775
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.195625584167365e-05,
+      "loss": 0.2233,
+      "step": 1776
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185072406768535e-05,
+      "loss": 0.1374,
+      "step": 1777
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.174522633031885e-05,
+      "loss": 0.1314,
+      "step": 1778
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163976275713646e-05,
+      "loss": 0.1689,
+      "step": 1779
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153433347565919e-05,
+      "loss": 0.1611,
+      "step": 1780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142893861336656e-05,
+      "loss": 0.18,
+      "step": 1781
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132357829769651e-05,
+      "loss": 0.4179,
+      "step": 1782
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.121825265604514e-05,
+      "loss": 0.2783,
+      "step": 1783
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111296181576671e-05,
+      "loss": 0.287,
+      "step": 1784
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100770590417336e-05,
+      "loss": 0.1912,
+      "step": 1785
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090248504853497e-05,
+      "loss": 0.2106,
+      "step": 1786
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079729937607904e-05,
+      "loss": 0.343,
+      "step": 1787
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069214901399058e-05,
+      "loss": 0.2407,
+      "step": 1788
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.058703408941187e-05,
+      "loss": 0.2168,
+      "step": 1789
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048195472944233e-05,
+      "loss": 0.2373,
+      "step": 1790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.037691106113839e-05,
+      "loss": 0.2111,
+      "step": 1791
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027190321151329e-05,
+      "loss": 0.1708,
+      "step": 1792
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016693130753705e-05,
+      "loss": 0.3032,
+      "step": 1793
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006199547613613e-05,
+      "loss": 0.165,
+      "step": 1794
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.995709584419342e-05,
+      "loss": 0.3073,
+      "step": 1795
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9852232538548e-05,
+      "loss": 0.2345,
+      "step": 1796
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974740568599509e-05,
+      "loss": 0.187,
+      "step": 1797
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964261541328581e-05,
+      "loss": 0.3085,
+      "step": 1798
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.953786184712701e-05,
+      "loss": 0.2931,
+      "step": 1799
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.943314511418118e-05,
+      "loss": 0.2349,
+      "step": 1800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93284653410663e-05,
+      "loss": 0.2946,
+      "step": 1801
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.922382265435563e-05,
+      "loss": 0.2793,
+      "step": 1802
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.911921718057761e-05,
+      "loss": 0.289,
+      "step": 1803
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901464904621564e-05,
+      "loss": 0.2199,
+      "step": 1804
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891011837770802e-05,
+      "loss": 0.2307,
+      "step": 1805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880562530144771e-05,
+      "loss": 0.3637,
+      "step": 1806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870116994378228e-05,
+      "loss": 0.218,
+      "step": 1807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859675243101358e-05,
+      "loss": 0.1435,
+      "step": 1808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84923728893978e-05,
+      "loss": 0.2971,
+      "step": 1809
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838803144514519e-05,
+      "loss": 0.3218,
+      "step": 1810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828372822441996e-05,
+      "loss": 0.1835,
+      "step": 1811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.817946335333999e-05,
+      "loss": 0.2852,
+      "step": 1812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.807523695797692e-05,
+      "loss": 0.2897,
+      "step": 1813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797104916435582e-05,
+      "loss": 0.3614,
+      "step": 1814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.786690009845508e-05,
+      "loss": 0.2841,
+      "step": 1815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776278988620625e-05,
+      "loss": 0.2585,
+      "step": 1816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765871865349393e-05,
+      "loss": 0.2383,
+      "step": 1817
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755468652615556e-05,
+      "loss": 0.1741,
+      "step": 1818
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74506936299813e-05,
+      "loss": 0.2594,
+      "step": 1819
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.734674009071393e-05,
+      "loss": 0.2348,
+      "step": 1820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724282603404855e-05,
+      "loss": 0.3719,
+      "step": 1821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.713895158563258e-05,
+      "loss": 0.3083,
+      "step": 1822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703511687106552e-05,
+      "loss": 0.1776,
+      "step": 1823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693132201589886e-05,
+      "loss": 0.2796,
+      "step": 1824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682756714563586e-05,
+      "loss": 0.2518,
+      "step": 1825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672385238573143e-05,
+      "loss": 0.2293,
+      "step": 1826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.662017786159201e-05,
+      "loss": 0.2297,
+      "step": 1827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651654369857539e-05,
+      "loss": 0.1376,
+      "step": 1828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.64129500219905e-05,
+      "loss": 0.2286,
+      "step": 1829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630939695709739e-05,
+      "loss": 0.1814,
+      "step": 1830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.620588462910697e-05,
+      "loss": 0.1577,
+      "step": 1831
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610241316318089e-05,
+      "loss": 0.4429,
+      "step": 1832
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599898268443137e-05,
+      "loss": 0.2418,
+      "step": 1833
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589559331792114e-05,
+      "loss": 0.2598,
+      "step": 1834
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579224518866314e-05,
+      "loss": 0.2299,
+      "step": 1835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568893842162052e-05,
+      "loss": 0.1973,
+      "step": 1836
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558567314170636e-05,
+      "loss": 0.2277,
+      "step": 1837
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548244947378359e-05,
+      "loss": 0.3655,
+      "step": 1838
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537926754266486e-05,
+      "loss": 0.2338,
+      "step": 1839
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52761274731123e-05,
+      "loss": 0.4014,
+      "step": 1840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517302938983745e-05,
+      "loss": 0.2848,
+      "step": 1841
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50699734175011e-05,
+      "loss": 0.286,
+      "step": 1842
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496695968071309e-05,
+      "loss": 0.2576,
+      "step": 1843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486398830403222e-05,
+      "loss": 0.2351,
+      "step": 1844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476105941196605e-05,
+      "loss": 0.3632,
+      "step": 1845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465817312897076e-05,
+      "loss": 0.3005,
+      "step": 1846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.455532957945102e-05,
+      "loss": 0.237,
+      "step": 1847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445252888775985e-05,
+      "loss": 0.2674,
+      "step": 1848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434977117819844e-05,
+      "loss": 0.1236,
+      "step": 1849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.424705657501599e-05,
+      "loss": 0.2093,
+      "step": 1850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414438520240959e-05,
+      "loss": 0.2013,
+      "step": 1851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.404175718452404e-05,
+      "loss": 0.2954,
+      "step": 1852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.393917264545175e-05,
+      "loss": 0.2198,
+      "step": 1853
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.383663170923255e-05,
+      "loss": 0.2699,
+      "step": 1854
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373413449985353e-05,
+      "loss": 0.3124,
+      "step": 1855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.363168114124889e-05,
+      "loss": 0.1976,
+      "step": 1856
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352927175729986e-05,
+      "loss": 0.2758,
+      "step": 1857
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34269064718345e-05,
+      "loss": 0.3069,
+      "step": 1858
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332458540862746e-05,
+      "loss": 0.2222,
+      "step": 1859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.322230869140005e-05,
+      "loss": 0.2714,
+      "step": 1860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.312007644381983e-05,
+      "loss": 0.2341,
+      "step": 1861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.301788878950068e-05,
+      "loss": 0.1661,
+      "step": 1862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291574585200252e-05,
+      "loss": 0.3692,
+      "step": 1863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28136477548312e-05,
+      "loss": 0.1661,
+      "step": 1864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271159462143838e-05,
+      "loss": 0.3354,
+      "step": 1865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260958657522132e-05,
+      "loss": 0.273,
+      "step": 1866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250762373952282e-05,
+      "loss": 0.2567,
+      "step": 1867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240570623763089e-05,
+      "loss": 0.2899,
+      "step": 1868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.230383419277886e-05,
+      "loss": 0.2628,
+      "step": 1869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.220200772814504e-05,
+      "loss": 0.2144,
+      "step": 1870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.210022696685263e-05,
+      "loss": 0.2242,
+      "step": 1871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199849203196953e-05,
+      "loss": 0.2277,
+      "step": 1872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189680304650828e-05,
+      "loss": 0.1636,
+      "step": 1873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179516013342587e-05,
+      "loss": 0.2248,
+      "step": 1874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169356341562358e-05,
+      "loss": 0.1736,
+      "step": 1875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159201301594672e-05,
+      "loss": 0.1815,
+      "step": 1876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149050905718476e-05,
+      "loss": 0.2256,
+      "step": 1877
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13890516620709e-05,
+      "loss": 0.1997,
+      "step": 1878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.128764095328209e-05,
+      "loss": 0.205,
+      "step": 1879
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.118627705343882e-05,
+      "loss": 0.164,
+      "step": 1880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1084960085105e-05,
+      "loss": 0.1626,
+      "step": 1881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098369017078773e-05,
+      "loss": 0.4005,
+      "step": 1882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0882467432937284e-05,
+      "loss": 0.3327,
+      "step": 1883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078129199394686e-05,
+      "loss": 0.3189,
+      "step": 1884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068016397615246e-05,
+      "loss": 0.3157,
+      "step": 1885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057908350183279e-05,
+      "loss": 0.244,
+      "step": 1886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0478050693208995e-05,
+      "loss": 0.3199,
+      "step": 1887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037706567244463e-05,
+      "loss": 0.2685,
+      "step": 1888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0276128561645484e-05,
+      "loss": 0.193,
+      "step": 1889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017523948285941e-05,
+      "loss": 0.2728,
+      "step": 1890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007439855807613e-05,
+      "loss": 0.1907,
+      "step": 1891
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.997360590922718e-05,
+      "loss": 0.3031,
+      "step": 1892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987286165818573e-05,
+      "loss": 0.1955,
+      "step": 1893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9772165926766464e-05,
+      "loss": 0.31,
+      "step": 1894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9671518836725283e-05,
+      "loss": 0.2216,
+      "step": 1895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95709205097594e-05,
+      "loss": 0.36,
+      "step": 1896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9470371067506994e-05,
+      "loss": 0.3153,
+      "step": 1897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936987063154714e-05,
+      "loss": 0.1482,
+      "step": 1898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.926941932339973e-05,
+      "loss": 0.2866,
+      "step": 1899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9169017264525125e-05,
+      "loss": 0.1843,
+      "step": 1900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906866457632424e-05,
+      "loss": 0.2699,
+      "step": 1901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.896836138013826e-05,
+      "loss": 0.1867,
+      "step": 1902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886810779724853e-05,
+      "loss": 0.3086,
+      "step": 1903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.876790394887638e-05,
+      "loss": 0.3118,
+      "step": 1904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866774995618304e-05,
+      "loss": 0.1412,
+      "step": 1905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856764594026944e-05,
+      "loss": 0.3284,
+      "step": 1906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846759202217611e-05,
+      "loss": 0.2633,
+      "step": 1907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8367588322882936e-05,
+      "loss": 0.3124,
+      "step": 1908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8267634963309135e-05,
+      "loss": 0.2527,
+      "step": 1909
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.816773206431306e-05,
+      "loss": 0.2115,
+      "step": 1910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806787974669203e-05,
+      "loss": 0.3636,
+      "step": 1911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796807813118219e-05,
+      "loss": 0.2756,
+      "step": 1912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786832733845846e-05,
+      "loss": 0.2645,
+      "step": 1913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7768627489134166e-05,
+      "loss": 0.2406,
+      "step": 1914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766897870376113e-05,
+      "loss": 0.2439,
+      "step": 1915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756938110282943e-05,
+      "loss": 0.3155,
+      "step": 1916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746983480676724e-05,
+      "loss": 0.1942,
+      "step": 1917
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73703399359407e-05,
+      "loss": 0.2343,
+      "step": 1918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727089661065374e-05,
+      "loss": 0.2219,
+      "step": 1919
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7171504951148046e-05,
+      "loss": 0.1652,
+      "step": 1920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7072165077602736e-05,
+      "loss": 0.2446,
+      "step": 1921
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6972877110134404e-05,
+      "loss": 0.215,
+      "step": 1922
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687364116879674e-05,
+      "loss": 0.2485,
+      "step": 1923
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677445737358068e-05,
+      "loss": 0.2705,
+      "step": 1924
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667532584441403e-05,
+      "loss": 0.2507,
+      "step": 1925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657624670116142e-05,
+      "loss": 0.2148,
+      "step": 1926
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6477220063624104e-05,
+      "loss": 0.4317,
+      "step": 1927
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6378246051539916e-05,
+      "loss": 0.3139,
+      "step": 1928
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6279324784583e-05,
+      "loss": 0.1546,
+      "step": 1929
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.618045638236375e-05,
+      "loss": 0.3086,
+      "step": 1930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608164096442866e-05,
+      "loss": 0.3428,
+      "step": 1931
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5982878650260054e-05,
+      "loss": 0.1805,
+      "step": 1932
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5884169559276154e-05,
+      "loss": 0.3231,
+      "step": 1933
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5785513810830805e-05,
+      "loss": 0.3145,
+      "step": 1934
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568691152421332e-05,
+      "loss": 0.319,
+      "step": 1935
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558836281864843e-05,
+      "loss": 0.2853,
+      "step": 1936
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548986781329599e-05,
+      "loss": 0.253,
+      "step": 1937
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539142662725102e-05,
+      "loss": 0.2115,
+      "step": 1938
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.529303937954342e-05,
+      "loss": 0.2245,
+      "step": 1939
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.519470618913782e-05,
+      "loss": 0.2191,
+      "step": 1940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509642717493354e-05,
+      "loss": 0.2193,
+      "step": 1941
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.499820245576441e-05,
+      "loss": 0.1565,
+      "step": 1942
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490003215039858e-05,
+      "loss": 0.3273,
+      "step": 1943
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480191637753842e-05,
+      "loss": 0.228,
+      "step": 1944
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4703855255820355e-05,
+      "loss": 0.208,
+      "step": 1945
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4605848903814705e-05,
+      "loss": 0.2368,
+      "step": 1946
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4507897440025626e-05,
+      "loss": 0.2271,
+      "step": 1947
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.441000098289086e-05,
+      "loss": 0.1602,
+      "step": 1948
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4312159650781627e-05,
+      "loss": 0.2928,
+      "step": 1949
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.42143735620025e-05,
+      "loss": 0.3169,
+      "step": 1950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41166428347913e-05,
+      "loss": 0.2181,
+      "step": 1951
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401896758731886e-05,
+      "loss": 0.2805,
+      "step": 1952
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3921347937688936e-05,
+      "loss": 0.3829,
+      "step": 1953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382378400393808e-05,
+      "loss": 0.2484,
+      "step": 1954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372627590403547e-05,
+      "loss": 0.2405,
+      "step": 1955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36288237558828e-05,
+      "loss": 0.32,
+      "step": 1956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3531427677313987e-05,
+      "loss": 0.211,
+      "step": 1957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3434087786095264e-05,
+      "loss": 0.1786,
+      "step": 1958
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333680419992493e-05,
+      "loss": 0.1532,
+      "step": 1959
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3239577036433165e-05,
+      "loss": 0.1439,
+      "step": 1960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.314240641318191e-05,
+      "loss": 0.219,
+      "step": 1961
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304529244766481e-05,
+      "loss": 0.2088,
+      "step": 1962
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2948235257306923e-05,
+      "loss": 0.3566,
+      "step": 1963
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285123495946469e-05,
+      "loss": 0.7649,
+      "step": 1964
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2754291671425825e-05,
+      "loss": 0.3363,
+      "step": 1965
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2657405510408944e-05,
+      "loss": 0.1466,
+      "step": 1966
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2560576593563746e-05,
+      "loss": 0.233,
+      "step": 1967
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246380503797065e-05,
+      "loss": 0.2379,
+      "step": 1968
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.236709096064072e-05,
+      "loss": 0.2616,
+      "step": 1969
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227043447851554e-05,
+      "loss": 0.1082,
+      "step": 1970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.217383570846701e-05,
+      "loss": 0.2033,
+      "step": 1971
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2077294767297305e-05,
+      "loss": 0.238,
+      "step": 1972
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1980811771738656e-05,
+      "loss": 0.252,
+      "step": 1973
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1884386838453157e-05,
+      "loss": 0.2198,
+      "step": 1974
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1788020084032805e-05,
+      "loss": 0.2297,
+      "step": 1975
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1691711624999186e-05,
+      "loss": 0.2356,
+      "step": 1976
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.159546157780344e-05,
+      "loss": 0.2663,
+      "step": 1977
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149927005882602e-05,
+      "loss": 0.2689,
+      "step": 1978
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.140313718437668e-05,
+      "loss": 0.2039,
+      "step": 1979
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13070630706942e-05,
+      "loss": 0.2102,
+      "step": 1980
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121104783394636e-05,
+      "loss": 0.3345,
+      "step": 1981
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1115091590229746e-05,
+      "loss": 0.1987,
+      "step": 1982
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.101919445556953e-05,
+      "loss": 0.2709,
+      "step": 1983
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092335654591952e-05,
+      "loss": 0.1452,
+      "step": 1984
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0827577977161845e-05,
+      "loss": 0.1208,
+      "step": 1985
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073185886510694e-05,
+      "loss": 0.0943,
+      "step": 1986
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0636199325493286e-05,
+      "loss": 0.1631,
+      "step": 1987
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.054059947398735e-05,
+      "loss": 0.3266,
+      "step": 1988
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044505942618347e-05,
+      "loss": 0.2578,
+      "step": 1989
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0349579297603636e-05,
+      "loss": 0.2032,
+      "step": 1990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02541592036974e-05,
+      "loss": 0.2266,
+      "step": 1991
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015879925984166e-05,
+      "loss": 0.2169,
+      "step": 1992
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.006349958134067e-05,
+      "loss": 0.2408,
+      "step": 1993
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996826028342578e-05,
+      "loss": 0.3331,
+      "step": 1994
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987308148125533e-05,
+      "loss": 0.2637,
+      "step": 1995
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977796328991453e-05,
+      "loss": 0.2152,
+      "step": 1996
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968290582441527e-05,
+      "loss": 0.2525,
+      "step": 1997
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958790919969604e-05,
+      "loss": 0.3325,
+      "step": 1998
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949297353062179e-05,
+      "loss": 0.2403,
+      "step": 1999
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.939809893198365e-05,
+      "loss": 0.2204,
+      "step": 2000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930328551849903e-05,
+      "loss": 0.1694,
+      "step": 2001
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920853340481132e-05,
+      "loss": 0.146,
+      "step": 2002
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.911384270548976e-05,
+      "loss": 0.1693,
+      "step": 2003
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.901921353502941e-05,
+      "loss": 0.1184,
+      "step": 2004
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8924646007850825e-05,
+      "loss": 0.152,
+      "step": 2005
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8830140238300104e-05,
+      "loss": 0.1559,
+      "step": 2006
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.873569634064863e-05,
+      "loss": 0.1325,
+      "step": 2007
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.864131442909306e-05,
+      "loss": 0.1465,
+      "step": 2008
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85469946177549e-05,
+      "loss": 0.189,
+      "step": 2009
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8452737020680796e-05,
+      "loss": 0.1346,
+      "step": 2010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835854175184202e-05,
+      "loss": 0.1594,
+      "step": 2011
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8264408925134576e-05,
+      "loss": 0.2495,
+      "step": 2012
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8170338654378874e-05,
+      "loss": 0.1561,
+      "step": 2013
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8076331053319766e-05,
+      "loss": 0.3041,
+      "step": 2014
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.798238623562626e-05,
+      "loss": 0.3122,
+      "step": 2015
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788850431489156e-05,
+      "loss": 0.1683,
+      "step": 2016
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779468540463263e-05,
+      "loss": 0.187,
+      "step": 2017
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77009296182904e-05,
+      "loss": 0.2098,
+      "step": 2018
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760723706922944e-05,
+      "loss": 0.2765,
+      "step": 2019
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751360787073783e-05,
+      "loss": 0.1831,
+      "step": 2020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742004213602708e-05,
+      "loss": 0.1479,
+      "step": 2021
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732653997823193e-05,
+      "loss": 0.1933,
+      "step": 2022
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723310151041029e-05,
+      "loss": 0.2159,
+      "step": 2023
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7139726845543006e-05,
+      "loss": 0.2076,
+      "step": 2024
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7046416096533866e-05,
+      "loss": 0.2141,
+      "step": 2025
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695316937620923e-05,
+      "loss": 0.3633,
+      "step": 2026
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685998679731815e-05,
+      "loss": 0.2009,
+      "step": 2027
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.676686847253209e-05,
+      "loss": 0.1613,
+      "step": 2028
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6673814514444836e-05,
+      "loss": 0.2259,
+      "step": 2029
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658082503557234e-05,
+      "loss": 0.1934,
+      "step": 2030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.648790014835256e-05,
+      "loss": 0.225,
+      "step": 2031
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639503996514538e-05,
+      "loss": 0.1913,
+      "step": 2032
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63022445982325e-05,
+      "loss": 0.2027,
+      "step": 2033
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620951415981709e-05,
+      "loss": 0.1311,
+      "step": 2034
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6116848762023965e-05,
+      "loss": 0.2201,
+      "step": 2035
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602424851689924e-05,
+      "loss": 0.1418,
+      "step": 2036
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593171353641026e-05,
+      "loss": 0.1892,
+      "step": 2037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.583924393244543e-05,
+      "loss": 0.232,
+      "step": 2038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574683981681413e-05,
+      "loss": 0.2062,
+      "step": 2039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5654501301246574e-05,
+      "loss": 0.1667,
+      "step": 2040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556222849739361e-05,
+      "loss": 0.2173,
+      "step": 2041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547002151682669e-05,
+      "loss": 0.1818,
+      "step": 2042
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.537788047103756e-05,
+      "loss": 0.1752,
+      "step": 2043
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.528580547143837e-05,
+      "loss": 0.2265,
+      "step": 2044
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519379662936135e-05,
+      "loss": 0.1854,
+      "step": 2045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510185405605876e-05,
+      "loss": 0.2153,
+      "step": 2046
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.50099778627027e-05,
+      "loss": 0.2,
+      "step": 2047
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.491816816038503e-05,
+      "loss": 0.1828,
+      "step": 2048
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4826425060117205e-05,
+      "loss": 0.2565,
+      "step": 2049
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473474867283018e-05,
+      "loss": 0.211,
+      "step": 2050
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4643139109374166e-05,
+      "loss": 0.1841,
+      "step": 2051
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455159648051863e-05,
+      "loss": 0.1603,
+      "step": 2052
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446012089695211e-05,
+      "loss": 0.204,
+      "step": 2053
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436871246928206e-05,
+      "loss": 0.2007,
+      "step": 2054
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4277371308034744e-05,
+      "loss": 0.1664,
+      "step": 2055
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4186097523655066e-05,
+      "loss": 0.1554,
+      "step": 2056
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4094891226506496e-05,
+      "loss": 0.112,
+      "step": 2057
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400375252687088e-05,
+      "loss": 0.1367,
+      "step": 2058
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3912681534948366e-05,
+      "loss": 0.1115,
+      "step": 2059
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382167836085713e-05,
+      "loss": 0.1109,
+      "step": 2060
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373074311463344e-05,
+      "loss": 0.1361,
+      "step": 2061
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363987590623144e-05,
+      "loss": 0.167,
+      "step": 2062
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354907684552294e-05,
+      "loss": 0.1399,
+      "step": 2063
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345834604229739e-05,
+      "loss": 0.1387,
+      "step": 2064
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.336768360626171e-05,
+      "loss": 0.2267,
+      "step": 2065
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.327708964704012e-05,
+      "loss": 0.1924,
+      "step": 2066
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318656427417409e-05,
+      "loss": 0.1393,
+      "step": 2067
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.309610759712216e-05,
+      "loss": 0.1639,
+      "step": 2068
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3005719725259695e-05,
+      "loss": 0.2269,
+      "step": 2069
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.291540076787899e-05,
+      "loss": 0.1278,
+      "step": 2070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2825150834188974e-05,
+      "loss": 0.2683,
+      "step": 2071
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273497003331512e-05,
+      "loss": 0.2209,
+      "step": 2072
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26448584742993e-05,
+      "loss": 0.2967,
+      "step": 2073
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255481626609966e-05,
+      "loss": 0.2299,
+      "step": 2074
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246484351759049e-05,
+      "loss": 0.2807,
+      "step": 2075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2374940337562154e-05,
+      "loss": 0.2028,
+      "step": 2076
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.228510683472074e-05,
+      "loss": 0.2053,
+      "step": 2077
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219534311768826e-05,
+      "loss": 0.1917,
+      "step": 2078
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2105649295002236e-05,
+      "loss": 0.1936,
+      "step": 2079
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2016025475115736e-05,
+      "loss": 0.2657,
+      "step": 2080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1926471766397166e-05,
+      "loss": 0.2059,
+      "step": 2081
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183698827713015e-05,
+      "loss": 0.2406,
+      "step": 2082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1747575115513405e-05,
+      "loss": 0.1326,
+      "step": 2083
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165823238966061e-05,
+      "loss": 0.209,
+      "step": 2084
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156896020760033e-05,
+      "loss": 0.144,
+      "step": 2085
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147975867727571e-05,
+      "loss": 0.1678,
+      "step": 2086
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.139062790654458e-05,
+      "loss": 0.2338,
+      "step": 2087
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130156800317917e-05,
+      "loss": 0.2689,
+      "step": 2088
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.121257907486603e-05,
+      "loss": 0.2674,
+      "step": 2089
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1123661229205876e-05,
+      "loss": 0.1626,
+      "step": 2090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1034814573713495e-05,
+      "loss": 0.199,
+      "step": 2091
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0946039215817575e-05,
+      "loss": 0.1604,
+      "step": 2092
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085733526286065e-05,
+      "loss": 0.2541,
+      "step": 2093
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076870282209878e-05,
+      "loss": 0.2398,
+      "step": 2094
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068014200070168e-05,
+      "loss": 0.1782,
+      "step": 2095
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0591652905752444e-05,
+      "loss": 0.2674,
+      "step": 2096
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0503235644247416e-05,
+      "loss": 0.1259,
+      "step": 2097
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04148903230961e-05,
+      "loss": 0.2103,
+      "step": 2098
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.032661704912099e-05,
+      "loss": 0.2139,
+      "step": 2099
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.02384159290575e-05,
+      "loss": 0.1639,
+      "step": 2100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0150287069553763e-05,
+      "loss": 0.1986,
+      "step": 2101
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.006223057717059e-05,
+      "loss": 0.2588,
+      "step": 2102
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997424655838119e-05,
+      "loss": 0.1911,
+      "step": 2103
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.988633511957124e-05,
+      "loss": 0.1861,
+      "step": 2104
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9798496367038605e-05,
+      "loss": 0.2428,
+      "step": 2105
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97107304069933e-05,
+      "loss": 0.1579,
+      "step": 2106
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962303734555729e-05,
+      "loss": 0.2193,
+      "step": 2107
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953541728876441e-05,
+      "loss": 0.2908,
+      "step": 2108
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9447870342560226e-05,
+      "loss": 0.1686,
+      "step": 2109
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9360396612801896e-05,
+      "loss": 0.1656,
+      "step": 2110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9272996205258015e-05,
+      "loss": 0.1478,
+      "step": 2111
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918566922560857e-05,
+      "loss": 0.1563,
+      "step": 2112
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.909841577944474e-05,
+      "loss": 0.1787,
+      "step": 2113
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9011235972268814e-05,
+      "loss": 0.2235,
+      "step": 2114
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8924129909494e-05,
+      "loss": 0.165,
+      "step": 2115
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883709769644438e-05,
+      "loss": 0.2862,
+      "step": 2116
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8750139438354716e-05,
+      "loss": 0.2134,
+      "step": 2117
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866325524037034e-05,
+      "loss": 0.2062,
+      "step": 2118
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8576445207547116e-05,
+      "loss": 0.1868,
+      "step": 2119
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848970944485106e-05,
+      "loss": 0.1916,
+      "step": 2120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840304805715854e-05,
+      "loss": 0.1984,
+      "step": 2121
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8316461149255945e-05,
+      "loss": 0.2739,
+      "step": 2122
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8229948825839576e-05,
+      "loss": 0.2613,
+      "step": 2123
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8143511191515615e-05,
+      "loss": 0.1873,
+      "step": 2124
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.805714835079988e-05,
+      "loss": 0.142,
+      "step": 2125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.797086040811776e-05,
+      "loss": 0.1287,
+      "step": 2126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7884647467804135e-05,
+      "loss": 0.2473,
+      "step": 2127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7798509634103086e-05,
+      "loss": 0.1567,
+      "step": 2128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771244701116797e-05,
+      "loss": 0.2319,
+      "step": 2129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7626459703061176e-05,
+      "loss": 0.2384,
+      "step": 2130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7540547813754045e-05,
+      "loss": 0.2582,
+      "step": 2131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7454711447126686e-05,
+      "loss": 0.2156,
+      "step": 2132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.736895070696793e-05,
+      "loss": 0.1051,
+      "step": 2133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7283265696975145e-05,
+      "loss": 0.2409,
+      "step": 2134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.719765652075412e-05,
+      "loss": 0.1759,
+      "step": 2135
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7112123281819e-05,
+      "loss": 0.2324,
+      "step": 2136
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702666608359201e-05,
+      "loss": 0.1215,
+      "step": 2137
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6941285029403526e-05,
+      "loss": 0.1229,
+      "step": 2138
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6855980222491824e-05,
+      "loss": 0.1967,
+      "step": 2139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.677075176600297e-05,
+      "loss": 0.2041,
+      "step": 2140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6685599762990755e-05,
+      "loss": 0.2513,
+      "step": 2141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.660052431641645e-05,
+      "loss": 0.1883,
+      "step": 2142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.651552552914884e-05,
+      "loss": 0.1932,
+      "step": 2143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643060350396397e-05,
+      "loss": 0.207,
+      "step": 2144
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6345758343545124e-05,
+      "loss": 0.1682,
+      "step": 2145
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.626099015048251e-05,
+      "loss": 0.2482,
+      "step": 2146
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.617629902727341e-05,
+      "loss": 0.2404,
+      "step": 2147
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.609168507632187e-05,
+      "loss": 0.2132,
+      "step": 2148
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.600714839993863e-05,
+      "loss": 0.1719,
+      "step": 2149
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.592268910034097e-05,
+      "loss": 0.1865,
+      "step": 2150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.583830727965264e-05,
+      "loss": 0.2159,
+      "step": 2151
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.575400303990368e-05,
+      "loss": 0.1804,
+      "step": 2152
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5669776483030394e-05,
+      "loss": 0.1325,
+      "step": 2153
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5585627710875004e-05,
+      "loss": 0.1151,
+      "step": 2154
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5501556825185825e-05,
+      "loss": 0.1497,
+      "step": 2155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.541756392761693e-05,
+      "loss": 0.1481,
+      "step": 2156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.533364911972812e-05,
+      "loss": 0.1909,
+      "step": 2157
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524981250298475e-05,
+      "loss": 0.1477,
+      "step": 2158
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516605417875763e-05,
+      "loss": 0.1719,
+      "step": 2159
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.508237424832296e-05,
+      "loss": 0.1412,
+      "step": 2160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4998772812862044e-05,
+      "loss": 0.1788,
+      "step": 2161
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.491524997346142e-05,
+      "loss": 0.1496,
+      "step": 2162
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.483180583111241e-05,
+      "loss": 0.1897,
+      "step": 2163
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.474844048671132e-05,
+      "loss": 0.3288,
+      "step": 2164
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4665154041059125e-05,
+      "loss": 0.1657,
+      "step": 2165
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4581946594861414e-05,
+      "loss": 0.1671,
+      "step": 2166
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4498818248728235e-05,
+      "loss": 0.2211,
+      "step": 2167
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.441576910317403e-05,
+      "loss": 0.1832,
+      "step": 2168
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.433279925861744e-05,
+      "loss": 0.2241,
+      "step": 2169
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.424990881538127e-05,
+      "loss": 0.2092,
+      "step": 2170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.41670978736922e-05,
+      "loss": 0.2371,
+      "step": 2171
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.40843665336809e-05,
+      "loss": 0.2386,
+      "step": 2172
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4001714895381745e-05,
+      "loss": 0.2504,
+      "step": 2173
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.391914305873274e-05,
+      "loss": 0.1857,
+      "step": 2174
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3836651123575405e-05,
+      "loss": 0.2883,
+      "step": 2175
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3754239189654644e-05,
+      "loss": 0.2556,
+      "step": 2176
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.367190735661863e-05,
+      "loss": 0.1212,
+      "step": 2177
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.358965572401866e-05,
+      "loss": 0.1668,
+      "step": 2178
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3507484391309144e-05,
+      "loss": 0.2162,
+      "step": 2179
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3425393457847235e-05,
+      "loss": 0.2266,
+      "step": 2180
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.334338302289301e-05,
+      "loss": 0.2417,
+      "step": 2181
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.326145318560916e-05,
+      "loss": 0.1905,
+      "step": 2182
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.317960404506093e-05,
+      "loss": 0.2081,
+      "step": 2183
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3097835700216e-05,
+      "loss": 0.1823,
+      "step": 2184
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3016148249944334e-05,
+      "loss": 0.233,
+      "step": 2185
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293454179301809e-05,
+      "loss": 0.2464,
+      "step": 2186
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.285301642811156e-05,
+      "loss": 0.2178,
+      "step": 2187
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.277157225380082e-05,
+      "loss": 0.2407,
+      "step": 2188
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.269020936856393e-05,
+      "loss": 0.2131,
+      "step": 2189
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260892787078059e-05,
+      "loss": 0.2588,
+      "step": 2190
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.252772785873213e-05,
+      "loss": 0.2057,
+      "step": 2191
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2446609430601294e-05,
+      "loss": 0.2116,
+      "step": 2192
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2365572684472245e-05,
+      "loss": 0.1795,
+      "step": 2193
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2284617718330335e-05,
+      "loss": 0.2297,
+      "step": 2194
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.220374463006204e-05,
+      "loss": 0.1603,
+      "step": 2195
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.212295351745487e-05,
+      "loss": 0.2196,
+      "step": 2196
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.204224447819712e-05,
+      "loss": 0.2546,
+      "step": 2197
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196161760987797e-05,
+      "loss": 0.3216,
+      "step": 2198
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.188107300998714e-05,
+      "loss": 0.1899,
+      "step": 2199
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.180061077591493e-05,
+      "loss": 0.1959,
+      "step": 2200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172023100495205e-05,
+      "loss": 0.2172,
+      "step": 2201
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163993379428948e-05,
+      "loss": 0.1109,
+      "step": 2202
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.155971924101837e-05,
+      "loss": 0.1599,
+      "step": 2203
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.147958744212996e-05,
+      "loss": 0.2298,
+      "step": 2204
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.139953849451536e-05,
+      "loss": 0.1316,
+      "step": 2205
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1319572494965554e-05,
+      "loss": 0.2022,
+      "step": 2206
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1239689540171215e-05,
+      "loss": 0.1877,
+      "step": 2207
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1159889726722626e-05,
+      "loss": 0.1664,
+      "step": 2208
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.108017315110949e-05,
+      "loss": 0.104,
+      "step": 2209
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.10005399097209e-05,
+      "loss": 0.1585,
+      "step": 2210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.092099009884519e-05,
+      "loss": 0.1869,
+      "step": 2211
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0841523814669806e-05,
+      "loss": 0.1345,
+      "step": 2212
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0762141153281197e-05,
+      "loss": 0.1621,
+      "step": 2213
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0682842210664655e-05,
+      "loss": 0.2979,
+      "step": 2214
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.06036270827043e-05,
+      "loss": 0.1742,
+      "step": 2215
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.05244958651829e-05,
+      "loss": 0.1912,
+      "step": 2216
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0445448653781773e-05,
+      "loss": 0.2072,
+      "step": 2217
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0366485544080613e-05,
+      "loss": 0.2804,
+      "step": 2218
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0287606631557473e-05,
+      "loss": 0.2338,
+      "step": 2219
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0208812011588574e-05,
+      "loss": 0.138,
+      "step": 2220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0130101779448216e-05,
+      "loss": 0.1853,
+      "step": 2221
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0051476030308712e-05,
+      "loss": 0.2387,
+      "step": 2222
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9972934859240097e-05,
+      "loss": 0.2853,
+      "step": 2223
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9894478361210254e-05,
+      "loss": 0.2505,
+      "step": 2224
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.981610663108465e-05,
+      "loss": 0.2399,
+      "step": 2225
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.973781976362625e-05,
+      "loss": 0.2466,
+      "step": 2226
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9659617853495425e-05,
+      "loss": 0.1113,
+      "step": 2227
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958150099524979e-05,
+      "loss": 0.211,
+      "step": 2228
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.950346928334413e-05,
+      "loss": 0.2121,
+      "step": 2229
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9425522812130336e-05,
+      "loss": 0.099,
+      "step": 2230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9347661675857096e-05,
+      "loss": 0.2156,
+      "step": 2231
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.926988596867003e-05,
+      "loss": 0.2145,
+      "step": 2232
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.919219578461142e-05,
+      "loss": 0.286,
+      "step": 2233
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9114591217620158e-05,
+      "loss": 0.2801,
+      "step": 2234
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9037072361531592e-05,
+      "loss": 0.1193,
+      "step": 2235
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8959639310077437e-05,
+      "loss": 0.1333,
+      "step": 2236
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.888229215688566e-05,
+      "loss": 0.1819,
+      "step": 2237
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8805030995480365e-05,
+      "loss": 0.1828,
+      "step": 2238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872785591928172e-05,
+      "loss": 0.1764,
+      "step": 2239
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8650767021605686e-05,
+      "loss": 0.1504,
+      "step": 2240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.857376439566414e-05,
+      "loss": 0.2525,
+      "step": 2241
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8496848134564603e-05,
+      "loss": 0.199,
+      "step": 2242
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.842001833131015e-05,
+      "loss": 0.2189,
+      "step": 2243
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8343275078799348e-05,
+      "loss": 0.2036,
+      "step": 2244
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826661846982609e-05,
+      "loss": 0.211,
+      "step": 2245
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819004859707951e-05,
+      "loss": 0.2085,
+      "step": 2246
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8113565553143894e-05,
+      "loss": 0.231,
+      "step": 2247
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8037169430498456e-05,
+      "loss": 0.2869,
+      "step": 2248
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.796086032151738e-05,
+      "loss": 0.199,
+      "step": 2249
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7884638318469648e-05,
+      "loss": 0.2801,
+      "step": 2250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.780850351351886e-05,
+      "loss": 0.1868,
+      "step": 2251
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.773245599872323e-05,
+      "loss": 0.1658,
+      "step": 2252
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7656495866035405e-05,
+      "loss": 0.1535,
+      "step": 2253
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.758062320730238e-05,
+      "loss": 0.1565,
+      "step": 2254
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7504838114265363e-05,
+      "loss": 0.227,
+      "step": 2255
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742914067855975e-05,
+      "loss": 0.1769,
+      "step": 2256
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7353530991714815e-05,
+      "loss": 0.1745,
+      "step": 2257
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7278009145153837e-05,
+      "loss": 0.1339,
+      "step": 2258
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7202575230193873e-05,
+      "loss": 0.1274,
+      "step": 2259
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7127229338045624e-05,
+      "loss": 0.1592,
+      "step": 2260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.705197155981337e-05,
+      "loss": 0.1317,
+      "step": 2261
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697680198649487e-05,
+      "loss": 0.2171,
+      "step": 2262
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6901720708981205e-05,
+      "loss": 0.1285,
+      "step": 2263
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.682672781805673e-05,
+      "loss": 0.1769,
+      "step": 2264
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6751823404398845e-05,
+      "loss": 0.1559,
+      "step": 2265
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6677007558578037e-05,
+      "loss": 0.2089,
+      "step": 2266
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6602280371057697e-05,
+      "loss": 0.282,
+      "step": 2267
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6527641932194004e-05,
+      "loss": 0.2113,
+      "step": 2268
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.645309233223583e-05,
+      "loss": 0.2087,
+      "step": 2269
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6378631661324616e-05,
+      "loss": 0.1296,
+      "step": 2270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6304260009494286e-05,
+      "loss": 0.1757,
+      "step": 2271
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.622997746667113e-05,
+      "loss": 0.2748,
+      "step": 2272
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615578412267371e-05,
+      "loss": 0.1543,
+      "step": 2273
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6081680067212644e-05,
+      "loss": 0.1454,
+      "step": 2274
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.60076653898907e-05,
+      "loss": 0.2433,
+      "step": 2275
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5933740180202514e-05,
+      "loss": 0.1634,
+      "step": 2276
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5859904527534563e-05,
+      "loss": 0.2362,
+      "step": 2277
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5786158521165016e-05,
+      "loss": 0.1933,
+      "step": 2278
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.571250225026368e-05,
+      "loss": 0.2451,
+      "step": 2279
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.563893580389183e-05,
+      "loss": 0.174,
+      "step": 2280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5565459271002156e-05,
+      "loss": 0.1808,
+      "step": 2281
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5492072740438567e-05,
+      "loss": 0.1745,
+      "step": 2282
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.541877630093621e-05,
+      "loss": 0.1625,
+      "step": 2283
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.534557004112128e-05,
+      "loss": 0.2763,
+      "step": 2284
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527245404951094e-05,
+      "loss": 0.2337,
+      "step": 2285
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5199428414513192e-05,
+      "loss": 0.26,
+      "step": 2286
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5126493224426785e-05,
+      "loss": 0.1818,
+      "step": 2287
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5053648567441112e-05,
+      "loss": 0.2643,
+      "step": 2288
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498089453163609e-05,
+      "loss": 0.18,
+      "step": 2289
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4908231204982103e-05,
+      "loss": 0.1944,
+      "step": 2290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4835658675339757e-05,
+      "loss": 0.1854,
+      "step": 2291
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.476317703045995e-05,
+      "loss": 0.2177,
+      "step": 2292
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.46907863579837e-05,
+      "loss": 0.1963,
+      "step": 2293
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461848674544196e-05,
+      "loss": 0.1725,
+      "step": 2294
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454627828025563e-05,
+      "loss": 0.2011,
+      "step": 2295
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.447416104973539e-05,
+      "loss": 0.2363,
+      "step": 2296
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4402135141081585e-05,
+      "loss": 0.1741,
+      "step": 2297
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.433020064138415e-05,
+      "loss": 0.2115,
+      "step": 2298
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.425835763762253e-05,
+      "loss": 0.2971,
+      "step": 2299
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.418660621666544e-05,
+      "loss": 0.2673,
+      "step": 2300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4114946465270937e-05,
+      "loss": 0.1829,
+      "step": 2301
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.404337847008624e-05,
+      "loss": 0.1895,
+      "step": 2302
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3971902317647586e-05,
+      "loss": 0.1977,
+      "step": 2303
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3900518094380177e-05,
+      "loss": 0.1863,
+      "step": 2304
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.382922588659804e-05,
+      "loss": 0.1226,
+      "step": 2305
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3758025780503978e-05,
+      "loss": 0.1831,
+      "step": 2306
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3686917862189438e-05,
+      "loss": 0.1454,
+      "step": 2307
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3615902217634302e-05,
+      "loss": 0.2535,
+      "step": 2308
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3544978932706985e-05,
+      "loss": 0.1624,
+      "step": 2309
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3474148093164172e-05,
+      "loss": 0.1666,
+      "step": 2310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34034097846508e-05,
+      "loss": 0.1348,
+      "step": 2311
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3332764092699888e-05,
+      "loss": 0.149,
+      "step": 2312
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3262211102732502e-05,
+      "loss": 0.1795,
+      "step": 2313
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3191750900057596e-05,
+      "loss": 0.1824,
+      "step": 2314
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3121383569871944e-05,
+      "loss": 0.2022,
+      "step": 2315
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3051109197260023e-05,
+      "loss": 0.1992,
+      "step": 2316
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.298092786719387e-05,
+      "loss": 0.3141,
+      "step": 2317
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291083966453307e-05,
+      "loss": 0.2012,
+      "step": 2318
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2840844674024574e-05,
+      "loss": 0.1317,
+      "step": 2319
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2770942980302655e-05,
+      "loss": 0.2148,
+      "step": 2320
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2701134667888757e-05,
+      "loss": 0.2018,
+      "step": 2321
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2631419821191412e-05,
+      "loss": 0.1639,
+      "step": 2322
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2561798524506138e-05,
+      "loss": 0.1697,
+      "step": 2323
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2492270862015374e-05,
+      "loss": 0.1298,
+      "step": 2324
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242283691778824e-05,
+      "loss": 0.1516,
+      "step": 2325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2353496775780645e-05,
+      "loss": 0.2275,
+      "step": 2326
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2284250519835025e-05,
+      "loss": 0.1664,
+      "step": 2327
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.221509823368032e-05,
+      "loss": 0.1924,
+      "step": 2328
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.214604000093182e-05,
+      "loss": 0.1365,
+      "step": 2329
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2077075905091117e-05,
+      "loss": 0.1336,
+      "step": 2330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2008206029545943e-05,
+      "loss": 0.2176,
+      "step": 2331
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1939430457570133e-05,
+      "loss": 0.2115,
+      "step": 2332
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.187074927232351e-05,
+      "loss": 0.2204,
+      "step": 2333
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1802162556851703e-05,
+      "loss": 0.2295,
+      "step": 2334
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.173367039408617e-05,
+      "loss": 0.1524,
+      "step": 2335
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1665272866844032e-05,
+      "loss": 0.1782,
+      "step": 2336
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.159697005782798e-05,
+      "loss": 0.2643,
+      "step": 2337
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1528762049626172e-05,
+      "loss": 0.2885,
+      "step": 2338
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1460648924712134e-05,
+      "loss": 0.222,
+      "step": 2339
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1392630765444665e-05,
+      "loss": 0.1839,
+      "step": 2340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1324707654067788e-05,
+      "loss": 0.2678,
+      "step": 2341
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1256879672710472e-05,
+      "loss": 0.3113,
+      "step": 2342
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.118914690338678e-05,
+      "loss": 0.1562,
+      "step": 2343
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1121509427995613e-05,
+      "loss": 0.2068,
+      "step": 2344
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1053967328320635e-05,
+      "loss": 0.1435,
+      "step": 2345
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0986520686030196e-05,
+      "loss": 0.3327,
+      "step": 2346
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0919169582677234e-05,
+      "loss": 0.2854,
+      "step": 2347
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0851914099699143e-05,
+      "loss": 0.2023,
+      "step": 2348
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.078475431841771e-05,
+      "loss": 0.223,
+      "step": 2349
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0717690320039042e-05,
+      "loss": 0.151,
+      "step": 2350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0650722185653338e-05,
+      "loss": 0.2586,
+      "step": 2351
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.058384999623496e-05,
+      "loss": 0.1909,
+      "step": 2352
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.051707383264224e-05,
+      "loss": 0.1778,
+      "step": 2353
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0450393775617395e-05,
+      "loss": 0.2364,
+      "step": 2354
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0383809905786443e-05,
+      "loss": 0.2595,
+      "step": 2355
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0317322303659092e-05,
+      "loss": 0.1451,
+      "step": 2356
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025093104962865e-05,
+      "loss": 0.1384,
+      "step": 2357
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0184636223971942e-05,
+      "loss": 0.1355,
+      "step": 2358
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.011843790684913e-05,
+      "loss": 0.1424,
+      "step": 2359
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0052336178303766e-05,
+      "loss": 0.1325,
+      "step": 2360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9986331118262567e-05,
+      "loss": 0.1698,
+      "step": 2361
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9920422806535387e-05,
+      "loss": 0.1426,
+      "step": 2362
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9854611322815086e-05,
+      "loss": 0.1364,
+      "step": 2363
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978889674667742e-05,
+      "loss": 0.29,
+      "step": 2364
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972327915758102e-05,
+      "loss": 0.1582,
+      "step": 2365
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.965775863486722e-05,
+      "loss": 0.2151,
+      "step": 2366
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9592335257759988e-05,
+      "loss": 0.2239,
+      "step": 2367
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.952700910536578e-05,
+      "loss": 0.2601,
+      "step": 2368
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9461780256673578e-05,
+      "loss": 0.2175,
+      "step": 2369
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9396648790554662e-05,
+      "loss": 0.1613,
+      "step": 2370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9331614785762576e-05,
+      "loss": 0.2344,
+      "step": 2371
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9266678320932997e-05,
+      "loss": 0.228,
+      "step": 2372
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9201839474583693e-05,
+      "loss": 0.1767,
+      "step": 2373
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9137098325114387e-05,
+      "loss": 0.2165,
+      "step": 2374
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9072454950806652e-05,
+      "loss": 0.1593,
+      "step": 2375
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9007909429823877e-05,
+      "loss": 0.1476,
+      "step": 2376
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.894346184021106e-05,
+      "loss": 0.1747,
+      "step": 2377
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.887911225989487e-05,
+      "loss": 0.2306,
+      "step": 2378
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8814860766683418e-05,
+      "loss": 0.2334,
+      "step": 2379
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8750707438266213e-05,
+      "loss": 0.245,
+      "step": 2380
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8686652352214108e-05,
+      "loss": 0.2071,
+      "step": 2381
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.862269558597911e-05,
+      "loss": 0.2542,
+      "step": 2382
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8558837216894398e-05,
+      "loss": 0.1655,
+      "step": 2383
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8495077322174157e-05,
+      "loss": 0.1428,
+      "step": 2384
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.843141597891346e-05,
+      "loss": 0.1895,
+      "step": 2385
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8367853264088253e-05,
+      "loss": 0.2501,
+      "step": 2386
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8304389254555244e-05,
+      "loss": 0.2041,
+      "step": 2387
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8241024027051745e-05,
+      "loss": 0.2813,
+      "step": 2388
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8177757658195683e-05,
+      "loss": 0.2384,
+      "step": 2389
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8114590224485385e-05,
+      "loss": 0.2044,
+      "step": 2390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8051521802299622e-05,
+      "loss": 0.2082,
+      "step": 2391
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.798855246789738e-05,
+      "loss": 0.1518,
+      "step": 2392
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.792568229741789e-05,
+      "loss": 0.2652,
+      "step": 2393
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.786291136688041e-05,
+      "loss": 0.2262,
+      "step": 2394
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7800239752184267e-05,
+      "loss": 0.1744,
+      "step": 2395
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.773766752910867e-05,
+      "loss": 0.1561,
+      "step": 2396
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7675194773312665e-05,
+      "loss": 0.2486,
+      "step": 2397
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7612821560335e-05,
+      "loss": 0.1773,
+      "step": 2398
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.75505479655941e-05,
+      "loss": 0.2745,
+      "step": 2399
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7488374064387913e-05,
+      "loss": 0.1722,
+      "step": 2400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.742629993189385e-05,
+      "loss": 0.1856,
+      "step": 2401
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7364325643168655e-05,
+      "loss": 0.3334,
+      "step": 2402
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.730245127314839e-05,
+      "loss": 0.2386,
+      "step": 2403
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.724067689664828e-05,
+      "loss": 0.149,
+      "step": 2404
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.717900258836266e-05,
+      "loss": 0.1199,
+      "step": 2405
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7117428422864834e-05,
+      "loss": 0.1699,
+      "step": 2406
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.705595447460706e-05,
+      "loss": 0.1077,
+      "step": 2407
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.699458081792039e-05,
+      "loss": 0.1648,
+      "step": 2408
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.693330752701463e-05,
+      "loss": 0.1315,
+      "step": 2409
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687213467597821e-05,
+      "loss": 0.1762,
+      "step": 2410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6811062338778107e-05,
+      "loss": 0.1565,
+      "step": 2411
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.675009058925977e-05,
+      "loss": 0.1686,
+      "step": 2412
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6689219501147034e-05,
+      "loss": 0.1759,
+      "step": 2413
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6628449148042025e-05,
+      "loss": 0.1614,
+      "step": 2414
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.656777960342504e-05,
+      "loss": 0.2443,
+      "step": 2415
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6507210940654504e-05,
+      "loss": 0.1539,
+      "step": 2416
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6446743232966855e-05,
+      "loss": 0.2666,
+      "step": 2417
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.638637655347648e-05,
+      "loss": 0.1294,
+      "step": 2418
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6326110975175534e-05,
+      "loss": 0.1659,
+      "step": 2419
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.626594657093401e-05,
+      "loss": 0.2018,
+      "step": 2420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6205883413499544e-05,
+      "loss": 0.2993,
+      "step": 2421
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6145921575497337e-05,
+      "loss": 0.2173,
+      "step": 2422
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6086061129430085e-05,
+      "loss": 0.196,
+      "step": 2423
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6026302147677875e-05,
+      "loss": 0.2303,
+      "step": 2424
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5966644702498146e-05,
+      "loss": 0.1632,
+      "step": 2425
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590708886602551e-05,
+      "loss": 0.2148,
+      "step": 2426
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5847634710271797e-05,
+      "loss": 0.1412,
+      "step": 2427
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5788282307125768e-05,
+      "loss": 0.1855,
+      "step": 2428
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5729031728353273e-05,
+      "loss": 0.2116,
+      "step": 2429
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5669883045596967e-05,
+      "loss": 0.1641,
+      "step": 2430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5610836330376343e-05,
+      "loss": 0.105,
+      "step": 2431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5551891654087558e-05,
+      "loss": 0.2715,
+      "step": 2432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.549304908800342e-05,
+      "loss": 0.1555,
+      "step": 2433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.543430870327327e-05,
+      "loss": 0.205,
+      "step": 2434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5375670570922894e-05,
+      "loss": 0.1932,
+      "step": 2435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5317134761854402e-05,
+      "loss": 0.1846,
+      "step": 2436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5258701346846215e-05,
+      "loss": 0.1302,
+      "step": 2437
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5200370396552954e-05,
+      "loss": 0.223,
+      "step": 2438
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5142141981505342e-05,
+      "loss": 0.2259,
+      "step": 2439
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5084016172110082e-05,
+      "loss": 0.2192,
+      "step": 2440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5025993038649867e-05,
+      "loss": 0.1581,
+      "step": 2441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4968072651283205e-05,
+      "loss": 0.2177,
+      "step": 2442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4910255080044389e-05,
+      "loss": 0.2019,
+      "step": 2443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4852540394843373e-05,
+      "loss": 0.2153,
+      "step": 2444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4794928665465702e-05,
+      "loss": 0.2636,
+      "step": 2445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4737419961572452e-05,
+      "loss": 0.2462,
+      "step": 2446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4680014352700121e-05,
+      "loss": 0.2599,
+      "step": 2447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4622711908260544e-05,
+      "loss": 0.1406,
+      "step": 2448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4565512697540818e-05,
+      "loss": 0.2093,
+      "step": 2449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4508416789703227e-05,
+      "loss": 0.2586,
+      "step": 2450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.445142425378513e-05,
+      "loss": 0.2149,
+      "step": 2451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4394535158698897e-05,
+      "loss": 0.1615,
+      "step": 2452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.433774957323184e-05,
+      "loss": 0.1091,
+      "step": 2453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.428106756604607e-05,
+      "loss": 0.1134,
+      "step": 2454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4224489205678492e-05,
+      "loss": 0.1623,
+      "step": 2455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.41680145605407e-05,
+      "loss": 0.127,
+      "step": 2456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4111643698918831e-05,
+      "loss": 0.1674,
+      "step": 2457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4055376688973588e-05,
+      "loss": 0.144,
+      "step": 2458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3999213598740057e-05,
+      "loss": 0.1215,
+      "step": 2459
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3943154496127709e-05,
+      "loss": 0.1175,
+      "step": 2460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3887199448920252e-05,
+      "loss": 0.1853,
+      "step": 2461
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3831348524775567e-05,
+      "loss": 0.1673,
+      "step": 2462
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3775601791225645e-05,
+      "loss": 0.1725,
+      "step": 2463
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3719959315676523e-05,
+      "loss": 0.2556,
+      "step": 2464
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3664421165408148e-05,
+      "loss": 0.2129,
+      "step": 2465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3608987407574303e-05,
+      "loss": 0.2502,
+      "step": 2466
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3553658109202593e-05,
+      "loss": 0.1787,
+      "step": 2467
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3498433337194294e-05,
+      "loss": 0.2452,
+      "step": 2468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3443313158324267e-05,
+      "loss": 0.2018,
+      "step": 2469
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3388297639240976e-05,
+      "loss": 0.2352,
+      "step": 2470
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3333386846466233e-05,
+      "loss": 0.2131,
+      "step": 2471
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3278580846395295e-05,
+      "loss": 0.1997,
+      "step": 2472
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3223879705296693e-05,
+      "loss": 0.2455,
+      "step": 2473
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3169283489312168e-05,
+      "loss": 0.201,
+      "step": 2474
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3114792264456598e-05,
+      "loss": 0.1851,
+      "step": 2475
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3060406096617883e-05,
+      "loss": 0.2747,
+      "step": 2476
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3006125051556928e-05,
+      "loss": 0.1699,
+      "step": 2477
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.295194919490753e-05,
+      "loss": 0.1902,
+      "step": 2478
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2897878592176238e-05,
+      "loss": 0.2164,
+      "step": 2479
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2843913308742394e-05,
+      "loss": 0.2386,
+      "step": 2480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2790053409857982e-05,
+      "loss": 0.2077,
+      "step": 2481
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2736298960647553e-05,
+      "loss": 0.1532,
+      "step": 2482
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2682650026108155e-05,
+      "loss": 0.241,
+      "step": 2483
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2629106671109258e-05,
+      "loss": 0.2217,
+      "step": 2484
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2575668960392662e-05,
+      "loss": 0.1904,
+      "step": 2485
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2522336958572433e-05,
+      "loss": 0.2386,
+      "step": 2486
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2469110730134848e-05,
+      "loss": 0.1434,
+      "step": 2487
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.24159903394382e-05,
+      "loss": 0.1009,
+      "step": 2488
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2362975850712899e-05,
+      "loss": 0.1986,
+      "step": 2489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.231006732806127e-05,
+      "loss": 0.2126,
+      "step": 2490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2257264835457515e-05,
+      "loss": 0.2356,
+      "step": 2491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2204568436747632e-05,
+      "loss": 0.3066,
+      "step": 2492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2151978195649316e-05,
+      "loss": 0.1777,
+      "step": 2493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2099494175751924e-05,
+      "loss": 0.1811,
+      "step": 2494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2047116440516393e-05,
+      "loss": 0.0891,
+      "step": 2495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1994845053275083e-05,
+      "loss": 0.1948,
+      "step": 2496
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1942680077231805e-05,
+      "loss": 0.3144,
+      "step": 2497
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1890621575461725e-05,
+      "loss": 0.2858,
+      "step": 2498
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1838669610911213e-05,
+      "loss": 0.116,
+      "step": 2499
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1786824246397876e-05,
+      "loss": 0.2198,
+      "step": 2500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.173508554461038e-05,
+      "loss": 0.1982,
+      "step": 2501
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1683453568108437e-05,
+      "loss": 0.1378,
+      "step": 2502
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1631928379322732e-05,
+      "loss": 0.2314,
+      "step": 2503
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1580510040554815e-05,
+      "loss": 0.1632,
+      "step": 2504
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1529198613977e-05,
+      "loss": 0.1921,
+      "step": 2505
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1477994161632366e-05,
+      "loss": 0.1915,
+      "step": 2506
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1426896745434657e-05,
+      "loss": 0.1196,
+      "step": 2507
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.137590642716817e-05,
+      "loss": 0.1382,
+      "step": 2508
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1325023268487712e-05,
+      "loss": 0.1761,
+      "step": 2509
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.127424733091853e-05,
+      "loss": 0.1175,
+      "step": 2510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1223578675856195e-05,
+      "loss": 0.132,
+      "step": 2511
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1173017364566585e-05,
+      "loss": 0.1333,
+      "step": 2512
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1122563458185786e-05,
+      "loss": 0.151,
+      "step": 2513
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1072217017719977e-05,
+      "loss": 0.24,
+      "step": 2514
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1021978104045438e-05,
+      "loss": 0.2362,
+      "step": 2515
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0971846777908413e-05,
+      "loss": 0.2853,
+      "step": 2516
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0921823099925065e-05,
+      "loss": 0.2061,
+      "step": 2517
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0871907130581394e-05,
+      "loss": 0.2156,
+      "step": 2518
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.082209893023316e-05,
+      "loss": 0.1581,
+      "step": 2519
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0772398559105801e-05,
+      "loss": 0.2974,
+      "step": 2520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0722806077294423e-05,
+      "loss": 0.1844,
+      "step": 2521
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0673321544763603e-05,
+      "loss": 0.1946,
+      "step": 2522
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0623945021347448e-05,
+      "loss": 0.3048,
+      "step": 2523
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0574676566749442e-05,
+      "loss": 0.089,
+      "step": 2524
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.052551624054241e-05,
+      "loss": 0.2759,
+      "step": 2525
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0476464102168437e-05,
+      "loss": 0.23,
+      "step": 2526
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042752021093878e-05,
+      "loss": 0.1553,
+      "step": 2527
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0378684626033807e-05,
+      "loss": 0.2585,
+      "step": 2528
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0329957406502955e-05,
+      "loss": 0.2131,
+      "step": 2529
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0281338611264601e-05,
+      "loss": 0.2163,
+      "step": 2530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0232828299106024e-05,
+      "loss": 0.3154,
+      "step": 2531
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0184426528683334e-05,
+      "loss": 0.2102,
+      "step": 2532
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0136133358521416e-05,
+      "loss": 0.1803,
+      "step": 2533
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0087948847013818e-05,
+      "loss": 0.1726,
+      "step": 2534
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0039873052422721e-05,
+      "loss": 0.1891,
+      "step": 2535
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.991906032878839e-06,
+      "loss": 0.1524,
+      "step": 2536
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.944047846381366e-06,
+      "loss": 0.2146,
+      "step": 2537
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.896298550797922e-06,
+      "loss": 0.1965,
+      "step": 2538
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.848658203864403e-06,
+      "loss": 0.2372,
+      "step": 2539
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.801126863185028e-06,
+      "loss": 0.2058,
+      "step": 2540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.753704586232183e-06,
+      "loss": 0.2403,
+      "step": 2541
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.706391430346384e-06,
+      "loss": 0.2313,
+      "step": 2542
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.659187452736219e-06,
+      "loss": 0.2898,
+      "step": 2543
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.612092710478248e-06,
+      "loss": 0.27,
+      "step": 2544
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.565107260516958e-06,
+      "loss": 0.2095,
+      "step": 2545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.51823115966467e-06,
+      "loss": 0.1507,
+      "step": 2546
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.471464464601521e-06,
+      "loss": 0.2088,
+      "step": 2547
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.424807231875287e-06,
+      "loss": 0.2655,
+      "step": 2548
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.378259517901478e-06,
+      "loss": 0.1224,
+      "step": 2549
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.331821378963113e-06,
+      "loss": 0.1944,
+      "step": 2550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.285492871210755e-06,
+      "loss": 0.2647,
+      "step": 2551
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.239274050662394e-06,
+      "loss": 0.2501,
+      "step": 2552
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.193164973203406e-06,
+      "loss": 0.1596,
+      "step": 2553
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.147165694586435e-06,
+      "loss": 0.2753,
+      "step": 2554
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.101276270431424e-06,
+      "loss": 0.2121,
+      "step": 2555
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.055496756225402e-06,
+      "loss": 0.2489,
+      "step": 2556
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.009827207322563e-06,
+      "loss": 0.1708,
+      "step": 2557
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.964267678944127e-06,
+      "loss": 0.1545,
+      "step": 2558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.918818226178272e-06,
+      "loss": 0.1525,
+      "step": 2559
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.87347890398007e-06,
+      "loss": 0.261,
+      "step": 2560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.828249767171471e-06,
+      "loss": 0.1927,
+      "step": 2561
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.783130870441148e-06,
+      "loss": 0.1258,
+      "step": 2562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.738122268344483e-06,
+      "loss": 0.1529,
+      "step": 2563
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.693224015303547e-06,
+      "loss": 0.1509,
+      "step": 2564
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.648436165606888e-06,
+      "loss": 0.1571,
+      "step": 2565
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.603758773409642e-06,
+      "loss": 0.1638,
+      "step": 2566
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.559191892733364e-06,
+      "loss": 0.2387,
+      "step": 2567
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.514735577465971e-06,
+      "loss": 0.1952,
+      "step": 2568
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.470389881361696e-06,
+      "loss": 0.1665,
+      "step": 2569
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.426154858041013e-06,
+      "loss": 0.1399,
+      "step": 2570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.382030560990606e-06,
+      "loss": 0.1757,
+      "step": 2571
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.33801704356325e-06,
+      "loss": 0.2174,
+      "step": 2572
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.294114358977734e-06,
+      "loss": 0.1954,
+      "step": 2573
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250322560318913e-06,
+      "loss": 0.2527,
+      "step": 2574
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.206641700537487e-06,
+      "loss": 0.2251,
+      "step": 2575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.163071832450086e-06,
+      "loss": 0.164,
+      "step": 2576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.11961300873907e-06,
+      "loss": 0.25,
+      "step": 2577
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.07626528195259e-06,
+      "loss": 0.1999,
+      "step": 2578
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.033028704504398e-06,
+      "loss": 0.2112,
+      "step": 2579
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.98990332867392e-06,
+      "loss": 0.2149,
+      "step": 2580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.94688920660609e-06,
+      "loss": 0.3022,
+      "step": 2581
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.90398639031129e-06,
+      "loss": 0.1741,
+      "step": 2582
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.86119493166535e-06,
+      "loss": 0.2895,
+      "step": 2583
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.818514882409466e-06,
+      "loss": 0.2402,
+      "step": 2584
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.775946294150094e-06,
+      "loss": 0.1703,
+      "step": 2585
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.733489218358925e-06,
+      "loss": 0.181,
+      "step": 2586
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.69114370637284e-06,
+      "loss": 0.2623,
+      "step": 2587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.648909809393768e-06,
+      "loss": 0.1187,
+      "step": 2588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.606787578488761e-06,
+      "loss": 0.2126,
+      "step": 2589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.564777064589757e-06,
+      "loss": 0.1843,
+      "step": 2590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.522878318493687e-06,
+      "loss": 0.2707,
+      "step": 2591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.481091390862283e-06,
+      "loss": 0.206,
+      "step": 2592
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4394163322220936e-06,
+      "loss": 0.1907,
+      "step": 2593
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.397853192964432e-06,
+      "loss": 0.2158,
+      "step": 2594
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.356402023345233e-06,
+      "loss": 0.2106,
+      "step": 2595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.31506287348509e-06,
+      "loss": 0.2336,
+      "step": 2596
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.273835793369099e-06,
+      "loss": 0.2653,
+      "step": 2597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.232720832846895e-06,
+      "loss": 0.2264,
+      "step": 2598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.1917180416325225e-06,
+      "loss": 0.1803,
+      "step": 2599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.1508274693044e-06,
+      "loss": 0.2294,
+      "step": 2600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110049165305233e-06,
+      "loss": 0.1789,
+      "step": 2601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.069383178942013e-06,
+      "loss": 0.1568,
+      "step": 2602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.028829559385897e-06,
+      "loss": 0.1761,
+      "step": 2603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.988388355672204e-06,
+      "loss": 0.1399,
+      "step": 2604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.948059616700298e-06,
+      "loss": 0.1384,
+      "step": 2605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.90784339123356e-06,
+      "loss": 0.2397,
+      "step": 2606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.867739727899336e-06,
+      "loss": 0.1506,
+      "step": 2607
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.827748675188894e-06,
+      "loss": 0.1264,
+      "step": 2608
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.787870281457254e-06,
+      "loss": 0.1277,
+      "step": 2609
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.748104594923299e-06,
+      "loss": 0.1375,
+      "step": 2610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.7084516636695905e-06,
+      "loss": 0.1655,
+      "step": 2611
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.668911535642386e-06,
+      "loss": 0.1535,
+      "step": 2612
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.629484258651508e-06,
+      "loss": 0.1598,
+      "step": 2613
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.590169880370345e-06,
+      "loss": 0.2634,
+      "step": 2614
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.550968448335781e-06,
+      "loss": 0.1943,
+      "step": 2615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.511880009948113e-06,
+      "loss": 0.2272,
+      "step": 2616
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.472904612471042e-06,
+      "loss": 0.3786,
+      "step": 2617
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.434042303031529e-06,
+      "loss": 0.203,
+      "step": 2618
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.395293128619861e-06,
+      "loss": 0.1912,
+      "step": 2619
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.356657136089483e-06,
+      "loss": 0.1368,
+      "step": 2620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.318134372157014e-06,
+      "loss": 0.1134,
+      "step": 2621
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.279724883402149e-06,
+      "loss": 0.1894,
+      "step": 2622
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.241428716267605e-06,
+      "loss": 0.1,
+      "step": 2623
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.203245917059119e-06,
+      "loss": 0.1881,
+      "step": 2624
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.165176531945293e-06,
+      "loss": 0.2428,
+      "step": 2625
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.12722060695764e-06,
+      "loss": 0.1268,
+      "step": 2626
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.089378187990458e-06,
+      "loss": 0.2073,
+      "step": 2627
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.051649320800801e-06,
+      "loss": 0.1796,
+      "step": 2628
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.0140340510084215e-06,
+      "loss": 0.1759,
+      "step": 2629
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.976532424095749e-06,
+      "loss": 0.1207,
+      "step": 2630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.939144485407755e-06,
+      "loss": 0.2686,
+      "step": 2631
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.901870280151978e-06,
+      "loss": 0.1489,
+      "step": 2632
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.864709853398409e-06,
+      "loss": 0.2701,
+      "step": 2633
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.827663250079507e-06,
+      "loss": 0.1171,
+      "step": 2634
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.790730514990039e-06,
+      "loss": 0.2932,
+      "step": 2635
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.753911692787128e-06,
+      "loss": 0.2091,
+      "step": 2636
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.7172068279901734e-06,
+      "loss": 0.202,
+      "step": 2637
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.680615964980762e-06,
+      "loss": 0.0949,
+      "step": 2638
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.644139148002614e-06,
+      "loss": 0.1215,
+      "step": 2639
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6077764211616035e-06,
+      "loss": 0.1751,
+      "step": 2640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.571527828425616e-06,
+      "loss": 0.1541,
+      "step": 2641
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.5353934136245344e-06,
+      "loss": 0.2365,
+      "step": 2642
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.49937322045021e-06,
+      "loss": 0.233,
+      "step": 2643
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.463467292456326e-06,
+      "loss": 0.3319,
+      "step": 2644
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.427675673058463e-06,
+      "loss": 0.1389,
+      "step": 2645
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.39199840553396e-06,
+      "loss": 0.3317,
+      "step": 2646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.356435533021887e-06,
+      "loss": 0.2574,
+      "step": 2647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3209870985230005e-06,
+      "loss": 0.2559,
+      "step": 2648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.285653144899682e-06,
+      "loss": 0.2612,
+      "step": 2649
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.250433714875879e-06,
+      "loss": 0.1202,
+      "step": 2650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.215328851037094e-06,
+      "loss": 0.1817,
+      "step": 2651
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.180338595830236e-06,
+      "loss": 0.2094,
+      "step": 2652
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.14546299156371e-06,
+      "loss": 0.1402,
+      "step": 2653
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.110702080407248e-06,
+      "loss": 0.1304,
+      "step": 2654
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0760559043919295e-06,
+      "loss": 0.203,
+      "step": 2655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.041524505410067e-06,
+      "loss": 0.1356,
+      "step": 2656
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.0071079252152156e-06,
+      "loss": 0.1402,
+      "step": 2657
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.972806205422087e-06,
+      "loss": 0.181,
+      "step": 2658
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.938619387506527e-06,
+      "loss": 0.1913,
+      "step": 2659
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904547512805446e-06,
+      "loss": 0.1151,
+      "step": 2660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8705906225167135e-06,
+      "loss": 0.1171,
+      "step": 2661
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.83674875769925e-06,
+      "loss": 0.1309,
+      "step": 2662
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.803021959272846e-06,
+      "loss": 0.2028,
+      "step": 2663
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7694102680181706e-06,
+      "loss": 0.2055,
+      "step": 2664
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.735913724576713e-06,
+      "loss": 0.2651,
+      "step": 2665
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.702532369450729e-06,
+      "loss": 0.2767,
+      "step": 2666
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.669266243003179e-06,
+      "loss": 0.2328,
+      "step": 2667
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.636115385457751e-06,
+      "loss": 0.1668,
+      "step": 2668
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.603079836898672e-06,
+      "loss": 0.2241,
+      "step": 2669
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.57015963727081e-06,
+      "loss": 0.2676,
+      "step": 2670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.537354826379525e-06,
+      "loss": 0.1991,
+      "step": 2671
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.504665443890677e-06,
+      "loss": 0.1661,
+      "step": 2672
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.472091529330535e-06,
+      "loss": 0.1784,
+      "step": 2673
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.439633122085773e-06,
+      "loss": 0.21,
+      "step": 2674
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.407290261403374e-06,
+      "loss": 0.1837,
+      "step": 2675
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.375062986390643e-06,
+      "loss": 0.2332,
+      "step": 2676
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3429513360150885e-06,
+      "loss": 0.2538,
+      "step": 2677
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.310955349104418e-06,
+      "loss": 0.1887,
+      "step": 2678
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.279075064346505e-06,
+      "loss": 0.2587,
+      "step": 2679
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.247310520289316e-06,
+      "loss": 0.2124,
+      "step": 2680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.21566175534085e-06,
+      "loss": 0.1739,
+      "step": 2681
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.184128807769138e-06,
+      "loss": 0.1212,
+      "step": 2682
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.152711715702151e-06,
+      "loss": 0.2056,
+      "step": 2683
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1214105171277884e-06,
+      "loss": 0.1849,
+      "step": 2684
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.090225249893831e-06,
+      "loss": 0.215,
+      "step": 2685
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.059155951707816e-06,
+      "loss": 0.2326,
+      "step": 2686
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.028202660137126e-06,
+      "loss": 0.1436,
+      "step": 2687
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.997365412608855e-06,
+      "loss": 0.2288,
+      "step": 2688
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9666442464097785e-06,
+      "loss": 0.1637,
+      "step": 2689
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9360391986863036e-06,
+      "loss": 0.2478,
+      "step": 2690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9055503064444385e-06,
+      "loss": 0.1763,
+      "step": 2691
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8751776065497626e-06,
+      "loss": 0.1863,
+      "step": 2692
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.844921135727342e-06,
+      "loss": 0.1583,
+      "step": 2693
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.814780930561701e-06,
+      "loss": 0.1744,
+      "step": 2694
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.784757027496788e-06,
+      "loss": 0.1412,
+      "step": 2695
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7548494628359187e-06,
+      "loss": 0.2082,
+      "step": 2696
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7250582727417548e-06,
+      "loss": 0.1647,
+      "step": 2697
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6953834932362264e-06,
+      "loss": 0.2277,
+      "step": 2698
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.665825160200531e-06,
+      "loss": 0.214,
+      "step": 2699
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.636383309375024e-06,
+      "loss": 0.118,
+      "step": 2700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.607057976359263e-06,
+      "loss": 0.1898,
+      "step": 2701
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5778491966118822e-06,
+      "loss": 0.2021,
+      "step": 2702
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5487570054506093e-06,
+      "loss": 0.1633,
+      "step": 2703
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.519781438052183e-06,
+      "loss": 0.2171,
+      "step": 2704
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.490922529452323e-06,
+      "loss": 0.1318,
+      "step": 2705
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4621803145457264e-06,
+      "loss": 0.1536,
+      "step": 2706
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4335548280859388e-06,
+      "loss": 0.1169,
+      "step": 2707
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4050461046854165e-06,
+      "loss": 0.1148,
+      "step": 2708
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.376654178815386e-06,
+      "loss": 0.1887,
+      "step": 2709
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.348379084805875e-06,
+      "loss": 0.1284,
+      "step": 2710
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.320220856845646e-06,
+      "loss": 0.1525,
+      "step": 2711
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.292179528982131e-06,
+      "loss": 0.1489,
+      "step": 2712
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.264255135121441e-06,
+      "loss": 0.2035,
+      "step": 2713
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2364477090282674e-06,
+      "loss": 0.3059,
+      "step": 2714
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.208757284325892e-06,
+      "loss": 0.2666,
+      "step": 2715
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1811838944961203e-06,
+      "loss": 0.276,
+      "step": 2716
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1537275728792503e-06,
+      "loss": 0.1842,
+      "step": 2717
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1263883526740035e-06,
+      "loss": 0.2343,
+      "step": 2718
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0991662669375254e-06,
+      "loss": 0.1764,
+      "step": 2719
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0720613485853533e-06,
+      "loss": 0.2604,
+      "step": 2720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.045073630391304e-06,
+      "loss": 0.1994,
+      "step": 2721
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.018203144987497e-06,
+      "loss": 0.3104,
+      "step": 2722
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.991449924864331e-06,
+      "loss": 0.208,
+      "step": 2723
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.964814002370375e-06,
+      "loss": 0.2423,
+      "step": 2724
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9382954097123882e-06,
+      "loss": 0.1449,
+      "step": 2725
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9118941789552654e-06,
+      "loss": 0.1985,
+      "step": 2726
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8856103420219606e-06,
+      "loss": 0.2424,
+      "step": 2727
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8594439306935396e-06,
+      "loss": 0.1222,
+      "step": 2728
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.833394976609027e-06,
+      "loss": 0.219,
+      "step": 2729
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8074635112654403e-06,
+      "loss": 0.1634,
+      "step": 2730
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.781649566017741e-06,
+      "loss": 0.2715,
+      "step": 2731
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7559531720788066e-06,
+      "loss": 0.2777,
+      "step": 2732
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.730374360519361e-06,
+      "loss": 0.1505,
+      "step": 2733
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7049131622679523e-06,
+      "loss": 0.1926,
+      "step": 2734
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.679569608110921e-06,
+      "loss": 0.2269,
+      "step": 2735
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6543437286923545e-06,
+      "loss": 0.1555,
+      "step": 2736
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6292355545140645e-06,
+      "loss": 0.2267,
+      "step": 2737
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.604245115935533e-06,
+      "loss": 0.195,
+      "step": 2738
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5793724431738666e-06,
+      "loss": 0.2174,
+      "step": 2739
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5546175663037965e-06,
+      "loss": 0.1165,
+      "step": 2740
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.529980515257635e-06,
+      "loss": 0.1473,
+      "step": 2741
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5054613198251862e-06,
+      "loss": 0.1998,
+      "step": 2742
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4810600096537683e-06,
+      "loss": 0.2253,
+      "step": 2743
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.45677661424818e-06,
+      "loss": 0.2037,
+      "step": 2744
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4326111629706217e-06,
+      "loss": 0.381,
+      "step": 2745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4085636850406656e-06,
+      "loss": 0.1458,
+      "step": 2746
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3846342095352527e-06,
+      "loss": 0.1758,
+      "step": 2747
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3608227653886706e-06,
+      "loss": 0.2187,
+      "step": 2748
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3371293813924445e-06,
+      "loss": 0.3045,
+      "step": 2749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3135540861953684e-06,
+      "loss": 0.2261,
+      "step": 2750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2900969083034518e-06,
+      "loss": 0.2186,
+      "step": 2751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2667578760798834e-06,
+      "loss": 0.3057,
+      "step": 2752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2435370177449787e-06,
+      "loss": 0.1735,
+      "step": 2753
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2204343613762003e-06,
+      "loss": 0.1693,
+      "step": 2754
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.197449934908025e-06,
+      "loss": 0.1693,
+      "step": 2755
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.174583766132043e-06,
+      "loss": 0.2154,
+      "step": 2756
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1518358826968064e-06,
+      "loss": 0.2313,
+      "step": 2757
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1292063121078564e-06,
+      "loss": 0.2164,
+      "step": 2758
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1066950817276944e-06,
+      "loss": 0.1327,
+      "step": 2759
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.084302218775691e-06,
+      "loss": 0.1538,
+      "step": 2760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.062027750328144e-06,
+      "loss": 0.1138,
+      "step": 2761
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0398717033181414e-06,
+      "loss": 0.1311,
+      "step": 2762
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.017834104535621e-06,
+      "loss": 0.1427,
+      "step": 2763
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9959149806272893e-06,
+      "loss": 0.1844,
+      "step": 2764
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.97411435809659e-06,
+      "loss": 0.187,
+      "step": 2765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.952432263303683e-06,
+      "loss": 0.173,
+      "step": 2766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9308687224654288e-06,
+      "loss": 0.1822,
+      "step": 2767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9094237616553155e-06,
+      "loss": 0.2673,
+      "step": 2768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8880974068034574e-06,
+      "loss": 0.2075,
+      "step": 2769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8668896836965711e-06,
+      "loss": 0.2304,
+      "step": 2770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8458006179779107e-06,
+      "loss": 0.1666,
+      "step": 2771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8248302351472679e-06,
+      "loss": 0.2669,
+      "step": 2772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.803978560560915e-06,
+      "loss": 0.152,
+      "step": 2773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7832456194316172e-06,
+      "loss": 0.2589,
+      "step": 2774
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7626314368285324e-06,
+      "loss": 0.1745,
+      "step": 2775
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7421360376772555e-06,
+      "loss": 0.1417,
+      "step": 2776
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.721759446759752e-06,
+      "loss": 0.1703,
+      "step": 2777
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7015016887143132e-06,
+      "loss": 0.3492,
+      "step": 2778
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6813627880355453e-06,
+      "loss": 0.1846,
+      "step": 2779
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6613427690743698e-06,
+      "loss": 0.2344,
+      "step": 2780
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.641441656037912e-06,
+      "loss": 0.2258,
+      "step": 2781
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6216594729895672e-06,
+      "loss": 0.1345,
+      "step": 2782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6019962438489022e-06,
+      "loss": 0.2295,
+      "step": 2783
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5824519923916647e-06,
+      "loss": 0.1683,
+      "step": 2784
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.563026742249718e-06,
+      "loss": 0.2061,
+      "step": 2785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5437205169110735e-06,
+      "loss": 0.2748,
+      "step": 2786
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5245333397197803e-06,
+      "loss": 0.2215,
+      "step": 2787
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5054652338759911e-06,
+      "loss": 0.2169,
+      "step": 2788
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4865162224358186e-06,
+      "loss": 0.321,
+      "step": 2789
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.467686328311435e-06,
+      "loss": 0.1443,
+      "step": 2790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4489755742709499e-06,
+      "loss": 0.2433,
+      "step": 2791
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4303839829384213e-06,
+      "loss": 0.1378,
+      "step": 2792
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4119115767938228e-06,
+      "loss": 0.2367,
+      "step": 2793
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.39355837817301e-06,
+      "loss": 0.1972,
+      "step": 2794
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3753244092677197e-06,
+      "loss": 0.2188,
+      "step": 2795
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.357209692125483e-06,
+      "loss": 0.1924,
+      "step": 2796
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3392142486496896e-06,
+      "loss": 0.2152,
+      "step": 2797
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3213381005994563e-06,
+      "loss": 0.1456,
+      "step": 2798
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3035812695896821e-06,
+      "loss": 0.2004,
+      "step": 2799
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2859437770910142e-06,
+      "loss": 0.1549,
+      "step": 2800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2684256444297492e-06,
+      "loss": 0.255,
+      "step": 2801
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.251026892787921e-06,
+      "loss": 0.1188,
+      "step": 2802
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2337475432031565e-06,
+      "loss": 0.2172,
+      "step": 2803
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.216587616568754e-06,
+      "loss": 0.211,
+      "step": 2804
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1995471336335939e-06,
+      "loss": 0.148,
+      "step": 2805
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1826261150021056e-06,
+      "loss": 0.1972,
+      "step": 2806
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1658245811343227e-06,
+      "loss": 0.127,
+      "step": 2807
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.149142552345761e-06,
+      "loss": 0.1629,
+      "step": 2808
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1325800488074524e-06,
+      "loss": 0.1597,
+      "step": 2809
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.11613709054591e-06,
+      "loss": 0.1279,
+      "step": 2810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.099813697443075e-06,
+      "loss": 0.1664,
+      "step": 2811
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.083609889236359e-06,
+      "loss": 0.1506,
+      "step": 2812
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0675256855185333e-06,
+      "loss": 0.1292,
+      "step": 2813
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.051561105737786e-06,
+      "loss": 0.2775,
+      "step": 2814
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0357161691976424e-06,
+      "loss": 0.2833,
+      "step": 2815
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0199908950569548e-06,
+      "loss": 0.2233,
+      "step": 2816
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.004385302329902e-06,
+      "loss": 0.2085,
+      "step": 2817
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.888994098859349e-07,
+      "loss": 0.2497,
+      "step": 2818
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.735332364498085e-07,
+      "loss": 0.1599,
+      "step": 2819
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.5828680060146e-07,
+      "loss": 0.2149,
+      "step": 2820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.431601207760876e-07,
+      "loss": 0.1417,
+      "step": 2821
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.281532152640826e-07,
+      "loss": 0.1596,
+      "step": 2822
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.132661022109967e-07,
+      "loss": 0.1245,
+      "step": 2823
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.98498799617542e-07,
+      "loss": 0.3124,
+      "step": 2824
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.838513253395686e-07,
+      "loss": 0.1425,
+      "step": 2825
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.693236970880204e-07,
+      "loss": 0.2077,
+      "step": 2826
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.549159324289568e-07,
+      "loss": 0.0893,
+      "step": 2827
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.406280487834761e-07,
+      "loss": 0.2384,
+      "step": 2828
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.264600634277364e-07,
+      "loss": 0.1416,
+      "step": 2829
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.12411993492923e-07,
+      "loss": 0.2074,
+      "step": 2830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.984838559652153e-07,
+      "loss": 0.1632,
+      "step": 2831
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.84675667685797e-07,
+      "loss": 0.278,
+      "step": 2832
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.709874453507793e-07,
+      "loss": 0.2063,
+      "step": 2833
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.574192055112672e-07,
+      "loss": 0.1712,
+      "step": 2834
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.439709645732373e-07,
+      "loss": 0.1493,
+      "step": 2835
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.306427387976045e-07,
+      "loss": 0.2384,
+      "step": 2836
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.174345443001551e-07,
+      "loss": 0.1928,
+      "step": 2837
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.043463970515585e-07,
+      "loss": 0.1313,
+      "step": 2838
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.913783128772888e-07,
+      "loss": 0.2457,
+      "step": 2839
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.78530307457681e-07,
+      "loss": 0.2245,
+      "step": 2840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.658023963278748e-07,
+      "loss": 0.2039,
+      "step": 2841
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.531945948777706e-07,
+      "loss": 0.1864,
+      "step": 2842
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.407069183520742e-07,
+      "loss": 0.2635,
+      "step": 2843
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.283393818502181e-07,
+      "loss": 0.1253,
+      "step": 2844
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.160920003263737e-07,
+      "loss": 0.1568,
+      "step": 2845
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.039647885894394e-07,
+      "loss": 0.2044,
+      "step": 2846
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.919577613029969e-07,
+      "loss": 0.1493,
+      "step": 2847
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.800709329853215e-07,
+      "loss": 0.1863,
+      "step": 2848
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.683043180093161e-07,
+      "loss": 0.1277,
+      "step": 2849
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.566579306025665e-07,
+      "loss": 0.1327,
+      "step": 2850
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.451317848472637e-07,
+      "loss": 0.1757,
+      "step": 2851
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.337258946802259e-07,
+      "loss": 0.2602,
+      "step": 2852
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.224402738928546e-07,
+      "loss": 0.1335,
+      "step": 2853
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.11274936131112e-07,
+      "loss": 0.1608,
+      "step": 2854
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.002298948955542e-07,
+      "loss": 0.3109,
+      "step": 2855
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.893051635412649e-07,
+      "loss": 0.117,
+      "step": 2856
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.785007552778331e-07,
+      "loss": 0.1811,
+      "step": 2857
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6781668316939754e-07,
+      "loss": 0.1886,
+      "step": 2858
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5725296013457986e-07,
+      "loss": 0.1668,
+      "step": 2859
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.468095989464627e-07,
+      "loss": 0.2117,
+      "step": 2860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3648661223261166e-07,
+      "loss": 0.1998,
+      "step": 2861
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2628401247504225e-07,
+      "loss": 0.1394,
+      "step": 2862
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1620181201019737e-07,
+      "loss": 0.1337,
+      "step": 2863
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.062400230289476e-07,
+      "loss": 0.2062,
+      "step": 2864
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9639865757655767e-07,
+      "loss": 0.2334,
+      "step": 2865
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.866777275526978e-07,
+      "loss": 0.1098,
+      "step": 2866
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.770772447113879e-07,
+      "loss": 0.189,
+      "step": 2867
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.675972206610534e-07,
+      "loss": 0.1145,
+      "step": 2868
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5823766686441387e-07,
+      "loss": 0.2007,
+      "step": 2869
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4899859463854987e-07,
+      "loss": 0.2445,
+      "step": 2870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3988001515488087e-07,
+      "loss": 0.1461,
+      "step": 2871
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.308819394390983e-07,
+      "loss": 0.2117,
+      "step": 2872
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.22004378371199e-07,
+      "loss": 0.1426,
+      "step": 2873
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.132473426854743e-07,
+      "loss": 0.276,
+      "step": 2874
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0461084297046524e-07,
+      "loss": 0.2576,
+      "step": 2875
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9609488966896306e-07,
+      "loss": 0.14,
+      "step": 2876
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8769949307803077e-07,
+      "loss": 0.2482,
+      "step": 2877
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.79424663348915e-07,
+      "loss": 0.2008,
+      "step": 2878
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.712704104871233e-07,
+      "loss": 0.18,
+      "step": 2879
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.632367443523354e-07,
+      "loss": 0.1912,
+      "step": 2880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5532367465843643e-07,
+      "loss": 0.2221,
+      "step": 2881
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.475312109735173e-07,
+      "loss": 0.2123,
+      "step": 2882
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3985936271978535e-07,
+      "loss": 0.1461,
+      "step": 2883
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3230813917365368e-07,
+      "loss": 0.2222,
+      "step": 2884
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2487754946565188e-07,
+      "loss": 0.2516,
+      "step": 2885
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1756760258047072e-07,
+      "loss": 0.1925,
+      "step": 2886
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1037830735690655e-07,
+      "loss": 0.1968,
+      "step": 2887
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0330967248786137e-07,
+      "loss": 0.1792,
+      "step": 2888
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.963617065203871e-07,
+      "loss": 0.1778,
+      "step": 2889
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8953441785557468e-07,
+      "loss": 0.2532,
+      "step": 2890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8282781474863174e-07,
+      "loss": 0.2088,
+      "step": 2891
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7624190530882712e-07,
+      "loss": 0.2439,
+      "step": 2892
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6977669749949076e-07,
+      "loss": 0.1695,
+      "step": 2893
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.634321991380139e-07,
+      "loss": 0.2294,
+      "step": 2894
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5720841789583773e-07,
+      "loss": 0.1801,
+      "step": 2895
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5110536129843145e-07,
+      "loss": 0.2434,
+      "step": 2896
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4512303672526984e-07,
+      "loss": 0.1213,
+      "step": 2897
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.392614514098889e-07,
+      "loss": 0.2212,
+      "step": 2898
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3352061243978588e-07,
+      "loss": 0.2281,
+      "step": 2899
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.279005267565081e-07,
+      "loss": 0.1365,
+      "step": 2900
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2946,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 7.102247586729984e+16,
+  "trial_name": null,
+  "trial_params": null
+}